Information Technology R&D: Critical Trends and Issues

Home , Convergent Technologies, Information industry, VLSI Project

February 1985

NTIS order #PB85-245660 —

Recommended Citation: Information Technology and R&D: Critical Trends and Issues (Washington, DC: U.S. Congress, Office of Technology Assessment, OTA-CIT-268, February 1985).

Library of Congress Catalog Card Number 84-601150

For sale by the Superintendent of Documents U.S. Government Printing Office, Washington, DC 20402 Foreword

New computer and communications technologies are obviously transforming American life. They are the basis of many of the changes in our telecommunications system and also a new wave of automation on the farm, in manufacturing and transportation, and in the office. They are changing the form and delivery of government services such as education and the judicial system. Information products and services have become a major and still rapidly growing component of our economy. A strong U.S. research and development effort has, in the past, been the source of much of this new technology. However, recent events, such as the restructuring of the U.S. telecommunications industry and the emergence of strong foreign competition for some technologies, have changed the environment for R&D. Con- sequently, the House Committee on Science and Technology, the House Commit- tee on Energy and Commerce, and its Subcommittee on Telecommunications, Con- sumer Protection, and Finance asked OTA to conduct an assessment of the current state of R&D in these critical areas. In this report, OTA examines four specific areas of research as case studies: computer architecture, artificial intelligence, fiber optics, and software engineering. It discusses the structure and orientation of some selected foreign programs. Finally, it examines a set of issues that have been raised in the course of the study: manpower, institutional change, the new research organizations that grew out of Bell Laboratories, and implications of trends in overall science and technology policy. Information technology research and development in the United States is a remarkably adaptable system— important changes may already be taking place in funding patterns, institutional structures, manpower development, and government policies. Hence, the policy issues for Congress are not so much to stimulate change as to remove barriers to productive change, and to monitor and maintain the health of the R&D enterprise in a time of rapid change in an industry so central to our national economy and security. OTA gratefully acknowledges the contributions of the many experts, within and outside the Government, who served as panelists, consultants, contractors, and reviewers of this document. As with all OTA reports, however, the content is the responsibility of OTA and does not necessarily constitute the consensus or endorsement of the advisory panel or the Technology Assessment Board.

JOHN H. GIBBONS Director OTA Information Technology Research and Development Project Staff

John Andelin, Assistant Director, OTA Science, Information and Natural Resources Division

Fred W. Weingarten, Communication and Information Technologies Program Manager

Project Staff Donna L. Valtri, Project Director’ Fred W. Weingarten, Project Director’ Chuck Wilk, Senior Analyst Linda G. Roberts, Senior Analyst Linda Garcia, Analyst M. Karen Gamble, Analyst Prudence S. Adler, Analyst Jim Dray, Research Analyst Earl Dowdy, Research Analyst Lauren Ackerman, Research Analyst John Williams3 Peg Kay4

Administrative Staff Elizabeth A. Emanuel, Administrative Assistant Shirley Gayheart, Secretary Jennifer Nelson, Secretary Marsha Williams, Secretary Renee S. Lloyd, Secretary

Contractors Myles Boylan J. F. Coates, Inc. Barbara Davies Russell Drew Sidney Fernbach Michael Goetz Stan Liebowitz Abbe Mowshowitz Larry Olson Bruce Owen John Young

‘Project Director from November 1982 to December 1983. ‘Acting Project Director from December 1983. ‘Detailee from the Federal Communications Commission from October 1982 to October 1983. 4Detailee from the National Bureau of Standards, May 1984. iv Information Technology Research and Development Advisory Panel

Roger G. Noll Chairman Professor of Economics, Stanford University Geneva Belford Donald McCoy Department of Computer Science Vice President and General Manager University of Illinois CBS Technology Center Steven Bisset 1thiel de sola Pool* President Professor of Political Science Megatest, Inc. Massachusetts Institute of Technology John E. Bryson Paul E. Ritt, Jr. Morrison& Foerster Vice President and Director of Research GTE Laboratories, Inc. NandiKishore M. Chitre Director Larry W. Sumney Systems Planning Division Executive Director INTELSAT Semiconductor Research Corp. Ralph E. Gomory Victor Vyssotsky Vice President and Director of Research Executive Director IBM Corp. Research, Information Sciences Thomas J. Watson Research Center AT&T Bell Laboratories JohnV. Barrington Robert E. Wesslund Director Vice President for Technology Exchange COMSAT Laboratories Control Data Corp. William C. Hittinger George R. White Executive Vice President Senior Research Fellow RCA/David Sarnoff Research Center The Harvard Business School Bruce Lusignan Director Communication Satellite Planning Center Stanford University

Advanced Computer Architecture Workshop Duane Adams Sidney Fernbach Consultant Consultant Defense Advanced Research Project Agency Control Data Corp. James C. Browne Dennis Gannon Department of Computer Science Assistant Professor University of Texas Department of Computer Science Purdue University Alan Charlesworth Senior Staff Engineer Robert G. Gillespie Floating Point Systems, Inc. Vice Provost for Computing University of Washington Kent K. Curtis Section Head, Computer Science David Kuck Division of Mathematical and Computer Department of Computer Sciences Sciences University of Illinois National Science Foundation

*Deceased Mar. 11, 1984 Neil Lincoln Burton J. Smith Vice President and Chief Architect Vice President ETA Systems, Inc. Research and Development Department Denelcor, Inc. Roger G. Nell Professor of Economics Harold Stone Stanford University Professor Department of Electrical and Computer Paul Ritt, Jr. Engineering Vice President and Director of Research University of Massachusetts GTE Laboratories, Inc. Jim Thornton Paul Schneck Chief Executive Officer Leader, Information Sciences Division Network System Corp. Office of Naval Research Kenneth G. Wilson Jacob Schwartz James A. Weeks Professor of Physical Director Science Division of Computer Sciences Newman Laboratory Courant Institute of Mathematical Sciences Cornell University

Fiber Optics Workshop

Thomas Giallorenzi, Chairman Herwig Kogelnick Superintendent, Optical Sciences Division Director Naval Research Laboratory Electronics Research Laboratory AT&T Bell Laboratories Michael Barnoski Consultant Henry Kressel Pacific Palisades, CA Senior Vice President E. M. Warburg, Pincus & Co. Melvin Cohen Director, Interconnection Technology William Streifer Laboratory Senior Research Fellow AT&T Bell Laboratories Xerox Corp. Anthony J. DeMaria John Whinnery Assistant Director of Research for Electrical Department of Electrical Engineering and and Electrooptics Technology Computer Science United Technology Research Center University of California, Berkeley Charles W. Deneka Amnon Yariv Director, Optical Waveguide Technology Professor of Electrical Engineering and Corning Glass Works Applied Physics California Institute of Technology Charles Kao ITT Executive Scientist ITT Corp.

Software Engineering Workshop Richard A. DeMillo Capers Jones Professor of Information and Computer Manager of Programming Technology Science Analysis Georgia Institute of Technology ITT Corp. George Dodd Brian Kernighan Department Head Head, Computing Structures Research Computer Science Department AT&T Bell Laboratories General Motors Research Laboratories vi Ann Marmor-Squires Paul Schneck Chief Technologist Leader, Information Sciences Division TRW Defense Systems Group Office of Naval Research Robert Mathis Ben Shneiderman Director Professor of Computer Science Ada Joint Program Office University of Maryland Department of Defense Steve Squires Harlan Mills Software Technology Programming Manager IBM Fellow Defense Advanced Research Projects Agency IBM Corp. Terry A. Straeter Leon J. Osterwell Vice President and Program Director Chairman General Dynamics Electronics Computer Science Department Raymond Yeh University of Colorado Professor of Computer Science C. V. Ramamoorthy University of Maryland Professor of Electrical Engineering and Computer Science University of California, Berkeley

Artificial Intelligence Workshop Bruce Bullock Kamron Parsaye Vice President President Teknowledge Federal Systems Silogic, Inc. Marvin Denicoff Stephen Polit Executive Manager of Advanced Development Thinking Machines Corp. Artificial Intelligence Group Digital Equipment Co. Ewald Heer Program Manager Azriel Rosenfeld Autonomous Systems and Space Mechanics Director Jet Propulsion Laboratory Center for Automation Research Computer Science Center Thomas F. Knight University of Maryland Assistant Professor of Electrical Engineering and Computer Science David Waltz Artificial Intelligence Laboratory Professor of Electrical Engineering and Massachusetts Institute of Technology Research Professor Coordinated Science Laboratory Mitchell Marcus University of Illinois Member of the Technical Staff Linguistics and Artificial Intelligence Patrick H. Winston Research Department Professor of Computer Science and Electrical AT&T Bell Laboratories Engineering and Director Artificial Intelligence Laboratory Ronald Ohlander Massachusetts Institute of Technology Program Manager Defense Advanced Research Projects Agency

vii Information Technology Research and Development Federal Workshop Charles Brownstein Robert Powers Head Chief Scientist Information Science Division Office of Science and Technology National Science Foundation Federal Communications Commission Douglas Crombie John Riganati Chief Scientist Division Chief National Telecommunications System Components Division Information Administration National Bureau of Standards John A. Daly Paul Schneck Special Assistant Leader, Information Science Division Agency for International Development Office of Naval Research Craig Fields Elias Schutzman Assistant Director for Systems Services Program Director Defense Advanced Research Projects Agency Electrical Optical Communications Program National Science Foundation Clinton W. Kelly, III Assistant Director, Systems Science Division Defense Advanced Research Projects Agency Robert Lovell Director of Communications National Aeronautics and Space Administration

Reviewers and Other Contributors Robert Ayres Kenneth Brown Peter Fuchs Carnegie-Mellon University American Enterprise Institute for Harvard University Richard Baxter Public Policy Research Osmund Fundingsland British Embassy Glen Chancy General Accounting Office Donald S. Beilman AT&T Bell Laboratories Robert Gallawa Microelectronics Center of North Gary Coleman National Bureau of Standards Carolina Department of Education Howard Gobstein John Biddle Arthur Cordell General Accounting Office Computer and Communications Science Council of Canada Robert A. Gross Industry Association Matthew Crugnale Columbia University Marlan Blissett Gnostic Concepts, Inc. Walter Hahn University of Texas Beverly Daniel George Washington University Erich Bloch General Accounting Office Glen Hansen Semiconductor Research Corp. George Dummer Department of Commerce IBM Corp. Massachusetts Institute of Richard Hersh Justin Bloom Technology University of Oregon Technology International, Inc. Lauren Erling Christopher Hill Bill Boesman ADAPSO Congressional Research Service Congressional Research Service Frank Fisher Joseph Hull Jane Bortnick Urban Institute Department of Commerce Congressional Research Service Penny Foster B. R. Inman Paul Bortz National Science Foundation Microelectronics & Computer Brown, Bortz & Covington Donald R. Fowler Technology Corp. Martha Branstad California Institute of Technology National Bureau of Standards

VIII ——

Charles L. Jackson Joel Moses Dan Slotnick Shooshan & Jackson, Inc. Massachusetts Instituteof University of Illinois Don Kash Technology Ollie Smoot University of Oklahoma Jane Myers Computer and Business Patent and Trademark Office Equipment Manufacturers Todd LaPorte Association University of California-Berkeley Tom Nardone Bureau of Labor Statistics Bill Stotesbery Christopher LeMaistre Microelectronics &Computer Rensselaer Polytechnic Institute Dorothy Nelkin Cornell University Technology Corp. Jean-Pierre Letouzey Gregory Tassey French Embassy WilliamF.Nelson GTE Laboratories, Inc. National BureauofStandards Thomas Linney Henry Taylor Council of Graduate Schools Arnie Packer Manpower Economist Naval Research Laboratory Harold Linstone Al Teich Portland State University Paul Penfield Massachusetts Instituteof American Association for John G. Linvill Technology Advancementof Science Stanford University Robert Pepper William Wells John Logsdon Department ofCommerce George Washington University George Washington University Irwin White Albert Lumbroso Paul Polishuk New York State Energy R&D French Embassy Information Gatekeepers, Inc. Agency JohnM. Richardson Enrique Marcatili National Research Council Robert Willard AT&T Bell Laboratories Information Industries Michel Robin Association Michael Marcus French Embassy Federal Communications John G. Williams Commission Martha G. Russell Carruthers, Deutsch, Garrison& Nick Metropolis University of Minnesota Williams Los Alamos National Laboratory Grant Saveers Paul Wilson Charles H. Miller Digital Equipment Corp. AT&T AT&T Wendy Schacht Paul Zurkowski Robert Miller Congressional Research Service Information Industries University of Houston Jurgen Schmandt Association Robert Morgan University ofTexas Washington University

In-House Reviewers Audrey Buyrn, Program Manager Martha Harris, Project Director Paul Phelps, Project Director Henry Kelly, Project Director John Alic, Project Director Barry Holt,Analyst Eric Bazques, Analyst

ix —

Contents

Chapter Page l. Introduction and Summary ...... 3 Z. The Environment for R&D in Information Technology in the United States ...... 25 3. Selected Case Studies in Information Technology Research and Development ...... 55 4. Effects of Deregulation and Divestiture on Research ...... 111 5. Education and Human Resources for Research and Development ...... 139 6. New Roles for Universities in Information Technology R&D...... 169 7. Foreign Information Technology Research and Development...... 201 8. Information Technology R&D in the Context of U.S. Science and Technology Policy...... 279 9. Technology and Industry ...... 307

Index ...... 339

x Chapter 1 Introduction and Summary Contents

Goals for Federal R&D Policy ...... 3 Principal Findings...... 5 The Nature of Information Technology R&D ...... 6 Software as Technology ...... 7 Multidisciplinary Nature ...... 7 Close Boundary Between Theory and Application ...... 8 Complexity ...... $ ...... 8 Case Studies ...... 9 Advanced Computer Architecture ...... 9 Software Engineering ...... 11 Fiber Optics ...... 12 Artificial Intelligence...... 13 Issues and Strategies ...... 14 Impacts of Telecommunications Deregulation ...... 14 Scientific and Technological Manpower ...... 16 Changing Roles of Universities ...... 17 Foreign Programs ...... 18 Science Policy ...... 19 Summary ...... 21 Chapter 1 Introduction and Summary

American society is becoming increasingly new technological products and bringing them reliant on the use of electronic information to the marketplace. Since research and devel- technology, principally computer and commu- opment (R&D) is generally considered to be a nication systems. Economists and social scien- key element driving innovation (although not tists point out that information is a basic re- the only one), Congress is concerned with the source for society and an important factor of general health of R&D in information tech- production for the economy. Information technologies. Concerns include the effects that nology provides business, governments, and changes in the structure of the U.S. telecom- citizens with the basic tools necessary to communications industry are having on industrial municate and use information. Furthermore, R&D, and the implications of new foreign pro- the information industry, itself, including both grams intended to challenge traditional U.S. those that produce electronic hardware and market leadership in some areas of computers those that offer information services, is an in- and communications. OTA examined these creasingly important part of U.S. industrial and related questions at the request of the strength. House Committee on Science and Technology, the House Committee on Energy and Com- Because of this reliance on information tech- merce, and its Subcommittee on Telecommu- nology, public attention has been drawn to the nications, Consumer Protection, and Finance. process of innovation in that field, developing

Goals for Federal R&D Policy

Currently, public concern with information nology is demonstrated by the predomi- technology is focused on its role in promoting nant role the Department of Defense economic competitiveness. However, Federal plays in support of R&D in that field interest in promoting R&D and technological (nearly 80 percent of direct government innovation, dating back to the writing of the funding in this area is supplied by DOD). Constitution, has always been motivated by ● Provide for Social Needs: Government several distinct goals, only one of which has has always treated the telecommunica- been economic growth. These goals, which tions system as a basic social infrastruc- may for any specific technology or program ture. One major goal of communication either complement or compete with one another, policymakers is “universal service. ” New are applicable to information technology. services such as cable television and in- ● Support National Defense: A modem mil- formation services delivered over telecom- itary force is dependent on communica- munication channels can potentially im- tions and computer technologies for many prove the quality of life for American purposes, including intelligence, missile citizens. For example, OTA, in a recent guidance, command and control, and report, pointed out the role information logistics. Projected future applications in- technology could play in improving the clude battlefield management, fully auto- quality of and access to education and mated weapons, and computer-based arti- also stressed the need for additional R&D ficial intelligence advisors for pilots. The focused on educational technology. Another military importance of information tech- recent OTA report has described the po-

3 —

4 ● Information Technology R&D: Critical Trends and Issues

tential benefits of information technology ● Enhance National Prestige: The United to handicapped individuals. States has always been a world leader in ● Promote Economic Growth: The informa- computer and communication technol- tion technology industry (those who make ogies, both in the discovery of new basic and sell or provide access to communica- knowledge and in its development into tions media) and the information indus- commercial products. In certain areas of try (those who use the new technologies marketable technology, such as super- to produce and sell new information serv- computers (very powerful computers, the ices and products) are a growing part of largest and fastest available on the mar- the U.S. economy. Their economic impor- ket) or communication satellites, U.S. tance is felt both domestically and inter- firms have in the past held an undisputed nationally. These industries also have an technological lead. Although such techno- important indirect effect, in that the tech- logical dominance has nearly disappeared nologies and services they produce con- in most cases and could not have been ex- tribute materially to the economy in such pected to continue as other nations devel- forms as productivity growth, better oped research capability and industrial quality of products, and improved mana- strength in information technology, pre- gerial decisionmaking. The health of the serving U.S. technological competitive- information industries depends in part on ness and leadership is still an important their ability to bring forth new products public policy goal with implications that ex- and develop new applications; this ability, tend beyond purely economic advantage. in turn, depends on R&D. ● Support Civilian Agency Missions: Many ● Advance Basic Understanding of the civilian Federal agencies need to use ad- World: Research in computer and commu- vanced information systems to perform nication science has contributed to basic their missions more efficiently or effec- understanding in fields that transcend tively. For example, computers and com- pure information technology. For exam- munication systems underlie the space ple, research in artificial intelligence has exploration programs of the National led to insights into brain functions and Aeronautics and Space Administration human thinking and problem solving. (NASA). Weather researchers working for Communications researchers have made the National Oceanic and Atmospheric a Nobel Prize-winning contribution to our Administration (NOAA) and fusion re- understanding of the origins of the uni- searchers for the Department of Energy verse. The study of computer languages need access to the most advanced super- has both benefited from and contributed computers. Although the support of mis- to theories of the evolution and structure sion agencies for basic research is limited, of natural human languages. Similarly, in- they have funded development and, often, formation technology researchers have applied research directed to meeting their contributed to knowledge in such other specific needs. fields as fundamental mathematics, logic, Over the years, these Federal interests have and the theory of genetic coding. led to a wide array of policies, programs, and Moreover, computers have become in- agencies designed to promote the conduct of dispensable research tools in many areas research and development in the United States of science ranging from physics and as- and to appropriate its benefits to American tronomy to biology and social science. society. Many of them have helped stimulate Some scientists now speak of “computa- the development of computers and communi- tional research” as a new, wholly different cation technologies. form as important to the progress of knowledge as is experimental and theoret- The principal issue examined in this study is ical research. whether those policies, programs, and agencies Ch. I—Introduction and Summary ● 5 are now adequate and appropriate for advanc- their particular characteristics, and the rapidly ing information technologies in light of the escalating world competition in producing and emerging needs of society for these technologies, selling products and services based on them.

Principal Findings

In summary, OTA’s study of information national programs designed to foster in- technology research and development reached formation technology. the following conclusions: ● In response to these new pressures, indus- ● Information technology is an area in try support is growing rapidly for short- which the United States has historically term applied research and developmental shown great strength, both in basic sci- work, both within industrial labs and ence and marketed technology; and this through support of university work. In- strength has benefited the Nation in sev- dustry has generally looked, and still eral ways. It has contributed to the growth looks, to the academic community to per- of a strong economic sector, the informa- form and the Federal Government to fund tion industry. It has contributed to na- the long-term basic research that will tional security. It has stimulated the underlie future technological advances. evolution of basic science. ● Universities, traditionally viewed as Most areas of information technology ex- centers for basic research, are reexamin- amined in this study, including microelec- ing their roles with respect to applied re- tronics, fiber optics, artificial intelligence, search and are forming new types of rela- computer design, and software engineer- tionships with industry and government. ing, are still in the early stages as tech- State governments, in particular, looking nologies. Much improvement in techno- for stimuli to economic development see logical capability remains to be developed, new high-technology industry, including and that improvement will depend on the information industry, as interested in both fundamental research and techno- being located near strong university basic logical development. Hence, R&D will be research programs. Many states have an important factor in stimulating contin- formed new organizations within their uing innovation. universities in order to strengthen research, ● By most measures, U.S. research and de- focus it toward problems of interest to in- velopment in information technology is dustry, and facilitate the transfer of tech- strong and viable; however, those tradi- nology from the research laboratory to in- tional measures may not be realistic guides dustrial application. With most of these to the future needs of the United States experiments, it is too early to tell whether for R&D in these areas. In particular, in- they will be successful in stimulating increasing competition from foreign nations dustrial development and whether they (in particular, Japan) as well as the grow- will have beneficial or negative effects on ing intrinsic importance of information the universities. technology to the United States, suggest ● The Department of Defense is the pre- that more attention should be paid by pol- dominant source of Federal support of in- icymakers to the needs for R&D in infor- formation technology research and devel- mation technology. Many of these na- opment, providing nearly 80 percent of tions, motivated both by economic concerns the funding. Although some spillover to and broader social goals, have developed the civilian sector from DOD research can — —

6 ● Information Technology R&D: Critical Trends and Issues

be expected, its work is predominately erations against both first amendment focused on military requirements and can- rights and the possible negative impacts not be relied on to provide for all civilian on domestic R&D. needs. An important issue for considera- Instruments for scientific research are tion by Congress is whether increased growing more sophisticated and are be- funding by nondefense agencies of long- coming obsolete at an increasingly rapid term research in information technology pace. As a result, they are more expensive is needed to focus more work on civilian and yet must be replaced more frequently needs. if researchers are to keep at the forefronts ● There is substantial concern that techni- of knowledge. In particular, computer cal and scientific information flows be- software researchers need access to so- tween the United States and other coun- phisticated computational facilities and tries are unbalanced outward. Proposals data communication networks, and those to redress that balance focus both on in- working on microelectronics need access creasing access to foreign information and to chip processing facilities and computer- on restricting outflows of U.S. informa- aided-design tools. tion. Policies that would enhance the ac- Policies designed to stimulate information cess of U.S. information scientists to the technology R&D need to be evaluated for work of foreign researchers (through such possibly significant tradeoffs and exter- vehicles as translation services, travel nal costs in other areas. Some costs are grants, and foreign science exchange pro- clear. For example, R&D tax incentives grams) would help their own work. Such are a short-term loss of Federal revenues. programs would also help science policy Policies designed to draw more highly experts to evaluate the state of foreign trained people into information technol- technology more accurately. ogy may create manpower shortages in Steps have also been taken to tighten other areas of research. Devoting educa- controls over the export of technical in- tional resources to information technol- formation on national security grounds. ogy education and training draws them In considering such controls, Congress away from other areas of need, at a time needs to balance national security consid- when overall resources are not growing.

The Nature of Information Technology R&D

For purposes of this study, OTA defines the price. This high rate of technological improve term “information technology to refer to elec- ‘merit in microelectronics will continue into the tronic hardware that is used to create, com- 1990s and probably beyond. Changes in the municate, store, modify, or display informa- information technology products and services tion and to programming or “software” that available in the marketplace will likely be as is developed to control the operation of that revolutionary in the next decade as they have hardware. Modern information technology is been over the last one that has seen the ap- based on the microelectronic chip, the large- pearance and growth of such technologies as scale integrated circuit that over the last dec- the personal computer, fiber optics, satellite ade has transformed the computer and com- communications, the video cassette recorder, munications industries through steady and and two-way cable television. Industry will rapid increase of performance and decrease in compete worldwide to bring these new prod-

Ch. l—introduction and Summary ● 7

Photo credit: Arthur A. Merin, National Magnetic Fusion Energy Computer Center, Lawrence Livermore National Laboratory The use of computer modeling and graphics to study fusion energy. Outer magnetic surface of four-period helical axis stellarator (HELIAC) ucts and services to use, and an important ele- of images on a video screen. Research in infor- ment of that competition will be research and mation technology includes research on both development. hardware and software technology, and on the boundaries between them, for many computa- Information technology has characteristics tional techniques may appear in either hard- that distinguish it from many other areas of ware or software. R&D and that affect policies designed to stimulate and direct it. Software as an area of research presents Federal R&D support with some unusual problems. For example, intellectual property as ex- Software as Technology pressed in software is difficult to protect. It “Software,” or programs, are sets of instruc- can also be hard to distinguish between R&D tions that direct a computer in its tasks. The in computer programming, work which ad- term can also refer to information; for exam- vances the state of software technology, and ple, computer data bases, the information con- using existing techniques to write new pro- tent transmitted over communication lines, or grams and maintain old ones—a distinction that contained on a video disk. Because soft- that drew particular attention in the debate ware often can be the most complex part of over R&D tax credits, for example. an information system, research and development on software is considered to be of equal Multidisciplinary Nature importance to research on hardware technology. Yet, to some people, software does not R&D in information technology spans a “look” like technology. The concepts and tech- wide variety of disparate fields of work. For example: niques that underlie programs are intangible, and their embodiment is usually in a form such • Solid-state physicists work to improve the as a list of instructions on paper or a sequence performance of the semiconductor devices

38-802 u - 85 - 2 8 ● Information Technology R&D: Critical Trends and Issues

that form the heart of modern computer neers, but with a much broader base of re- and communication systems. searchers and users in many disciplines. ● Mathematicians study abstract theories of computation, looking for methods of Close Boundary Between faster calculations, for ways to write Theory and Application error-free software, and to transmit the maximum amount of information reliably Both the rapid pace of commercialization over communication lines. and the intrinsic nature of the technology re- ● Psychologists and brain physiologists sult in a short time lag between basic research find clues for writing “smarter” programs results and commercialization. Because infor- that can solve more complex problems. mation systems are so flexible and because ● Linguistic theorists contribute to the de- manufacturers are using powerful new design sign of new languages for programming tools to create new types of computers, less computers and build the foundations for time is required to implement a new concept. computer programs that can understand It may take only a few months for the results and automatically translate human lan- of a basic research project in pure mathematics guages. to be realized in a commercial software package. ● Users of computer systems, ranging from This close relationship between basic re- physical scientists to graphic artists, de- search and commercial application, possibly velop new types of software for their ap- most closely paralleled by biotechnology, cre- plications. In this development process, ates unusual and difficult issues for institu- they make fundamental contributions to tions, such as universities, that are tradi- software technology. tionally concerned only with basic research ● Sociologists, psychologists, and the management scientists search for the most ef- published in the public domain. These issues fective ways to design office automation include publication policies, intellectual prop- and decision support systems. Their re- erty rights, faculty conflicts of interest, competition with the private sector, and research search ranges from human factors, which priorities. Federal agencies that fund basic re- examines direct human-machine interaction, to investigations of organizational search, such as NSF, also find themselves deal- decisionmaking. ing with these issues more frequently. ● A few computer scientists, sociologists, Complexity historians of science, and experts in other fields, examine the broader questions of Modem computer/communications systems the interactions of information technol- are among the most complex technologies ever ogy with society. assembled by human beings. Computers consist of millions of logical subunits. Computer As a result of this breadth of subject area, programs can contain millions of instructions. relevant Federal programs of R&D support These complex machines, running complex are distributed among many Government programs, are then interlinked through com- agencies and among many organizational munications networks. To be able to under- parts of those agencies. The National Science stand, predict, and control the behavior of Foundation (NSF), for example, has at least these technologies requires a powerful theory four divisions, each in a different directorate, of complex processes. No such theory yet ex- that support research directly related to com- ists, although it remains a major goal of computers and communications systems. Another puter science. important implication is that manpower programs must be concerned not just with the It is because of this complexity that, even supply of “core” computer scientists and engi- though computers and communication sys- Ch. I—Introduction and Summary ● 9 terns are artificial devices—technologies cre- puters. Writing large, complex programs to do ated and built by human beings-there is basic unusual tasks is a common research method- science devoted to understanding their behav- ology aimed at understanding basic principals ior. This science has a strong experimental underlying the structure of problems and component to it. Many computer scientists software. and engineers need access to large-scale com-

Case Studies

Information technology is a broad subject significantly, new market possibilities have ap- area. To better understand and describe R&D peared for inexpensive special purpose hard- issues, four specific areas of technology were ware that takes advantage of these character- examined in more detail. These case studies istics. The general purpose computer system covered both software and hardware and both of the future may serve mainly as a routing computer and communications technologies. switch, sending a computing task to the appro- In addition, all four technologies have raised priate one of several different specialized proc- policy issues in the last few years. essors attached to it.

Advanced Computer Architecture Supercomputers, the label given to the most powerful machines on the market at any time, Although computers have increased in so- have received significant press attention late- phistication over the years, and individual ly. Until recently, U.S. firms have been the models differ from one another in detail, the only significant marketers of supercomputers. higher level logical design, called the “architec- However, Japanese firms have now brought ture, ” of most commercially available com- out supercomputers that, by some measures, puters is based on concepts that date back a seem to perform comparably to U.S. machines. few decades. In addition, Japan has embarked on two major supercomputer projects. One is designed New concepts of computer architecture now to develop the next generation of high-speed being explored in the laboratory will underlie numerical calculating machines; the other, the two types of innovation: 1) the development Fifth-Generation Computer Project based on of highly specialized, low-cost computer mod- artificial intelligence theories, is intended to ules that do specific types of tasks at extreme- produce a machine that performs reasoning ly high speeds; and 2) the future generations functions similar to human thinking processes. of supercomputers. Low-cost specialized computers are designed The Federal Government is considering sev- to perform specific types of computational eral responses to maintain leadership in this tasks very efficiently. When hardware is ex- technology. For example, the Defense Ad- pensive, as computer hardware has been in the vanced Research Projects Agency (DARPA) past, users usually try to allocate its cost over is beginning a “Strategic Computing Pro- a variety of applications. Hence, the so-called gram” to develop artificial intelligence capa- general-purpose computer, which performs a bilities for the needs of the Department of De- wide variety of computations relatively well, fense. The National Science Foundation has has been and continues to be the mainstay of embarked on a program intended both to in- the computer industry. However, most types crease their support of computer architecture of computational tasks have unique character- research and to put more supercomputer ca- istics. Since hardware costs have dropped pability in the hands of scientific users. Both

IO ● Information Technology R&D: Critical Trends and Issues

Photo credit: AT&T A microprocessor on a chip. This chip contains a entire computer, as powerful as some minicomputers

NASA and the Department of Energy have and easier access to supercomputers from the developed similar plans to increase the ad- United States than their counterparts at vanced supercomputing capability available American universities. to their researchers. In addition to providing access to supercom- If they were fully implemented, these Fed- puters for agency supported researchers, such eral plans would constitute an important re- Federal procurements would also stimulate sponse to the growing need for access to ad- the market for the next generation machine, vanced computing resources. Over the last a market that has traditionally been small and, decade, computing resources available to basic hence, unattractive to computer manufactur- researchers, particularly in universities, have ers. As the principal user of them, the Federal steadily fallen behind the state of the art. Al- Government has always had a major influence though the U.S. computer industry has tradi- on the supercomputer market. This Federal tionally led the market in the supercomputer stimulus may be particularly important for ex- field, basic researchers in other countries such perimental machines that are based on new ar- as West Germany and Japan may have more chitectural concepts. Markets and applications Ch. 1-Introduction and Summary ● 11 for these machines are particularly unpre- tem. Software was often given away free by dictable. computer manufacturers, and the costs of developing custom programs were much less Should Congress decide that additional Fed- than the costs of purchasing and maintaining eral emphasis on computer design is in the na- the machines. The situation is now reversed. tional interest, support of basic research should In most cases, costs of developing and main- also be expanded to provide the scientific base taining software are now the dominant costs for succeeding generations of advanced ar- of creating and operating large computer ap- chitecture computers, large and small. Al- plications. Although developments in micro- though R&D on the next generation supercom- electronics have steadily and rapidly reduced puters and other specialized architectures is the cost and increased the performance of com- underway in a few industrial laboratories, puter hardware, improvements in the produc- firms participating in this specialized market tivity of programmers has come much more tend to be small. Longer term basic research slowly. is usually left to be performed in academic laboratories and receives little industrial support. Since the limits of cost, reliability, and time Hence, Federal support of basic research can required to create new applications are increas- have significant effect. ingly dictated by software instead of hardware considerations, research directed toward im- In the past, responding to funding limita- proving the productivity of programmers and tions as well as the complexity of implementing designers will have high leverage. “Software computer hardware in a laboratory, agencies have usually kept basic research in machine engineering” is the term used to describe this area of research. architecture at the theoretical, or “paper and pencil” stage. This restriction has not allowed Research in software engineering ranges new concepts to be tested as hardware. To widely in content and approach. At one ex- allow such testing, some of these conceptual treme is highly theoretical work directed at de level basic research projects need to be funded veloping a fundamental understanding of the at a level that allows them to be carried to ex- nature of programs and “proving” in some perimental and even prototype stages. Micro- mathematical sense that they act as intended. electronics allows such hardware experimenta- At the other end of the spectrum are behavior- tion to be done more easily and cheaply than al and management scientists concerned with in the past, because prototypes can now be understanding how people interact with com- assembled from chips that form large logical puter systems and how to manage program- modules containing thousands of functions. ming projects. Not all research related to understanding programming and program Software has also been and continues to be behavior is called “software engineering”; a major problem area. Taking full advantage many fields of computer science and engineer- of a supercomputer’s speed and power requires ing contribute. the design of specialized programs. Signifi- cantly more research needs to be done on The Department of Defense, faced with an numerical computational techniques and pro- annual software expense estimated at $4-$8 gramming theory for new computer archi- billion, supports a large effort in software engi- tectures. neering. DOD has funded the design of a new programming language, ADA, and a “programming environment” called STARS that Software Engineering provides a set of automated tools for use by the program designer. These tools are intended The computer industry has transformed to be standards, required to be used in the de- over the few decades of its existence from be- velopment of DOD software. ing hardware dominant to being software dominant. In the early years, hardware was the Because the economic stakes are so high, re- most expensive component of a computer sys- searchers at industry centers such as the Bell 12 ● Information Technology R&D: Critical Trends and Issues

Laboratories and IBM’s T. J. Watson Re- can transmit far more information than can search Laboratories have also been active in copper wires or coaxial cables of the same size. software engineering research. University re- Since information is transmitted as light, fi- searchers in software engineering, supported ber optics are resistant to both electrical in- by Federal agencies, have concentrated on un- terference and eavesdropping. The potential derstanding better the theoretical underpin- capacity and economy of optical fiber is such nings of computer programs. that it not only will help keep communications costs down, but, over the long term, it will re- Other countries have also recognized the im- lieve the growing congestion of the radio spec- portance of software engineering. It has been trum and may ease some of the demands for made an explicit part of the British informa- satellite communications. tion technology research program. The Japa- nese, realizing that software has been a major Optical fiber is already used for communi- weakness in their attempts to be competitive cation lines between some major cities in the in the worldwide computer market, are experi- United States, and a transatlantic cable using menting with what they call “software fac- optical transmission will be laid in the late tories. ” The French have always had strength 1980s. By the end of the 1980s optical fiber in the area of programming, particularly the will be used heavily in nearly every stage of theory of programming languages. The De- the telecommunication systems, from local in- partment of Defense language, ADA, was de- traoffice networks to long-distance lines. veloped in a French laboratory. Although European firms are also develop- Although the potential benefits to the Na- ing capability in fiber optics, the principal tion of advances in software engineering are foreign competition has been from the Japa- high, the implications for Federal support are nese, whose technology is roughly at a par less clear. Some experts arguing that lack of with that available from U.S. firms. Since a fundamental research is not the key bottle- large world market for fiber optical hardware neck, point to the failure of designers to apply is growing, innovation in this technology is existing basic theories and research method- critical to U.S. competitiveness in telecom- ologies to the development of large-scale sys- munications. tems. To do so would require closer industry- The technology of the fibers, themselves, is university researcher interaction and might be still advancing rapidly along three general helped by the establishment of some large lab- lines: 1) techniques to increase the information oratory facilities to study design problems on capacity of the fibers; 2) improvements to the a realistic scale. transparency of the glass, which allow longer Other experts, pointing out that we still lack distances between “repeaters,” devices that the fundamental theoretical breakthrough nec- detect the signals and retransmit them on essary to develop a true discipline of software down the line; and 3) improved process tech- engineering, argue that the problems will per- nologies for manufacturing fibers. Hence, even sist until such a theory is established. Ex- though fiber optics is now in the commercial panded long-term Federal support of a broad marketplace, the need for basic research con- range of fundamental research on software is tinues and the current state-of-the-art is a long necessary to develop a sound theoretical basis way from any fundamental limits to this tech- to programming. nology. Applied research and development on fiber Fiber Optics optics, because it is expensive and capital- intensive, is concentrated in a few large lab- The last decade has seen the rapid develop- oratories; and, since the economic reward for ment of a new communications medium, infor- technological advance is clear, it is done pri- mation transmitted by pulses of laser light marily by industry. However, long-term devel- through thin glass fibers. These optical fibers opment of fiber optics will depend on funda- Ch. I—Introduction and Summary ● 13

include the ability to recognize and translate human speech, to prove the truth of mathematical statements, and to win at chess. For two reasons public attention has recently focused on an area of AI called expert systems. First, expert systems are seen to be the first significant commercialization of 25 years of research in artificial intelligence; and, secondly, they form the basis of the Japanese Fifth Generation Computer Project.

“ An expert system is an “intelligent” computer information retrieval system designed for use in a specific decisionmaking task, say medical diagnosis. It stores not only data, itself, but the rules of inference that describe how an expert uses the data to make decisions. By asking questions and suggesting courses - of action, the expert system interacts with a decisionmaker to help solve a problem. AI researchers now understand this process well

l ,. enough to be able to build profitable commercial systems for applications in very narrow areas of specialization, for example, advising on repair of telephone cable or diagnosing pulmonary disease.

Photo credit: AT& T Although few computer scientists doubt that sometime in the next century, people will Telephone lines circa 1887. Fiber optics is only the most recent of many technological developments that interact with computers differently than they have increased the capacity and shrunk the physical do now and that computers will show behavior size of communication lines that we would now call “intelligent,” they differ over the nearer term significance of expert mental research in materials and solid-state systems. Some maintain that expert systems electronics. Increased Federal funding of re- represent an interesting and profitable, yet search in universities would achieve the dual small, application area. In their view, AI con- objectives of stimulating fundamental research tinues to move slowly and is a long way from and training people at the Ph.D. level in fiber realizing the potential predicted for it. Others optics technology. A current lack of trained argue that expert systems are a model of how research personnel has been pointed out to be most computing will be done within a decade; a handicap to progress in this field. and, hence, that future U.S. leadership in computer science will depend on an aggressive pro- Artificial Intelligence gram of research in this field. “Artificial Intelligence (AI)” is a term that The Department of Defense has always been has historically been applied to a wide vari- a dominant supporter of AI research, includ- ety of research areas that, roughly speaking, ing work on expert systems. It has funded this are concerned with extending the ability of the work by establishing and funding large-scale computer to do tasks that resemble those per- laboratories at a few academic and nonprofit formed by human beings. These capabilities research centers. Civilian support is limited. ——

74 ● Iformation Technology R&D: Critical Trends and IssueS

NSF has steadily supported basic research in centers working in the field, and competition AI, but has done so at a relatively low fund- for limited human resources is sharp. In paring level, and NASA has funded work support- ticular, experienced faculty are being drawn ing applications to autonomous spacecraft. off campus just as demand for newly trained DOD experience suggests that AI research is graduates is growing rapidly. best performed in large laboratories with ac- The increased public attention and excite- cess to largc-scale computing facilities and col- ment about the promise of AI holds some lections of specialized software. Some areas of dangers to the field. If Federal and private research will begin to require access to special- support is contingent on high expectations for ized hardware and the facilities to design and significant short-term advances, failure to build new systems. meet those expectations could result in an This field in particular is experiencing an abrupt withdrawal of funding. Yet AI is a very acute faculty shortage. The new commercial difficult field of research, just in its infancy. excitement in A/I comes after a long period Much basic science remains to be learned in of relative disinterest by funding agencies. As AI, and steady long-term support is required a result, there are few researchers and research for that learning.

Issues and Strategies

The level and nature of R&Din all fields, in- rect issues, particularly as they relate to including information technology, is influenced formation technology. by many social, economic, and political factors beyond those specifically related to science Impacts of Telecommunications and technology. For example, the overall cost Deregulation of capital and the general health of the national economy will affect management’s abil- During the last few years, the fundamental ity and incentives to invest in R&D. Trade and structure of the telecommunications industry export policies that affect the access of domes- has changed. AT&T has been subdivided into tic firms to foreign markets or vice versa will several smaller (although still very large) in- influence the willingness of U.S. firms to independent firms. Many telecommunication vest in innovation. products and services traditionally provided by regulated monopolies are now provided Some observers have suggested that senior through a competitive marketplace. Improved corporate managers in many firms have down- innovation and international competitiveness played investment in R&D, which promises in the telecommunications industry is one of risky and long-term payoffs, in favor of short- the benefits intended by proponents of deregu- term gains. According to this view, although lation, and OTA expects such increased inno- these managers may have been reacting ra- vation will occur over the next decade. The tionally in terms of immediate economic incen- open question is whether that burst of innova- tives and the desires of stockholders, the re- tion will be at the expense of basic research sult of their decisions has been a low rate of that would lay the foundation for future inno- innovation and a long-term loss of competi- vation. tiveness. However, although technological development should increase, many science policy ex- Although these broad factors set an over- perts have worried about the impact of deregu- all environment for R&D decisions, Congress lation on basic research. In particular, they also deals with several specific policy issues view Bell Laboratories as a national scientific that directly relate to science and technology. resource and have expressed concern that a In this study, OTA concentrated on those di- new, competitive AT&T will have decreased

Ch. I—Introduction and Summary ● 15

Photo credit: AT&T Alexander Graham Bell opening the first telephone line to Chicago 16 ● Information Technology R&D: Critical Trends and Issues commitment to the basic research that has At the same time, the former Bell Labora- traditionally marked at least part of the Lab- tories was a major performer of information oratories’ activities. technology research in the United States, and the divestiture in the telecommunications in- OTA found that, although laboratory man- dustry is a structural and regulatory change agement will be admittedly more interested in unprecedented in scale. The impact of this the contribution of research programs to change on industrywide research including AT&T business, short-term effects on basic re- that done at AT&T Bell Laboratories and Bell- search will likely be small. In the first place, core should be closely monitored by Congress. most large firms in historically competitive information technology areas, such as computers, have a history of supporting long-term Scientific and Technological Manpower research. Furthermore, AT&T senior manage- The Federal Government has historically ment has a strong appreciation for the contri- assumed a role in stimulating the production bution that research at the Laboratories has of technical manpower in areas of national in- made to the growth of the telephone system. terest. Policy has focused on increasing both They are unlikely to make a sudden shift in the quantity and quality of technical gradu- policy in the foreseeable future. Finally, most ates, but it has also played an important role of the areas in which AT&T Bell Laboratories in equity. Federal scholarship and fellowship has particular scientific strength are those programs are vital mechanisms to eliminate that, although fundamental, are clearly related financial barriers to entry into technological to information technology-e. g., solid-state careers. physics and computer science. Since the line between basic and applied research in these Information technology is a fast-growing fields tends to be particularly hazy, fundamen- field of national importance, and Federal pol- tal work is likely to continue to be supported icies for developing R&D manpower in infor- even were management to adopt a short time mation technology may be deemed warranted horizon for results. by Congress. However, those policies should be broadly focused rather than aimed at nar- Long-term prospects for basic research at. row fields of specialization, and they should AT&T Bell Laboratories will depend, in part, be consistent and steady over the long term. on AT&T’s success as a competitor in the computer and telecommunications industry. This Any policy designed to increase technical long-term success will depend on many factors manpower will not take full effect for several including the regulatory environment, the eco- years. The delays are inherent in the educa- nomic climate, the skill of AT&T management tion system. They represent both the time it in operating under new rules, and the con- takes educational institutions to respond by tributions made by the Laboratories to in- developing new instructional programs and novation. hiring staff, and the time it takes for a student to pass through the system. Hence, these The divestiture has also introduced a new policies should be directed, not to short-term, and potentially important organization, Bell but to long-term needs. If addressed to cur- Communication Research (Bellcore), to the in- rent shortages, Federal programs run the risk formation technology R&D arena. Bellcore of overproduction by the time they take effect, serves the seven Bell operating companies, for needs may have changed or labor market which hold it jointly. It commands the re- forces such as high wages may have already sources of a significant number of researchers encouraged adequate new entrants in the field. from the pre-divestiture Bell Laboratories. Bellcore’s relationships with the operating Detailed long-term needs for technical man- companies and its research activities are in the power are hard to assess in a field changing genesis stage at present; thus, some time will as rapidly as information technology. Hence, be required to determine its overall contribu- projections of future requirements differ sig- tion to the national R&D effort. nificantly from one to another. They do not Ch. I—introduction and Summary ● 17 account well for the appearance of new fields Lawrence Radiation Laboratory at the Univer- of specialization, such as the recent emerging sity of California, Berkeley, and the Jet Pro- area of fiber optics, nor the sudden growth in pulsion Laboratory at the California Institute importance of older fields of research such as of Technology. artificial intelligence. Thus, policies that deal Once again, in response to some of the con- broadly with science and engineering manpow- cerns raised earlier in this chapter, many er in all fields are most likely to be successful. universities are experimenting with new struc- OTA found that, although some shortages tures designed with the dual objectives of im- exist in specific fields, the long-term outlook proving the quality of technical education on is promising. Enrollments at university pro- campus and improving the flow of research grams, both at the undergraduate and gradu- results into industrial innovation. In contrast ate level have been increasing rapidly, and to many past experiments, these new struc- educators from some institutions have ex- tures are, by and large, not in response to Fed- pressed concern that in a few years there may eral programs. Rather, many of them respond be an overabundance of computer scientists to initiatives both from industry looking for and engineers, at least at the bachelor’s degree closer ties with academic programs and from level. Similarly, as competition for admission States who see strong academic technical pro- to undergraduate and graduate programs grows, grams as attracting high-technology industry so should the quality of the graduates. and, thus, serving as stimuli for economic development. Science education policy also addresses equity issues. Undergraduate and graduate Some potential issues have arisen, but to technological education is increasingly expen- date problems appear to have been or are be- sive. Access barriers to high-paying careers in ing resolved to mutual satisfaction by nego- information technology could become formi- tiation between the academic institutions and dable for some parts of society, due both to the industrial sponsors. They include the fol- cost and poor preparation at the precollege lowing: level. To help offset these barriers, Federal pol- ● How intellectual property rights are dis- icy may need to address scholarships, fellow- tributed among researchers, their institu- ships, traineeships, and other forms of direct tions, and the industrial sponsors. assistance to students, as well as the improve- ● When and under what circumstances re- ment of precollege science education. search results may be published. ● Who establishes research priorities and Changing Roles of Universities standards of scientific quality. The United States has traditionally looked Over the longer term, other issues will also become important, such as the overall influ- to its university system whenever national ence on the directions of academic basic re- needs called for improved technological inno- search. Another long term concern is the po- vation and the development of new expertise. tential imbalance of campus resources and This dependence on the education system, attention between science and technology on probably unparalleled in the world, has often the one hand and other important scholarly resulted in experiments in institutional struc- fields such as foreign studies, social science, tures associated with higher education. The the arts, or the humanities. Morrow Act established the land grant colleges to develop a scientific basis for agricul- Another question of equity concerns the bal- ture and to train farmers in modern tech- ance between institutions, themselves. In the niques. During World War II and after, major first place, are these programs “zero-sum federally funded research laboratories were es- games” in which the gains of a few institutions tablished in association with universities. Ex- are mainly at the expense of all the others? amples include the Lincoln laboratories at the Secondly, although the top research institu- Massachusetts Institute of Technology, the tions have been able to negotiate arrange- 18 ● Information Technology R&D: Critical Trends and Issues ments that, in their view, do not threaten their tronics industry and developing technol- other academic roles, less influential schools ogy for social applications. will have far less bargaining power. ● ESPRIT, is a pan-European program intended to draw together the technical Since many of these institutional experi- resources of Europe to focus on R&D in ments are new, it is too early to tell whether information technology. they will be successful in meeting their spon- ● The British Alvey program, constituted sor’s expectations or whether the negotiated in part as a British response to the Japa- solutions to the major issues will work. How- nese Fifth Generation Computer Project, ever, it is important that Federal science agen- is a program of government support for cies watch these developments carefully for research and development in semiconduc- two reasons: tors and computer software. ● Many of these new research institutions Regardless of assessments of the specific will be performers of federally funded re- prospects for any one program or its implica- search or will be cosponsors of federally tions for U.S. policy, they demonstrate in total funded research. the new competitive world that is developing ● Many of the issues under negotiation be- in information technology. No longer can the tween universities and industry, such as United States expect to maintain unques- ownership of intellectual property rights tioned technological leadership and unchal- and restrictions on publication are cur- lenged domination of world markets in infor- rently echoed as problems in government/ mation technology. Other nations now see it university relationships. in their own interests to build scientific and industrial strength in these areas and are taking steps to do so. U.S. science and technology policy will need to adapt to take this new reality into account. Foreign Programs However, the foreign programs being estab- Programs in other countries designed to pro- lished do not necessarily constitute useful mote innovation in information technology models for U.S. policies, for several reasons. have received attention in the United States. In the first place, many of them are too new These programs have raised both concern to determine success. Secondly, each is tai- about increased competition in technologies lored to the unique patterns of government/in- in which the United States has traditionally dustry/academic relationships as they exist in been a world leader, and they have provided country of origin and may not be workable in models to those who suggest that the United the country. Finally, many of these programs States also needs to develop more coherent are designed to address particular bottlenecks programs to foster innovation by domestic in- to innovation that may exist in the specific dustries. Some examples of such programs are country. U.S. science and technology has a dif- the following: ferent balance of strengths and weaknesses. For example, some countries have a lack of ● The Japanese Fifth Generation Computer Project is an attempt to move beyond cur- venture capital, others a shortage of scientific rent concepts of computer design to devel- manpower, and still others have strong cultur- op an entirely different type of machine al barriers between business and academic in- based on artificial intelligence concepts. stitutions that impede technology transfer. ● The French La Filiere Electronique Pro- Information technology R&D is, in many re- gram is a five-year national information spects, an international activity. For example, technology R&D program with long-term some domestic firms are owned or partially goals of strengthening the French elec- owned by foreign firms (and vice versa). Many Ch. I—Introduction and Summary ● 19 large computer and communications compa- several programs administered by different nies are multinationals that operate labora- agencies. Despite occasional calls for centraliz- tories in several nations; and many companies ing R&D support within a single science and have agreements to exchange and cross-license technology agency, this historical approach technology with foreign counterparts. Scien- was accepted for two basic reasons. In the first tists have always viewed basic research as an place, agencies with specific technological international activity. Thus, international con- needs were considered to be best suited to sup- ferences, scientific journals, and exchange of port R&D focused on their needs. Secondly, researchers are common. multiple sources have been considered healthy Programs designed to strengthen and pro- for the support of science, since diversity and tect technology as a purely domestic resource redundancy are both important attributes of need to balance this goal against the natural scientific research. limitations posed by this international char- The issue of government organization is be- acter. For example, the managers of ESPRIT, ing reexamined, in the case of information a program designed to stimulate European technology. Arguments in favor of a more cen- technology, must decide whether and on what tralized and/or coordinated approach are based basis to admit or deny the participation of both on the changing nature of R&D and on multinationals based, at least partly, outside the challenges posed by new foreign programs of Europe. Companies concerned about what stressing R&D in fields such as microelectron- they may view as stringent controls over tech- ics, computer systems, and communications nology in one country can simply move their technology. R&D efforts to their laboratories in other countries. Research is becoming increasingly expensive. Research equipment such as a microelectronic fabrication facility or a supercomputer can cost several million dollars in capital costs, plus several hundred thousands of dollars a Science Policy year to maintain and operate. Salaries of research-level technologists are escalating, and Federal programs designed to stimulate in- much research in information technology now formation technology R&D exist in the much requires large teams of different specialists broader context of Federal science and tech- and technicians. This cost and complexity of nology policy. R&D not only may make redundancy and duplication of effort a luxury, but may even make Congress needs to ask whether existing some types of research impossible to initiate science and technology policy serves the needs without coordination of Federal programs. of information technology R&D both in terms of its unique characteristics and the potential Some foreign research programs targeted at importance of the technology to the nation. specific technologies have claimed successes Similarly, it is important to ask whether pol- in the sense of capturing a world market-the icies designed specifically to support informa- most notable being the Japanese program for tion technology R&D are consonant with the the so-called “64k” computer memory chip for broader scope of science and technology policy. computers that stores about 64,000 characters of information on a microchip. While Japanese OTA identified three major sets of policy success in this market may be attributed to issues that, although applicable to a broad many factors, the existence of a specific gov- range of science and technology, seem particu- ernment program may have been a major in- larly important in the context of information fluence. Seeing such programs targeted at technology. selected technologies in Japan and Europe has Institutional Focus. —The Federal Govern- led some experts to call for the United States ment has traditionally and purposefully sup- to respond with a similar approach to domes- ported research and development through tic R&D. 20 ● /formation Technology R&D: Critical Trends and Issues

There is no doubt that past DOD and De- partment of Energy support of R&D in such areas as microelectronics and computer design has been a spur to the entire information technology industry. However, there is evidence that, as the technology becomes more mature, military requirements have begun to diverge from civilian interests. For example, in the area of supercomputers and very high speed integrated circuits, there is a significant doubt that the DOD Strategic Computing and very high speed integrated circuit programs, although responsive to DOD needs, will also serve civilian needs. Hence, there may be a need to boost civilian support in information R&D. Influence of International Competitive- ness.—The growing concern over economic issues, in particular, international competitive ness has led to issues in which attempts to treat technology as a national resource that Photo credit: NASA can be held and protected conflict with tradi- A communication satellite being launched from a tional approaches to science policy. space shuttle. Satellite communication technology was stimulated by both Federal military and civilian For example, concern about the interna- R&D programs tional flow of scientific and technical information conflict with the historical science policy Such programs, if initiated, would not likely goal of establishing international cooperation be restricted to information technology alone, in science and technology. Open publication but would be focused on a wide range of tech- of some scientific material has been challenged nologies deemed critical. It would be a signif- on the grounds that the information should be icant departure from the past approach that restricted to the United States. Some have featured multiple government programs of suggested that the practice of admitting support for basic research, agency-by-agency foreign science students, who comprise a sig- support of applied research based on mission nificant percentage of degree candidates in needs, and, with a few exceptions, private sec- information technology, conflicts with U.S. tor support for innovation and product devel- economic interests. Restrictions have been im- opment. posed on foreign scholars attempting to attend U.S. conferences and seminars. Military and Domestic R&D.–Military funding tends to dominate the Federal R&D Other policies assume that the United States budget, particularly in the case of information does not need to draw on foreign science. technology. Such a high level of DOD involve- Hence, research project support typically ment is not surprising considering the impor- gives little attention to the potential benefits tance of information technology to its mission. of travel by U.S. scientists to foreign research The policy issue is whether the high level of laboratories and conferences. Unlike the prac- defense support provides adequate underpin- tices of other countries, little support is pro- nings for the broad development of civilian vided to provide translations and analyses of technology or whether a case exists for strong foreign research papers and monographs. Yet civilian agency support of R&D. U.S. scientists, in general, do not have facil- Ch. I—Introduction and Summary ● 21

ity in some of the significant scientific lan- information scientists do not even have facil- guages of today, particularly Japanese. Due ity with traditional scientific languages like to the weakening of foreign language require- French and German. ments in some graduate programs, many U.S.

Summary

During the past decade, information tech- to monitor and manage this change. In par- nology has grown to major economic and so- ticular, Congress— will need to: cial importance in the- United States and ● determine and maintain an appropriate abroad. The desire to maintain a competitive level and balance of R&D support for in- posture in worldwide markets as well as the formation technology (including research urge to realize potential social benefits of new on the social impacts of these tech- computer and communications systems has fo- nologies), cused public attention on the process of inno- ● remove unintended barriers to govern- vation, in particular, on R&D. OTA has found ment and private sector R&D efforts, that, in response to these pressures, the sys- ● assure that these changes in the R&D system has been remarkably adaptable. Federal tem do not have unintentional side effects programs are changing rapidly in response to new perceived needs; science and engineering on other sectors or goals of the U.S. system of science and technology, and enrollments are increasing in response to an ● anticipated need for more technological man- assure access by researchers in all disciplines to powerful new computational and power; and universities and industry are data communication technologies. changing their traditional patterns of research and relationships. The policy problem will be Chapter 2 The Environment for R&D m● Information Technology m● the United States Contents

Page Introduction ...... 25 Concepts for R&D ...... 27 The Roles of the Participants and the R&D Environment ...... 28 Federal Government Role in R&D ...... 28 Industrial R&D ...... 34 Universities’ Role in R&D ...... 37 Conflicts in Perspectives, Goals, and Policies ...... 42 Measures of the Health of U.S. R&D in Information Technology, ...... 43 Information Industry Profile ...... 43 U.S. Patent Activity ...... 45 A Synthesis: The Changing U.S. R&D Environment ...... 51 Observations ...... 52

Tables Table No. Page l. Federal Obligations for Total Research and Development: Fiscal Year 1984 ...... 28 2, Federal 0bligations for Basic Researching Information Technology-Related Fields ...... 29 3. Federa1 0bligations for Applied Research in Information Technology-Related Fields...... 29 4. The Top 15 in R&D Spending ...... 44 5. Federal and Department of Defense Obligations for Basic Research, Applied Research, and Development, 1983 ...... 45 6. Federal and Department of Defense Obligations for Basic and Applied Research by Field, 1983...... 46 7. Technology Distribution of the Top 50 U.S. Patent Electrical Categories 1978-80 ...... 46 8. Percentages of U.S. Patents Granted in Information Technology and in All Technologies, 1981 ...... 49 9. Foreign-Origin U.S. Patents in Some Components of Information Technology. . . 49 IO. Foreign-Owned U.S. Patents unselected Telecommunications Classes, 1982 ...... 49

Figures Figure No. Page l. Interrelationships in the R&D Process ...... 26 2. Science, Engineering, and Technician Employment Within High-Technology Industries, 1980...... 35 3. Federal 0bligations to Universities and Colleges for R&D Plant ...... 39 4. Estimated Ratio of Civilian R&D Expenditures to Gross National Product for Selected Countries ...... 45 5. U.S. Patents in Information Technology, SIC Codes 357, 365,366,367 ...... 48 6. Share of Foreign Patenting in the United States for the Three Most Active Countries in Selected Product Fields: 1981 ...... 50 7. U.S. Patent Activity of 10 Foreign Multinational Corporations, 1969-80 ...... 50 Chapter 2 The Environment for R&D in Information Technology in the United States

Introduction

“Information technology” is a generic term The world’s major countries are coming to for a cluster of technologies (discussed in detail view the development of high technology— in ch. 9) that provide automated capabilities and particularly information technology-as for a key to economic gains, important social objectives, and national defense and prestige. ● data collection; These countries have adopted national indus- ● data input; trial policies in information technology, in- ● information storage and retrieval; vesting hundreds of millions of dollars in the ● information processing; hope of achieving preeminence, both in R&D • communication; and and in commercial markets. This growth of ● information presentation. foreign competition2-especially from Japan– The information technology industry has be- has stimulated a concomitant growth of R&D come an integral component of U.S. industrial in the United States. strength. In common with other high-tech- l The trend toward internationalizing R&D, nology industries, the robustness of informa- manufacturing, and distribution is increasing. tion technology depends in part on a base of American companies are deciding that tech- research and development (R&D). However, nological strength can also be improved by several interacting factors are straining long- licensing technology from other domestic and established U.S. policies vis-a-vis research and foreign businesses, by acquiring equity posi- development: tions in firms with needed technology, and by ● rising costs and complexity of R&D; establishing R&D and manufacturing opera- ● intensive competition for both domestic tions in foreign countries in order to obtain and foreign markets; access to rapidly changing commercial ap- ● limited resources; and plications. ● accelerating technological advances. Figure 1 shows some important components This report describes those factors and ex- of the R&D process and diagrams their inter- amines their effects on the information tech- relationships. The remainder of this report will nology industry and its R&D base, raising focus on those components. questions both about current policies and ● about proposals for improving the competitive This chapter describes some of the key position of the United States in international players in the process and discusses some information technology markets. measures of health of the information technology industry. ● Chapter 3 Presents four case studies, each ‘The term high technology is used throughout this chapter dealing with an important element in the to refer to those industries characterized by a high proportion cluster of technologies that comprise in- of R&D expenditures per employee, or a significantly larger proportion of skilled workers than the industry average, and a formation technology. rapidly evolving underlying technological base. Thus computers and electronics are within the scope of this meaning; auto and ‘See also ht.ermtiond Competitiveness in EZectrom”cs (Wash- steel manufacturing are not, in spite of the recent trend to mod- ington, DC: U.S. Congress, Office of Technology Assessment, ernize and automate. OTA-ISC-200, November 1983).

26 ● Information Technology R&D: Critical Trends and Issues

SOURCE: Office of Technology Assessment.

● Chapter 4 discusses the recent divestiture ogenous elements: the science and tech- of AT&T in the context of its potential nology policies of foreign governments in effect on the R&D activities of Bell Lab- chapter 7; the science and technology pol- oratories. icies of the United States in chapter 8. ● Chapter 5 considers the availability of ● Chapter 9 describes the technological trained personnel to the R&D process. underpinnings of information technology, ● Chapter 6 examines new university-indus- the directions of key research areas, and try institutional relationships and their the characteristics of the information changing roles in the R&D environment. technology industry. ● Chapters 7 and 8 focus on some of the ex- Ch. 2—The Environment for R&D in Information Technology in the United States ● 27

Concepts for R&D

R&D includes a wide variety of activities– fuzziness has led to differing judgments as to ranging from investigations in pure science to which projects are applied and which are basic, product development. Because segments of and to confusion in data collection, because the information technology draw on so many terms are not applied uniformly by Federal science, engineering and other disciplines— agencies, industry, or academia. computer science, manufacturing, electrical, Apart from the difficulty in drawing a clean mechanical and industrial engineering, physics, line between basic and applied research, there chemistry, mathematics, psychology, linguis- is an additional problem in identifying the set tics—it is difficult to assign particular efforts of industries that collectively comprise the in- to the general category of information tech- formation industry. For purposes of this re- nology. port, we use the term to include electronics, As defined by the National Science Foun- computer, and telecommunications equipment dation (NSF), R&D is categorized as follows:3 manufacturers, providers of computer-based services, and commercial software developers. ● Research is systematic study directed toward fuller scientific knowledge or un- There are, then, two major areas of ambi- derstanding of the subject studied. guity in any discussion of information tech- ● In basic research the objective is to gain nology R&D: ambiguities inherent in designat- fuller knowledge or understanding of the ing an effort as “information technology” and fundamental aspects of phenomena and ambiguities arising from the overlap of basic of observable facts without specific ap- and applied research. Because of this, quan- plications toward processes or products tification of R&D efforts in information tech- in mind. nology is necessarily approximate and the ● In applied research the objective is to gain numbers cited in this report should be re- knowledge or understanding necessary garded as estimates, not as “hard” data. for determiningg the means by which a rec- One further term, “innovation,” should also ognized and specific need may be met. be clarified. As used in this report, innovation ● Development is the systematic use of the is a process that includes research and devel- knowledge or understanding gained from opment, manufacturing or production, and dis- research directed toward the production 4 tribution. The Nation’s innovative capacity of useful materials, devices, systems, or depends on the effective functioning of all methods, including design and develop- parts of the process. Success in the market- ment of prototypes and processes. place requires proficiency in some—not well The definitions of “applied” and “basic” re- understood—combination of those parts. There search are especially troublesome in a field as are other factors that influence marketplace dynamic as information technology, in which success, such as the timing of the introduction laboratory concepts evolve into marketable of commercial products, the influence of entre products very rapidly. In the area of artificial preneurs, and a variety of government policies. intelligence, for example, the work is often Thus, while excellence in research and devel- basic in the sense that it seeks new ways of opment provides no assurance of leadership understanding complex symbolic processes, in the commercial marketplace, it may very and applied in the sense that much of the work well be a necessary (if not sufficient) ingredient is directed at prototype applications. This of success.

‘National Science Foundation, FederaJ Funds for Research and Development Fiscal Yws 1981, 1982, and 1983, Volume ‘International Competition in Advanced Technology: Deci- XXXI Detailed Statistical Tables, NSF 82-326 (Washington, sions for Amen”ca, Office of International Affairs, National Rt+ DC, U.S. Government Printing Office, 1982), p. 1. search Council, 1983, pp. 21-22. 28 ● Information Technology R&D: Critical Trends and Issues

The Roles of the Participants and the R&D Environment

Industry, universities, and Government sor of research, the Federal Government funds (State, local, and Federal) are the three key roughly half of the total R&D carried out in contributors to R&Din the United States. The the United States6 and about two-fifths of the effectiveness of the Nation’s R&D is depen- research by the electrical machinery/commu- dent on the vitality of each of the participants nications industry (a key component of the in- and on their interrelationships. Federal sup- formation technology complex). In addition, port of R&D and Federal policies that main- the Federal Government itself performs about tain a healthy economy encourage industrial $11 billion of R&D7 (table 1) in its own and con- investment in R&D.5 A vigorous industry, in tract laboratories. turn, provides a large Federal and State tax Beyond that, the Government helps to shape base, making possible added support for aca- the environment in which private firms make demic institutions. For their part, well- their R&D decisions. In some cases, this is a financed academic institutions generate well- result of deliberate Government policy in- trained personnel as well as the dynamic knowl- tended to stimulate (or suppress) industry in- edge base necessary to fruitful R&D efforts. vestment. At other times, the environment is In addition to describing the roles of each affected by uncoordinated actions—intended of the participants and the environment for to serve other purposes-taken by a variety R&D, this section identifies some of the di- of Federal entities including the Federal Re- verse changes taking place in the R&D en- serve Board, the courts, regulatory bodies and vironment-those already in place, and others a plethora of executive branch agencies includ- in transition-that may profoundly modify ing the Departments of Justice, Commerce, some longstanding institutional patterns. State, and Defense, the National Security Council, and the Environmental Protection Federal Government Role in R&D Agency. The Federal Government plays several key Government Funding of R&D in roles in information technology R&D. As a Information Technology major user of information technology products (about 6 percent of the total automated data The Federal Government provided about 65 processing market and most of the market for percent of the funding in 1982 for R&D in supercomputers), its requirements are of conscience and engineering fields in the Nation’s siderable interest to the industry. As a spon- ———— ..— ‘Federal Support for R&D and Innovation, Congressional ‘David M. Levy and Nestor E. Terleckyj, Effects of Govern- Budget Office, April 1984, p. iii. ment R&D on Pn”vate Investment and %xiucti”w”ty: A Macro ‘Pmbalde Levels of R&D Exped”tums in 1984: Fomcas t and sconomi”c Analym”s, National Planning Association, revised Jan. Analysis, Columbus Division of Battelle Memorial Institute, 5, 1983, pp. 17-19. December 1983, p. 1.

Table 1 .–Federal Obligations for Total Research and Development: Fiscal Year 1984 (Estimated) (millions of dollars)

Extramurala United States and Territories

C C C FFRDCS FFRDCS FFRDC S administered administered Other administered Us. Industrial by industrial Universities by universities nonprofit by nonprofit State and local supported Total Intramuralb firms firms and colleges and colleges institutions restitutions governments Foreignd Total all agencies $45,497.0 $10,969.9 $22,957.4 $1,614.4 $5,270.7 $2,291.9 $1,335.5 $6832 $189.1 $1848 NOTES: aAll organizations outside the Federal Government that perform with Federal funds, bAgencies of the Federal Government. cFederally funded R&D centers. dForeign Citizens, organizations, governments, or international organizations, suchas NATO, UNESCO, WHO, performing work abroad financed by the Federal Government. SOURCE: “Federal Funds for Research and Development, Fiscal Years 1982, 1983, 1984, ” vol. XXXII, Detailed Statistical Tables NSF 83-319, p. 30 Ch. 2—The Environment for R&D in Information Technology in the United States ● 29 universities and colleges.8 Data collected by trical engineering. These categories alone the National Science Foundation (see tables amount to $933 million in basic and applied 2 and 3) indicate that Federal obligations for R&D funding. In addition, there are other basic research in fields related to information R&D areas related to information technology technology included $103.7 million for com- —e.g., mathematics, physics, and materials puter science and $115.4 million for electrical sciences. engineering in fiscal year 1984. For applied re- The Department of Defense (DOD), which search, funding levels are $145.8 million for has the largest of the Federal agency R&D computer science and $568.3 million for elec- budgets, is becoming increasingly dependent —. .— on electronics and computer science. By 1985, 8Nationa.1 Science Foundation, Early Release of Summary Sta- tistics on Academic Science/Engineering Resources, Division those fields will absorb nearly 25 percent of of Science Resources Studies, December 1983. Also see Fed- the total DOD R&D spending.’ Within DOD, eral R&D Funding The 197585 Decade, National Science Foun- dation, March 1984, p. 11. the Defense Advanced Research Projects Agency (DARPA) has heavily funded efforts in artificial intelligence, microelectronics, networking and advanced computer architecture. Table 2.—Federal Obligations for Basic Research in Information Technology-Related Fields No other agency compares with DOD, which (millions of dollars) accounts for about 60 percent of the Federal l0 Computer Electrical R&D budget. NSF, for example, accounts for Year science engineering Total about 3 percent of the Federal total, with a 1974 ...... NA $38.45 NA fiscal year 1983 appropriations of just over $1 1975 ...... 47.76 billion” and $1.32 billion and $1.5 billion in 1976 ...... $26.59 53.08 $79.67 12 1977 ...... 31.02 55.14 86.16 fiscal years 1984 and 1985 respectively. NSF 1978 ...... 40.28 57.41 97.70 funding, which is heavily weighted toward the 1979 ...... 42.96 62.03 104.98 “basic” end of the R&D scale, supports re- 1980 ...... 46.22 70,59 116.80 1981 ...... 52.21 78.51 130.71 search through grants, scholarships, univer- 1982 ...... 67.45 93.63 161.07 sity laboratory modernization, the establish- 1983a , . . . . . 80.25 91,89 172.14 ment of university-based “centers of excellence, ” 1984a ...... $103.66 $115.38 $219.04 aNatiOna} science Foundation estimates and similar programs. Many of the informa- NA—Not available. tion technology-related disciplines are funded SOURCE National Science Foundation, “Federal Funds for Research and De- velopment, Detailed Historical Tables” Fiscal Years 1955-S4, ” p 275 through NSF programs: communications, electrical engineering, optoelectronics, mathematics, physics, materials research, information Table 3.—Federal Obligations for Applied Research sciences, and so on. These information tech- in Information Technology-Related Fields nology-related fields accounted for over $90 (millions of dollars) million in fiscal year 1984. Computer Electrical Year science engineering Total Since 1972, NSF has been the primary Gov- 1974 ...... NA $230.79 NA ernment force in creating university-indus- 1975 ...... 239.20 try cooperative research centers, providing 1976 ...... $46.99 244.61 $291.60 some of their startup funds, planning grants, 1977 ...... 58.34 327.59 385.93 1978 ...... 66.97 375.22 442.19 and advice during their first years of opera- 1979 ...... 63.31 355.84 418.15 ..__ ——.—.— - 1980 ...... 82.38 446.56 528.93 Wited in a speech by Dr. Leo Young, Office of the Under Sec- 1981 ...... 69.32 478.17 547.48 retary of Defense for Research and Engineering, DOD, at the 1982 ...... 103.49 518.56 622.05 IEEE 1984 Conference on U.S. Technology Policy, Feb. 22, 1983 ...... 121.18 525.75 646.92 1984, Washington, DC. 1984a , . . . . . $145.85 $568.33 $714.18 Ioprobab]e Leve]s of R&D Experd”tures in 1984, op. cit. aNatlonal science Foundation estlfnates. “Federal Funds for Research and Development, Fiscal Years, NA—Not available 1981, 1982, and 1983, op. cit., p. 25. lZFi9c~ yea 1985 Nat,ion~ Budget Wti- SOURCE National Science Foundation, “Federal Funds for Research and De. Science Foundation velopment, Detailed Historical Tables: Fiscal Years 1955-84, ” p 327 mate to Congress. 30 . Information Technology R&D: Critical Trends and Issues tion. The centers are expected to become self- The Pattern of Government Funding of R&D supporting. 13 By the end of fiscal year 1984, in Information Technology NSF was involved with 20 of these centers and The Federal Government has had a long his- expects to make awards to at least 10 new tory of funding R&D in information technol- centers in 1985. ogy-related fields. It is currently the major A number of the centers are involved with sponsor of those types of information technol- information technology-related research. For ogy R&D in which it has special interests. example: These include artificial intelligence, supercom-

● puters, software engineering, and very large Rensselaer Polytechnic Institute’s Center scale integrated circuits (VLSI), all areas in for Interactive Computer Graphics is do- their technological infancy and with enormous ing applied research in computer-aided de- potential for military as well as commercial ap- sign (CAD). plications. There is a long list of related tech- ● North Carolina State University’s Com- nologies that have been stimulated by Govern- munications and Signal Processing Lab- ment-often defense or other mission agencies oratory is primarily engaged in basic re- —sponsorship of R&D including radar, guid- search. ance systems, satellite communications, and ● Both basic and applied research is being many others. performed at the University of Rhode Island’s Robotics Center. There are some historic examples of inten- ● Ohio State University is doing basic and sive Government sponsorship of technologi- applied research in robotic welding. cal development in areas where the potential ● Georgia Institute of Technology’s Center benefit was expected to be great, but the risks for Material Handling is engaged in ap- and costs of research were high and therefore plied research. unattractive to industry —e.g., computers, aviation and communications satellites. One The National Science Board has recommended of the classic illustrations of a successful, ma- broadened NSF support for engineering re- jor Government contribution to information search, an area long neglected in favor of the technology R&D is in the field of satellite com- agency’s traditional emphasis on basic scien- munications. The National Aeronautics and tific research. Congress approved about $150 Space Administration (NASA) (which current- million for NSF grants for engineering rely accounts for about 7 percent of the Federal search in fiscal 1985, about 10 percent of the 14 R&D budget) had the leading role in pioneer- agency’s total research budget. The Presi- ing technological progress toward commercial dent’s budget request for fiscal year 1985 calls development, accelerating the time frame for for more funding for this purpose, as well as the introduction of this technology, influenc- for establishment of cross-disciplinary engi- ing the structure of the U.S. domestic and in- neering centers at universities which would, ternational telecommunications common carrier among other effects, promote research on com- 15 industries, and effecting significant cost sav- puters and manufacturing processes. Gov- ings over the long run.17 ernment funding of engineering equipment and facilities at universities decreased from In these cases, the Government, through the $42 million to $17 million per year between undertaking of a number of risky and expen- 1974 and 1981,16 but has been rising since sive R&D programs and with extensive pri- then. vate sector involvement, developed a large pool of baseline technology that served to laDoD i9 aISO supporting this program through m $8 mill- prove the feasibility of geostationary satellite ion grant, primarily for laboratory equipment. “’’National Science Foundation Starts to Broaden Support of Engineering Research, ” The Chronicle of Higher Education, Jan. 18, 1984, p. 17, and interviews with NSF officials, Janu- 171Mofis Teub~ ad Edwud Steinmuller, Government POl- ary 1985. icy, Innovation and Economic Growth: Lessons From a Study “Ibid. of Satelh”te Comrnum”c.ations, Research Policy 11 (1982) 27-287, 16Probable Levels of R&D Expemh.tums in 1984, op. cit. North Holland Publishing Co. Ch. 2—The Environment for R&D in Information Technology in the United States ● 31 communications. These R&D programs were Other Federal Government Policies for the purposes of proving the feasibility of The Federal Government has many other various technological advances such as geosta- means for promoting (or suppressing) private tionary orbiting satellites, electromagnetic sector R&D activities including antitrust pol- propagation of signals from outer space, icy, patent policy, tax credits, technology traveling wave tubes, automatic station keep- transfer from Federal laboratories and feder- ing, and aircraft communications. The NASA ally funded R&D centers, and the promotion programs initiated to undertake the extensive of Research and Development Limited Part- R&D included the SCORE, ECHO, and RELAY nerships (RDLP). Export controls, whether for programs, the SYNCOM series of launches national security or political purposes, serve that paved the way for Intelsat I, the first as a negative influence in promoting private commercial communications satellite, and the sector R&D. A major source of corporate fund- Applications Technology Satellites series. The ing for R&D, international sales, is lessened, costs for the RELAY, ECHO, and SYNCOM and the open exchange of technical data is Programs alone through 1965 were over $128 limited. Six policies intended to promote pri- million-an amount that few companies could– or would—commit, particularly considering vate sector R&D are reviewed below. that the feasibility of synchronous satellite PATENT POLICY operation was seriously questioned.18 Previous policies assigning Federal owner- It is also interesting to note that these ship of patents based on Government-funded NASA programs likely had some important R&D have been modified in recent years with side-effects on the structure of the U.S. inter- the intent to stimulate patenting and commer- national satellite communications industry. cialization of invention. Public Law 96-517 Because AT&T was the only private company (1980), which permits small businesses, not- to have heavily invested its own funds for sat- for-profit institutions, and universities to ob- ellite communications R&D—with focus on tain patents based on Government-sponsored the nonsynchronous TELSTAR system—it is R&D, is intended to encourage university- likely that AT&T would have dominated the industry collaboration and patenting. The new international and domestic satellite com- Government’s right to patent ownership was munications services industry. Instead, the further reduced by Presidential Memorandum NASA programs, through continuous trans- (February 1983). This memorandum modified fer of technology to, and close interaction with, the Federal Acquisition Regulations by ex- commercial firms stimulated the competition tending the concepts of the current law to that followed the 1972 Federal Communica- allow all Government contractors to retain tion Commission’s decision allowing open en- patent rights. try into the domestic satellite communications services industry. There are obvious tensions in this situation, since it is sometimes argued that the public The market for the supply of satellite com- should own patents derived from research it munications equipment was also open to com- has funded. The counter-argument is that petition due to the expertise gained by NASA Government-owned patents tend not to be- contractors. In addition, the international sat- come commercialized and the public reaps no ellite network that evolved is owned and oper- real benefit from them. For example, Federal ated by INTELSAT, an international consor- efforts to license its patents have resulted in tium, with the U.S. portion owned and oper- a meager 4 percent being licensed, in contrast ated by COMSAT, a broadly based private/ with 33 percent for university-owned patents .19 public corporation.

19 Lm~ing Felker, Us. I)epartrnmt of Commerce, Office of Productivity, Technology, and Innovation, during interviews 181 bid., p, 277. with OTA staff, January 1984. 32 • Information Technology R&D: Critical Trends and Issues

Allowing universities and businesses to retain The Act has probably had an effect on the ownership stimulates commercialization, but activities of the Federal Laboratory Consor- may also have the effect of distorting the tium (FLC). In 1984, the FLC established an university’s traditional role as a developer of award for excellence in technology transfer fundamental knowledge. and issued 26 such awards. The Federal Lab- oratories, however, are mission-oriented; and TECHNOLOGY TRANSFER no Federal Laboratory has a mission empha- sizing the development of commercial technol- The Stevenson-Wydler Technology Innova- 24 - tion Act of 198020 was an attempt “to improve ogies. Thus, the concept of cooperative gen the economic, environmental, and social well- eric research laboratories envisioned by the being of the United States,” through such Act has not been tested. means as: establishing organizations in the executive branch to study and stimulate tech- TAX CREDITS nology; promoting technology development Tax credits for businesses performing R&D through the establishment of centers for in- have been expanded through the Economic dustrial technology; stimulating improved uti- Recovery Tax Act (ERTA) of 1981, which will lization of federally funded technology devel- expire in 1986 unless extended.25 A key pro- opment by State and local governments and vision of ERTA allows companies to claim 25 the private sector; and by other activities.21 percent tax credits for their qualified R&D The act has been selectively implemented. costs above their average expenditures for the Most of the Federal Laboratories have estab- prior 3-year period. The law also allows for in- lished Offices of Research and Technology creased deductions for manufacturer’s dona- Applications (ORTAs) which collect and dis- tions of new R&D equipment to universities, seminate the results of their respective Lab- and provides a new capital cost recovery sys- oratory’s research. The Center for the Utili- tem for R&D equipment (modified later by the zation of Federal Technology, in the National Tax Equity and Fiscal Responsibility Act of Technical Information Service of the Depart- 1982–TEFRA). ment of Commerce, serves as a central clear- Opinion is mixed as to whether this tax inghouse. credit is effective in stimulating R&D invest- However, the heart of the act, the Coopera- ment. One study,26 based on the limited avail- tive Generic Technology Program, has never able data, observes that the tax credit may been implemented. In February 1984, the Sec- well be helpful in encouraging increased R&D retary of Commerce issued the first report to budgets. However, a current study finds very the President and the Congress on the prog- little effect on increased R&D spending due ress of Federal activities conducted pursuant to the tax credit.27 Battelle Memorial Institute to the Act.22 It appears that much of the work attributes at least part of the increased indus- cited in the Report as “Stevenson-Wydler” try investment in R&D to the tax credits.28 activities would have been performed even if the act had not existed. For instance, the new patent policies discussed above and the R&D Limited Partnership (RDLP) discussed below .— were both cited as “Stevenson-Wydler’ initia- ~ioTA MemormdD, “l)weloprnent and Diffusion of COlll- tives.23 mercial Technologies: Does the Federal Government Need to Redefine Its Role?” March 1984, p. 26. “public Law 97-34, August 1981. ——— ‘Eileen L. Collins, An Early Assessment of Thins li&D In- ‘Public Law 96-480. centives Prow”ded by the Econonu”c Recovwy Tax Act of 1981, “For more details see the Stevenson-Wydler Technology In- National Science Foundation, PRA Report 8307, April 1983. novation Act of 1980, Report to the President and the Congress “preliminary findings of an ongoing study by Edwin Mans- from the Secretary of Commerce. February 1984. field, financed by the National Science Foundation. ‘gIbid. as~obab]e ~ve]s of R&D Expencli”tures in 1984, pp. 2, 11- “Ibid., p. 4. 12, op. cit. Ch. 2—The Environment for R&D in /formation Technology in the United States • 33

R&D LIMITED PARTNERSHIPS used. However, until recently businesses have The Department of Commerce has been pro- been exceptionally cautious about such ven- meting wider use of R&D Limited Partner- tures because of concern over litigation. ships (RDLPs), and offers advisory assistance Some of the questions that arise in consid- to businesses in their use. RDLPs are intended ering more liberalized interpretation of anti- to attract venture capital to commercial R&D trust legislation concerning joint research are: by limiting the potential losses to the venture capitalists while still permitting them to re- ● Will U.S. companies, long accustomed to tain patent rights, if any, and to have pros- performing much of their R&D individu- pects for receiving royalties or a subsequent ally, be able to adapt swiftly to a different buy-out by the company. RDLPs are some- mode of operation? What will be the real times used for conducting R&D with relatively commitment to shared research? How will short-term payoffs—e.g., 3 to 4 years. The use intellectual property issues be resolved? of RDLPs primarily affects the segments of ● Will there be new opportunities for collu- the innovation sequence from prototype through sion among joint R&D partners that re- product development. create historical antitrust problems? ● Will joint R&D dilute the benefits of com- During 1982, $275 million is estimated to petition even in basic research? Will small have been invested in RDLPs, mainly through firms be disadvantaged? large brokerage houses. In 1983, the amount is estimated at $490 million; in 1984, it was in the closing days of the 98th Congress, the $220 million.29 Although these investments National Cooperative Research Act of 1984 have tended to go for biotechnology, some was passed. The Act eliminates the treble dam- have been allocated to information technology age penalties in antitrust cases involving joint projects in fields such as computers, software, R&D ventures when those ventures meet the microelectronics, telecommunications, robot- conditions of the Act, in particular, by provid- ics, and artificial intelligence. The drop in funding prior notification to the Federal Trade ing in 1984 is believed by investment bankers Commission and the Justice Department. to be due to two trends. First, there is a general drop in investor interest in high tech- INDUSTRIAL POLICY nology. Second, some investors appear to be An important topic debated in Congress concerned about possible changes in tax laws concerns industrial policy and the appropriate that may give less favorable treatment of role of Government in strengthening industry. R&D tax deductions and capital gains. One approach would provide an environment generally conducive to industry reinvestment, ANTITRUST POLICY productivity improvements, and increased There have been administrative proposals competitiveness through tax, antitrust, pat- and congressional bills that would limit the ent, and other policies. A different approach use of the treble damage penalty against com- would assist selected industries, create a high- panies found guilty of antitrust violations and level industry-labor-Government advisory establish clearer guidelines for companies con- council, and provide loans and loan guar- sidering cooperative research activities. antees. There have been arguments noting that the Among the issues that surround the debates antitrust laws have not had a chilling effect are whether the Government could be effec- on cooperative research since they are rarely tive in selecting industries for support; whether businesses, without further encouragement, would invest their resources in areas most ZgData based on interviews by OTA staff with officials from the Office of Productivity, Technology, and Innovation, U.S. beneficial to the Nation’s competitiveness; and Department of Commerce, and key sources in the investment whether foreign national industrial policies banking community, Jan. 29, 1985. pose insuperable problems for U.S. businesses. —

34 • Information Technology R&D: Critical Trends and Issues

There may also be important lessons from the between market dominance and substantial various foreign experiences with targeted in- losses. The information technology industry’s dustrial policies, many of which may not be significant investments in R&D, the high mo- suitable as models for the United States (see bility of technical personnel, and the increas- Ch. 7: Foreign Information Technology R&D). ing internationalization of R&D, encourage rapid diffusion of technical data and frequent Industrial R&D introduction of new products. These characteristics intensify the need for legal protection The corporate motivation for performing of new ideas and products. R&D centers about the need to maintain or improve market share and profitability for both Certain areas of information technology are the short and long term. For high-technology especially vulnerable to “borrowing” and the businesses in general, R&D plays a critical– degree of legal protection available is uncer- although not singular-role in helping the firm tain. Software, for instance, can be copywriter to sustain or improve its competitive position. but cannot be patented except in certain instances. Policy is being made in the courts, vir- Funding and Licensing of Research tually on a case-by-case basis, and the re- The information technology industry in the sultant ambiguities satisfy no one. The entire United States spent about $10.8 billion for problem of intellectual property rights has be- R&D in 1983. As is typical of a high-tech- come a matter for national attention. nology industry, the firms in information technology often spend a large proportion of their Industry-University Links sales revenue on R&D—for supercomputer Chapter 6 of this report describes in detail manufacturers the ratio was nearly 20 percent the relationship of the information technology in 1982, and it is believed to have been com- industry and the universities. International parable in 1983. In 1983, overall R&D spend- competition is causing U.S. industry to being by computer manufacturers rose by 19.5 come increasingly sensitive to the importance percent; spending by computer software and of academia both as a performer of informa- service vendors rose by 38.9 percent; and tion technology-relevant basic research and as telecommunications R&D rose by 31 percent. a supplier of trained personnel. The importance of R&D to these firms can be seen from the fact that even during the recent The information technology industry is a economic recession they continued to make major “consumer” of technically trained per- substantial R&D investments. sonnel. As shown in figure 2, the office/computing and communications industries are ri- While information technology companies valed only by the aircraft and parts industry perform much of their R&D in-house, they also in terms of overall employment of scientists, make use of research originating in univer- engineers, and technicians. According to sta- sities, other companies, and the Federal Gov- tistics compiled by the National Center for ernment through licensing and other arrange- Education Statistics,30 some 21,400 electrical, ments for technology transfer. Licensing and electronic, and communications engineering cross-licensing are often used as means of ac- and 25,500 computer and information science quiring technology quickly and for recovering majors graduated in 1982. Within those disci- R&D expenses. plines, less than 2 percent are unemployed.31 Protection of Research Results There is some controversy surrounding the putative shortage of future manpower for in- Leadtimes in research and product develop- formation technology research (discussed at ment are very important for capturing markets, recovering R&D expenses, and contributing ~ONation~ Center for Education Statistics, Survey of Emned to profitability and further R&D investment. Degrees Conferred, reported to OTA by Dr. Vance Grant, Jan. 3, 1984. A 6-month leadtime in getting products into ‘] Congressional Budget Office, Defense Spendeng and the commercial markets can make the difference Economy, Table A-7, p. 59, February 1983.

Ch. 2—The Environment for R&D in Information Technology in the United States ● 35

Figure 2.—Science, Engineering, and Technician Employment Within High-Technology Industries, 1980 (Thousands) o 20 40 60 80 100 120 140

Machinery, except electrical Office and computing - Construction and related

General industrial

Metalworking Other

Electrical machinery Communication I Electronic components Electronic industrial appliances

Other

Transportation equipment Aircraft and parts J Guided missiles/space vehiclesa Motor vehicles Other

Instruments

Measuring control devices

Engineering/scientific equipment

Photographic equipment

Other

Chemicals Industrial organic

Plastic materials Industrial inorganic

Drugs Other

confidentiality rules and/or Statistical reliability data release SOURCE: National Foundation, Changing Employment Patterns of Scientists, Engineers, and Industries, 1977-80 36 ● Information Technology R&D: Critical Trends and Issues length inch. 7) but there is certainly no present tices), restrictions on foreign direct invest- oversupply of well-trained R&D personnel nor ments, exclusion of U.S. subsidiaries from is there doubt that industry is dependent on R&D programs funded by the host govern- the university system to produce the man- ment, preferential treatment of domestic pro- power necessary to maintain a sufficient level ducers in government procurement, and ex- of R&D. port credits.32 Industry is both influenced by, and influ- Jointly Funded Research ences, university training programs. In addition to its needs for traditionally trained Recently, the industry has made what may graduates, there is a growing need for gradu- be the beginning of a major shift from its tradi- ates with multidisciplinary training. For ex- tional pattern of conducting independent R&D, ample, companies involved in fiber optic com- toward undertaking some joint or cooperative munications require researchers trained in efforts. There are a number of examples in both physics and electronics-a combination which companies are jointly supporting basic which is not part of traditional curricula. A and some applied research through newly similar situation applies in the “expert” sys- formed cooperative organizations, which rely tems field, where a wider range of skills are heavily on university and corporate research- needed and broad scientific training is espe- ers. Among these new organizations are the cially valued. These shifting industry needs Microelectronics Center of North Carolina, the place demands on universities to alter their Semiconductor Research Corp., and the Micro- curricula and to create multidisciplinary in- electronics and Computer Technology Corp. A stitutional structures, and they increasingly detailed discussion of these arrangements and require frequent retraining of professional the policy issues arising from them is con- technical personnel. tained in chapter 6. In some cases, information technology firms These cooperative research efforts were requiring special skills not normally produced spurred by escalating R&D costs, by a per- by academia have compensated for the short- ceived limited supply of science and engineer- fall by providing additional cross-training for ing talent, and by the apparent erosion of in- employees, by helping selected universities to formation technology industry’s international develop new curricula, or by furnishing sup- competitive position. Some leaders in the in- plemental teaching staff. For example, in 1983 dustry argue that it has neither the resources IBM announced that it would make $10 mil- nor the time to continue its established pat- lion available to support university research- tern of across-the-board duplicative R&D. This ers and another $40 million earmarked for the does not mean that information technology development of curricula in computer-aided de- companies intend to slacken their competitive sign and manufacturing. R&D work vis-a-vis proprietary technologies. If anything, the cooperative projects are ex- Foreign Government Policies pected to lead to more innovation and more Policies and practices of foreign govern- competition at the level of the participating ments and companies can influence the prof- companies. itability of U.S. companies or deny them mar- Cooperative research programs require a kets, and thus effectively restrict their ability careful distinction between proprietory and to invest in R&D. These policies and practices nonproprietory technology. Nonproprietary include pricing exports at below cost in order technology is made up of: to capture larger market share and the advantage of scale, targeting specific advanced tech- ● generic technology, consisting of scien- nology markets through government-spon- tific and engineering principles that form sored industrial strategies, creating nontariff 32 For mom ~t~g, s Internat.iorlal Competition ~ Adv~~ barriers (e.g., discriminatory certification prac- Technology: Decisions for America, op. cit., pp. 28-37. Ch. 2—The Environment for R&D in Information Technology in the United States ● 37

a competitively neutral technology base the result of decades of dedicated and expen- that can be shared by all firms without sive research-efforts which few commercial reducing the potential benefits for any one firms would be willing to undertake. In each firm; and of the four areas of information technologies ● infratechnology, consisting of the knowl- selected for the chapter 3 case studies (ad- edge base necessary to implement product vanced computer architecture, fiber optics, and process design concepts. It includes software engineering, and artificial intelli- such things as basic data characterizing gence), universities have made and are con- materials, test methods, and standards. tinuing to make valuable research contribu- Like generic technology, the infratechnol- tions in technologies that the private sector ogy is competitively neutral. commercializes. The various cooperative arrangements are con- The intensity and breadth of university re- cerned with the nonproprietary technologies. search is dependent on a wide variety of factors, ranging from the prestige of the institu- Cooperative Government-industry develop- tion, graduate enrollment, ability to retain ment of nonproprietary technical standards is qualified faculty and researchers, adequacy of a related area important to industry. A recent funding for researchers, adequacy of facilities example is the cooperative effort of the Na- and laboratory equipment, affiliations with tional Bureau of Standard’s Institute for Com- major companies, and increasingly on interac- puter Sciences and Technology (ICST) and 12 tions with other researchers domestically and information technology firms in developing internationally. It is also dependent on a large and demonstrating networking technology for proportion of foreign graduate students– as office systems. A similar cooperative project many as 50 percent in some universities-paris aimed at developing networking technology ticularly in disciplines such as engineering. for the factory floor. These efforts are based on the development of nonproprietary stand- The intensity of university R&D is also de- ards. The programs, which began joint dem- pendent on the level of funding for research onstrations in 1984, permit the products of dif- provided by the Federal and State govern- ferent manufacturers to work together comments, as well as by industry. About 85 per- patibly and therefore expand the market for cent of the funding for university and college them. R&D came from external sources.33 Federal and State governments as well as industry In the long term, continued expansion of provide funding for research, for scholarships, U.S. cooperative research activities could have for laboratory equipment, and for real estate. policy implications for the appropriate amounts The universities accounted for about one-half and focus of Federal funds for R&D, for uni- of all basic research expenditures in 1984, with versities’ needs for outside support, and for 70 percent of their funding provided by the invigorating segments of the university re- Federal Government. search environment—and the potential for altering the status of U.S. R&D relative to Laboratory Research Instrumentation other nations. and Facilities During the past few years, problems con- Universities’ Role in R&D cerning the obsolescence of university laboratory research instrumentation and facilities, The exceptionally broad nature of the under- and a lack of access to supercomputer equip- pinnings of information technology, and the ment, have been recognized in many academic escalating complexity associated with con- disciplines. This problem is not specific to in- tinued advances based on research, indicate an increasing role for universities. Major ad- $~Nation~ &ience Foundation, Early Release Of summ~’ vances in fundamental knowledge are often Statistics, etc., table 1, December 1983. 38 ● Information Technology R&D: Critical Trends and Issues formation technology, but this field is among equipped with supercomputers.3g In part as a those affected. A decline in university research response, the Federal Government is increas- capabilities could result in a significant decline ingly sharing its supercomputers with its con- in the overall rate of the Nation’s scientific tractors, many of which are universities. Dur- advance. ing fiscal year 1984, $6 million was authorized for NSF to buy access to the equivalent of one One study notes that when the appropriate supercomputer for scientific use. This amount instrumentation needed to conduct specific re- was increased to $40 million in fiscal year 1985 search is unavailable, then research objectives 34 and will contribute to the establishment of six are altered to match that which is available. or seven supercomputing centers nationwide This source also reports that leaders within over the next 5 years. In addition, four U.S. the scientific community have estimated the universities have recently acquired supercom- cost of updating university research equip- puters. ment to lie between $1 billion and $4 billion. In particular, instrumentation with costs be- Further acquisitions of supercomputers by tween $100,000 and $1 million at U.S. research universities may be curtailed both by finan- universities is reported as becoming obsolete.35 cial limitations and by the difficulty of assem- One estimate stated that selected instrument bling the expert staff needed to maintain the costs have increased fourfold since 1970.36 facilities. Consequences of obsolescence are likely to include foregone opportunities to per- Compounding the problem is the fact that form frontier university research, less-than- the most up-to-date research equipment has 37 optimal training for graduate students, a con- a short lifetime-only 3 to 8 years. Another tinuation of the migration of faculty and new study which compared university laboratory graduates to industrial laboratories, and a de- equipment with that of industrial laboratories terioration in the quality of U.S. instrumen- found that the median age of university lab- tation, because university researchers tradi- oratory equipment was twice that of the equip- tionally provide valuable feedback and ment found in the laboratories of companies 38 innovative improvements to the instrument performing high-quality research. manufacturing community. The same study also noted that until re- A number of factors have contributed to the cently, not a single top-line supercomputer obsolescence of university laboratory equip- was installed in service at a U.S. university. ment. Among these are the long-term decrease A number of foreign universities, however, are in Federal funding for R&D plant in universities and colleges since 196540 (fig. 3); an ap- ‘—~4TeStim~y of Charles A. Bowsher, Comptroller General, GAO, before the Senate Committee on Commerce, Science, and proximately four- to six-fold increase between Transportation, Subcommittee on Science, Technology, and 1970-80 in the costs of state-of-the-art instru- Space, Research Instrumentation Needs of Universities, May 27, 1982. — ——— ‘sObsolescence of Scientific Instrumentation in Research ~gArnOng these me: West Germany’s Max Pl~* Institute UniversJ”ties, Emerging Issues in Science and Technology, 1981, and the Universities of Karlsruhe, Stuttgart, Berlin, and KFA; A Compendium of Working Papers for the National Science Japan’s Universities of Tokyo and Nagoya; England’s Univer- Foundation, National Science Board, p. 49. sities of London and Manchester France’s Ecole Polytech- ‘Su”ence, vol. 204 (1979), p. 1365, as reported in Obsolescence nique; and in Sweden, onehalf time access by universities to of Sci”entifi”c Instrumentation in Research Um”versities. a major auto manufacturer’s supercomputer. “International Competition in Advanced Technology: Deci- 40’’The Nation’s Deteriorating University Research Facilities, sions for Ameri”ca, op, cit., pp. 47-48. A Survey of Recent Expenditures and Projected Needs in Fif- ‘L. Berlowitz, R. A. Zdanis, J. C. Crowley, and J. C. Vaughn, teen Universities,” prepared for the Committee on Science and “Instrumentation Needs of U.S. Universities,” ~“ence, vol. 211, Research of the Association of American Universities, July Mar. 6, 1981, p. 1017,as reported in Obsolescence of Scientific 1981, p. 4. This survey covered 15 leading universities and six Instrumentation in Research Um”versities. academic disciplines.

Ch. 2—The Environment for R&D in Information Technology in the United States ● 39

120

100

0 1963 1965 1967 1969 1971 1973 1975 1977 1979 1980 19811982 Fiscal year SOURCE: National Science Foundation.

mentation41; and the short lifetime of state- identify critical areas affected, the trends or of-the-art equipment and high maintenance rate of change, and likely influence of new ini- costs. In addition, during periods of decreased tiatives to relieve the problems. funding, university laboratory administrators A GAO investigation into the instrumenta- tend to forego instrumentation purchases 42 tion of obsolescence issue found “a tremen- rather than reduce project staffs. Further, dous lack of information”: an absence of trend until recently, Federal funding for research data on nationwide research equipment ex- projects did not allow for purchase of instru- penditures by universities, a lack of consen- mentation if it was to be shared with other sus on university laboratory needs, and no projects. comprehensive indexes that would measure 43 44 Quantification of the Problem changes in the price of equipment or the costs to maintain it. GAO also found that the Until recently, much of the information con- rapidly increasing costs for instrumentation cerning the extent of the obsolescence prob- in conjunction with relatively level funding of lem has been anecdotal. However, there are basic research (in constant 1972 dollars) at now a number of initiatives to provide statis- universities and colleges for the period 1968- tical data quantifying its scope and the effect 81 combined to have a “large effect on re- on specific disciplines. In addition to the need for suitable data collection, there is a need to ..— [GAO testimony]. —. ——.— illustration of cost escalation is the $100,000 ‘l’’ Revitalizing Laboratory Instrumentation, The Report of electron microscope of the 1960s, which could distinguish ob- a Workshop of the Ad Hoc Working Group on Scientific In- jects smaller than of a meter. By the scan- strumentation, ” National Research Council, March 1982, p. 1. ning transmission electron microscopes had improved the reso- This source estimates a sixfold increase in costs, while others lution by a factor of 1,000, and cost more than $1 million. estimate a fourfold increase. (Testimony of Dr. Edward A. Knapp, Director, NSF before the ‘Testimony of Charles A. Bowsher, Comptroller General, House Subcommittee on Science, Research, and Technology, GAO, op. cit. Feb., 1984.)

38-802 0 - 85 . 4 40 ● Information Technology R&D: Critical Trends and Issues

searchers’ acquisition and maintenance of re- ● The replacement value of all instrumenta- search equipment. ” tion in use was estimated at 42 percent

45 above the original purchase price (almost A National Science Foundation study sur- matching the inflation rate). veyed the status of university laboratory re- ● Two-thirds of all research instrument sys- search instrumentation in 1982 in three ma- tems in use in 1982 were acquired partly jor disciplines. The study, when completed in or entirely with Federal funds. 1985, will provide some useful insights into the extent of the problem nationwide. The study These preliminary findings indicate the need polled 43 academic institutions concerning the to develop data providing a comprehensive condition of their instrumentation in computer picture of university research instrumentation. sciences, physical sciences, and engineering The 43 universities surveyed account for 94 disciplines in order to develop national esti- percent of the R&D expenditures in each of mates of the findings. Preliminary data from the three disciplines (computer sciences, phys- the study, which covers equipment with pur- ical sciences, engineering) covered and had in- chase prices ranging from $10,000 to $1,000,000 strumentation inventories that cost nearly $1 in use in 1982, may not either confirm or refute billion-a significant portion of which was the notion of a serious problem with instru- funded by the Federal Government. Exactly mentation obsolescence in these three fields, how much total funding is needed to equip the but does seem to demonstrate that the prob- laboratories adequately is not known. How- lem may not be uniform. For example: ever, it is possible to make some very approximate, inferential estimates based on the avail- ● University officials classified 26 percent able data. For example: given an instrument of the research equipment listed in these inventory of $1 billion and assuming that the fields in 1982 as “not in current use. ” equipment has a 4-year lifespan, one-quarter Some portion of this undoubtedly is obso- of the equipment ($250 million current dollars) lete. Seventeen percent of the laboratory would be needed annually to upgrade the equipment associated with computer sci- equipment assigned to those three disciplines ence research was obsolete; 24 percent of in the 43 universities. the physical sciences and engineering equipment was obsolete. Remedial Activities ● One-half of all of the academic research instruments in the fields surveyed that Federal agencies, State governments, and were still in use in 1982 were purchased industry have begun to address the instrumen- during the 1978-82 period. Only 12 per- tation problem. For example, the Department cent of the computer science instrumenta- of Defense initiated a $150 million 5-year pro- tion was purchased prior to 1972, and 78 gram in fiscal year 1983 to fund instrumenta- percent was purchased during 1978-82. tion in areas of research in support of its mis- ● Concerning state of-the-art equipment, 98 sion. DOD’s University Research Instrumen- tation Program is based in part on a 1980 percent of the computer science equip- 46 ment in this category had been acquired study of the instrumentation needs of U.S. since 1978, compared with 80 percent university laboratories to conduct defense- of the engineering research equipment. related research. The pervasiveness of the Eighty-four percent of all of the state-of- problem is illustrated by the estimated 2,500 the-art equipment surveyed was listed as responses from the academic research commu- in excellent condition, as compared with nity to an initial DOD invitation for proposals 42 percent of all equipment covered by the for funding. survey.

‘American kmciation of Universities Report to the National “’’One Fourth of Academic Research Equipment Classified Science Foundation, Scientific Instrumentation Needs of Re Obsolete, ” Science Resources Studies Highlights, NSF, 1984. search Universities, June 1980. See also 13erlowitz, et. al., op. cit. Ch. 2—The Environment for R&D in Information Technology in the United States ● 41

In addition, NSF’s appropriation increased State Government and Industry Initiatives from $195 million in fiscal year 1984 to about Among the various State government ini- $234 million in fiscal year 1985 for support of tiatives to improve university research capa- advanced instrumentation. Some $122 million bilities are those of North Carolina, Massachu- in fiscal year 1985 (up from $104 million in setts, New York, California, and Minnesota fiscal year 1984) will be allocated to research (see ch. 6). instrumentation for individual project grants, and the remainder for instrumentation for Industry is also contributing at a significant multi-user regional instrumentation centers level to academic information technology re- and major equipment in national centers. search and education. For example, seven computer vendors alone have made recent commit- The universities are also concerned about ments to contribute some $180 million in cash obsolete or inadequate research facilities, or and equipment to universities. One source buildings–which Federal agencies have not “conservatively” estimates the level of dona- funded since the 1960s. One preliminary esti- tions of computer equipment to higher insti- mate of the funds needed to fully upgrade fa- tutions of education to exceed $100 million in cilities at the Nation’s major research univer- 1982. Among the major contributors were sities is between $990 million and $1.3 billion 47 IBM, Digital Equipment Corp., Apple Com- per year. NSF is leading the interagency puter, Inc., Hewlett-Packard Co., Wang Lab- Steering Committee on Academic Research oratories, Inc., NCR Corp., and Honeywell, Facilities (which includes representation from Inc. DOD, DOE, the National Institutes of Health, and U.S. Department of Agriculture) to ad- Two other recent examples further illustrate dress this issue. The committee is expected to the trend: Brown University built a $1.5 mil- recommend that a study be initiated to clarif y lion computer science facility based on contri- requirements for additional buildings or mod- butions from IBM, Xerox Corp., Gould, Inc., ernization programs for the Nation’s academic and others; the University of California at research institutions. In addition, the National Berkeley has commitments of $18 million in Science Board addressed this issue during its cash and equipment from firms such as Fair- June 1984 session and recommended that fund- child Camera & Instrument Corp., Advanced ing for facilities become a component of the Micro Devices, Bell Laboratories, Digital fiscal year 1986 NSF budget. It recommended Equipment Corp., GE, Harris, Hewlett-Pack- that NSF conduct pilot programs for R&D fa- ard, Hughes Aircraft, IBM, Intel, National Semiconductor, Semiconductor Research Corp., cilities construction in three areas of priority 48 research (large-scale computing, engineering Tektronix, Texas Instruments, and Xerox. research centers, and biotechnology), and that Changing University Role NSF support the Committee by obtaining improved data on the condition of university fa- The role of university research maybe at the cilities. The House Authorization bill for the threshold of significant change. Faced with the fiscal year 1984 budget of the Department of increasing expense and risks associated with Defense directed that agency to determine the research, a limited supply of trained personnel need to modernize university science and engi- (especially in needed multidisciplinary skills), neering laboratories for national security pur- and intensifying competition, U.S. industry is poses. Congress has requested NIH to make taking steps to bolster the universities’ role a similar determination with respect to its in the performance of research in information mission.

47 Adequacy of Academic Research Facilities, A Brief Report 48These donations are seen as motivated by business strate- of a Survey of Recent Expenditures and Projected Needs in gies, and to some extent, by the 1981 changes in the Federal Twenty-Five Academic Institutions, National Science Founda- tax regulations which provide tax advantages for donations of tion, April 1984. new equipment to schools. 42 . /n formation Technology R&D: Critical Trends and Issues technology. State governments are promoting despite some differing perspectives among their universities’ research capabilities to at- participants, and some discords in goals and tract technology-intensive industry. policies. For example: The National Science Foundation’s support 1. National economic goals for improving of engineering research at universities is also productivity and competitiveness in inter- contributing to the change. With a $150 mil- national markets are supportive of a lion appropriation for fiscal year 1985, NSF healthy information technology industry. will increase the range of engineering projects However, productivity improvements are supported and will help to establish more uni- viewed by some as possibly resulting in versity engineering centers to promote re- fewer employment opportunities, and search on computers, manufacturing proc- lower job skill requirements and pay esses, 49 and other nationally important tech- levels. nologies. Many of these joint activities are em- 2. National science endeavors are fostered phasizing strengthened linkages among the by measures that increase fundamental ties, promoting entrepreneurship, and improv- knowledge, consistent with university ing the overall scientific and technological and scientists’ objectives-including the base of State and local communities.60 They sharing of research results international- are also serving to add to both the supply and ly—but may be contrary to national secu- the quality of degreed professionals, and to rity objectives of controlling technology modernize the tools available in participating transfer and industry concerns for pro- university laboratories. tecting research data. 3. Universities’ and scientists’ interests in The universities may find themselves in a conducting undirected (basic) research position in which they are looked to as criti- may be in conflict with industry’s and cal to U.S. competitiveness in domestic and mission agencies’ need for achieving spe- world markets, to our ability to maintain tech- cific results. The current trend toward in- nological prominence and to remain reason- creased university-industry collaboration ably self-sufficient in critical areas for national and toward university patenting may re- security purposes. Undoubtedly, for these and sult in more directed university research other reasons—such as the growing need for and less independence of universities. life-long education for many professionals– 4. U.S. policies toward opening university there will be forces for change in the role of admission to foreign students have been universities. enormously successful, but other regulations encourage emigration of aliens after Conflicts in Perspectives, Goals, graduation, thus depriving U.S. firms and and Policies academic institutions of needed talent. These various participants-academia, in- The most striking observation concerning dustry, and Federal and State governments– the roles of the various participants is that work together in a sort of dynamic balance, they are in a state of flux. To date, the directions of the changes appear to be: 1) modified interrelations among the participants in the 49Nation~ &ience Foundation starts to broaden suPPort of engineering research, The Chrom”cZe of I+@her Education, Jan. R&D process, 2) a significantly larger role in 18, 1984, p. 17, and updated by NSF officials, January 1985. research for participating universities, and 3) ‘For more detailed information, see Technology, Innovation, a potential strengthening of national capabil- and Regional Economic Development, Background Paper No. 2, Encouraging High Technology Development, Office of Tech- ities to conduct R&D in information technology Assessment, February 1984. nology. Ch. 2—The Environment for R&D in Information Technology in the United States ● 43

Measures of the Health of U.S. R&D in Information Technology

There is no single indicator of the health of (SIC) for manufacturing companies without R&D in information technology. However, a accounting for the significant revenues from combination of indirect indicators can provide services. One statistical comparison will serve an impression of its overall vigor in the United to illustrate the problem of noncomparability: States. Indicators of industry growth include In 1982, shipments of all electronic computing the level of funding for R&D, the availability equipment establishments totaled $34.1 bil- of trained personnel, trade balances for infor- lion; in the same year, data processing reve- mation technology exports and imports, and nues of the top 100 information technology patent trends. Although the indicators used companies totaled $79.4 billion.sl are not comprehensive, taken together they Because of these data inconsistencies the in- portray a robust industry with significant formation presented is skewed by the noncom- growth in sales and investment in R&D. They parable databases; and, for that reason, quan- also show that these industries account for the tifications can only be regarded as approximate. employment of a high proportion of the Na- tion’s technically trained work force as well as a substantial proportion of its industry- and Information Industry Profile government-funded R&D. Paradoxically, they The review presented in chapter 9 of busi- provide a varying contribution to the U.S. ness statistics for the U.S. information tech- trade balance, and a declining proportion of nology industry indicates that this industry the total number of information technology is generally robust as measured in a variety patents granted in the United States. of ways, and in comparison with U.S. indus- These indicators, while generally promising try as a whole. For example, for the 1978-82 in themselves, provide far less than a complete period: sales revenue grew by 66 percent com- picture of the state of health of U.S. R&D in pared with 40 percent for the composite U.S. the information technology industry. Interna- industry; profits grew by 36 percent compared ational competition in both information tech- with 6 percent for the composite. Profits- nology markets and in R&D is intensifying to-sales ratios were about 9 percent v. 5 per- and U.S. leadership in many of these areas is cent; and the growth in the number of employ- being seriously challenged. Also, as described ees averaged 12 percent v. a negative 8 per- previously, aspects of the R&D process are in cent.52 flux and it is too early to tell whether the Concerning R&D, the information technol- changes are for better or for worse. In addi- ogy industry is also vigorous. R&D expendi- tion, while this report focuses on information tures compare very favorably to the compos- technology R&D, several other factors play ite industry when measured as a percentage critical roles in U.S. competitiveness. These increase over the period, as a percentage of include marketing strategies, manufacturing sales, or in terms of R&D expenditures per em- capabilities, and global macroeconomic and ployee. This industry accounted for 28 percent trade conditions. of the total R&D spending by all industries. Beyond that, as noted earlier in this chapter, the “information technology industry” is an 51Based on a draft report, ‘l’he Computer Industry ~d ~n- ill-defined entity and the available statistics termtiomd Trade: A Summary of the U.S. Role, by Robert are often noncomparable. Much of our infor- G. Atkins, Information Processes Group, Institute for Com- puter Sciences and Technology, 1984. mation is based on statistics pertaining to a ‘zThese data primarily represent large firms. See table 52 (ch. small subset of the Standard Industrial Code 9) for limitations. —

44 ● Information Technology R&D: Critical Trends and Issues

In fact, a listing of the 15 U.S. companies that as a percentage of Gross National Product (ex- spend the most on R&D—as a percentage of cluding expenditures for defense and space) .66 sales, and in dollars spent per employee-is al- Figure 4 shows that both Japan and West Ger- most completely populated with information many have been outpacing the United States technology companies such as Cray Research, (as well as France and the United Kingdom) Telesciences, Advanced Micro Devices, LTX, by this measure for more than a decade. Both Amdahl, Computer Consoles, and Convergent of these countries have relatively small R&D Technologies53 (table 4). expenditures for defense and space purposes as a percentage of GNP—e.g., 2.5 and 5.6 per- The electrical and communications industry cent in 1981 for Japan and West Germany, re- increased R&D budgets by approximately 13 spectively, in contrast with 31 percent for the percent in 1984 and 1985. These increases United States, 29 percent for France, and 30 resulted in large part from R&D on semicon- percent (in 1975) for the United Kingdom. ductors and telecommunications. During 1984 and 1985, electrical and electronics companies The Influence of DOD Funding of R&D in plan to accelerate their investment in commu- Information Technology nications R&D, including work on integrated power semiconductors and cellular radio.” DOD funding for R&D in information tech- Thus, the industry viewed broadly is com- nology reflects its growing dependence on this mitted to well-funded R&D. technology and its reluctance to be dependent on foreign sources for technology critical to Comparison of R&D Funding by national security. Defense spending for R&D Selected Countries generally has ranged from a high of 90 percent of Federal R&D spending in 1953 to a low of Funding levels for R&D are generally rec- 50 percent during 1976-80, and is expected to ognized as important to the innovation proc- rise to 70 percent in 1985.56 ess, as noted earlier. The United States has fallen behind two of its major competitors, as Table 5 shows the distribution of DOD fund- measured in terms of total outlays for R&D ing for 1983 among basic and applied research, .-— 63~ugjne9s week, R&I) Scoreboard 1982, June 20, 1983, 66% willi~ c. B~Srn~ U.S. Civilian and Defense R&D pp. 122-153. Funding: Some Trends and Comparisons with Selected In- “Science Resource Studies, Highlights, National Science dustrialized Nations, Congressional Research Service, Library Foundation, NSF 83-327, Dec. 15, 1983, p. 2, and NSF 84-329, of Congress, Aug. 26, 1983. Oct. 15. 1984. ‘Ibid.

Table 4.—The Top 15 in R&D Spending

In percent of sales In dollars per employee 1. TeleSciences...... 31 .6°/0 1. Ultimate ...... $37,089 2. Policy Management Systems ...... 26.6 2. Fortune Systems ...... 19,390 3. Fortune Systems ...... 22.3 3. TeleSciences...... 18,797 4. Management Science America ...... 20.8 4. Convergent Technologies ...... 18,721 5. King Radio...... 20.0 5. Activision...... 16,667 6. Dysan ...... 19.4 6. Cray Research...... 16,467 7. Advanced Micro Devices...... 19.4 7. Management Science America ...... 15,563 8. Modular Computer Systems...... 17.6 8. Amdahl ...... 15,413 9. ISC Systems ...... 16.6 9. Digital Switch ...... 15,017 10. Computer Consoles ...... 16.6 10. Policy Management Systems ...... 14,677 11. LTX ...... 16.4 11. Applied Materials ...... 14,545 12. Ramtek...... 15.6 12. Auto-trol Technology ...... 14,413 13. Applied Materials ...... 15.6 13. Computer Consoles ...... 13,816 14. Auto-trol Technology ...... 15.4 14. Network Systems ...... 13,292 15. Kulicke & Soffa Industries ...... 15.3 15. LTX ...... 13,229 SOURCE: Standard & Poor’s Compustat, Inc., as cited in Business Week, “A Deepening Commitment to R& D,” July 9, 1984, p. 64; and “The U.S. Still Leads the World in R&D Spending, ” June 20, 1983, p 122 Ch. 2—The Environment for R&D in Information Technology in the United States ● 45

Figure 4.— Estimated Ratio of Civilian R&D example, in funding for basic research in elec- Expenditures to Gross National Product trical engineering, DOD accounts for some 69 for Selected Countries percent; in computer sciences, 55 percent; and 2.6 ‘ in mathematics 42 percent. In applied re- 2.5 search, the DOD is a major Federal funder for 2.4 West electrical engineering (90 percent), computer sciences (87 percent), and mathematics (29 percent). These, as well as others, are disciplines supported primarily by the DOD R&D budget and that have a central influence on advances in information technology for the Na- tion.67 There have been many commercially appli-

1.7 cable advances in information technology that have their origin in, or had strong early support from, DOD funded R&D. These include very high speed integrated circuits (VHSIC), digital telecommunications, and new high-performance materials. However, there are some major disadvantages for the commercial sec- 1 tor to DOD funded R&D. Among these are: 1.1 France I security classifications which tend to slow ad- 1.0 , vancements in technology; rigid technical .9 -1 specifications for military procurements which have limited utility for commercial applica- 1961 63 65 67 69 71 73 75 77 78 81 83 tions; and the “consumption” of limited, val- Years uable scientific and engineering resources for military purposes, which may inhibit commercial developments. This issue is discussed in more detail in chapter 8. and development. DOD spending for basic research accounts for some 13.6 percent of the U.S. Patent Activity total Federal obligations, while applied re- It is generally accepted that patenting is a search accounts for 33.9 percent, and devel- measure, even if imperfect, of the effectiveness opment accounts for 71.3 percent. of R&D activities. A key observation is that Table 6 shows that the DOD R&D budget patenting in information technology is among dominates some fields of Government R&D spending for basic and applied research. For 571bid.

Table 5.—Federal and Department of Defense Obligations for Basic Research, Applied Research, and Development, 1983 (Estimated) (millions of dollars)

Total R&D Basic research Applied research Development Total Federal Government ...... $42,973.8 100 ”/0 $5,765.2 100.0 ”/0 $7,499.7 100.0% $29,708.9 1OO.OO/o Department of Defense ...... $24,519.6 57.1 % $ 782.1 13.60/o $2,543.9 33.9 ”/0 $21,193.6 71.30/0 SOURCE: National Science Foundation, “Federal Funds for Research and Development Fiscal Years 1981, 1962, and 1983,” vol. XXXI, Detailed Statistical Tables (NSF 82-326) (Washington, DC: U.S. Government Printing Office, 1982), pp. 174, 179, 181, 183, Percentages calculated from data in the table. As cited in Boesman, “U.S. Civilian and Defense R&D Funding; Some Trends and Comparisons With Selected Industrialized Nations,” Congressional Research Service, Library of Congress, Aug. 26, 1983. 46 • Information Technology R&D: Critical Trends and Issues

Table 6.—Federal and Department of Defense Obligations for Basic and Applied Research by Field, 1983 (Estimated) (millions of dollars)

DOD as a percentage of Total Federal total Federal funds DOD funds funds Basic research:

Electrical engineering ...... $103.8 $71.8 69.1 ‘/0 Computer sciences ...... 73.0 40.0 54.8 Mathematics ...... 100.7 42.6 42.3 Applied research: Electrical engineering ...... $520.5 $471.2 90.5 ”/0 Computer sciences ...... 91.5 79.4 86.7 Mathematics ...... 48.5 13.9 28.6 SOURCE: National Science Foundation, “Federal Funds for Research and Development, Fiscal Years 1981, 1982, and 1983,” vol. XXXI Detailed Statistical Tables (NSF 82-326) (Washington, DC: U.S. Government Printing Office, 1982), pp. 73, 75, 79, 82, 98, 101, 104, 109. Percentages calculated from data in the table. As cited in Boesman. the most intensive of all technologies. U.S. pa- Table 7.—Technology Distribution of the tenting of foreign origin68 in all technologies Top 50 U.S. Patent Electrical Categories 1978.80 has doubled in the past two decades to 41 Percent of percent—indicating escalating world competi- Ranked by actual file growth categories tion for U.S. patents in general. A small num- Semiconductors and circuits ...... 480/o ber of foreign multinational corporations have Computers ...... 15 ”/0 a dominant (but perhaps somewhat diminish- Other ...... 37 ”/0 Total number of patents in the ing) role in the proportion of foreign-origin 50 patent categories ...... 8,139 U.S. patents. These “multinationals” empha- SOURCE: Tenth Report, Technology Assessment and Forecast, U.S. Patent and size information technology patents. The over- Trademark Office, U.S. Department of Commerce, November 1981, pp. 16, 24 all picture derived from this review of patent data confirms the finding reported in chapter patents in the electrical category, with 548 3 that foreign competition in information tech- patent documents.59 nology is increasing. Solid-state devices, integrated circuits, and transistor categories together account for 24 U.S. Patent Data of the 50 categories in the total ranked by ac- 60 The top 50 electrical patent categories tual growth from 1978 to 1980. The percent (ranked by actual numeric growth in the num- growth in the number of these patents generally ranges from about 40 percent to 59 per- ber of patents) received 8,139 patents during 61 1978-80 time period (table 7). Within these cent. Solid-state devices account for 8 of the 11 highest growth entries. Lasers, laminag- categories, semiconductors and circuits ac- raphy, and fiber optics are also among the counted for 48 percent and computers 15 percent, respectively. In the computer category, information technology segments included in General Purpose Programmable Digital Com- the high patent growth entries. The two sub- puter Systems was the most active, as in pre- classes of fiber optics inventions show patent vious years, receiving 632 patent documents. Miscellaneous Digital Data Processing Sys- 5Tenth Report, Technology Assessment and Forecast, Pat- ent and Trademark Office, U.S. Department of (knmerce, No tems received the second largest number of vember 1981, pp. 14-18. ‘Actual growth is the numeric increase resulting from additions to the patent copies (including cross-reference copies) to the file in the 3-year period 1978-80, Ibid p. 11. “Percent growth, as used by the U.S. Patent and Trademark Office, is computed by dividing the actual growth for the 3-year 5aThe coUtV origin of a patent is determined by the coun- period exarnined (1978-80) by actual growth for the 8-year period try of residence of the first named inventor. (1975-80), and multiplying by 100. Ibid., p. 11. Ch. 2—The Environment for R&D in Information Technology in the United States ● 47

growth rates of over 70 percent. The listing, cially in countries that represent large in fact, is composed almost exclusively of in- potential markets. As a consequence, U.S. formation technology inventions. Computer patent statistics tend to mirror trends in technology patents showed growth rates of technological activity worldwide. over 70 percent for the 1978-80 period, well 2. Although foreign-origin patenting in all above the average of 46 percent for all tech- technologies averaged only 20 percent of nologies during the same time period.62 the total U.S. patenting for the years 1963-66, the percentage share has contin- There are reasons for caution against gen- ued to increase, reaching 40 percent of the eralizations concerning the use of patent sta- total for the year 1980, and 41 percent for tistics—e.g., variations in the importance and the 1981 to mid-1983 period. the degree of “invention” of different patents; the propensity (or absence of it) of some com- Figure 5 illustrates the long-term decline in panies, and perhaps countries, to patent as op- the number of U.S.-origin information technol- posed to using other alternatives-e. g., trade ogy patents granted in the United States be- secrets, or lead times in the market place; the tween 1968 and 1981, and the relative leader- cost factor as a disincentive to patenting, as ship position of Japan compared to France, well as concern for antitrust allegations based West Germany, and the United Kingdom. It on patent dominance; rapid technological is not clear as to why the total number of U.S. change (making patents of limited value); dif- patents has declined steeply between 1971 and ferences in the scope of patent categories that 1980, but the U.S. Patent and Trademark Of- may give a misleading impression of substan- fice advises that except for Japan, the trend tial amount of patenting activity in a broadly was worldwide during that period. (Note that scoped subcategory or vice versa. in 1979 a shortage of funds at the Patent Of- fice limited the number of U.S. patents Nevertheless, the evidence shown above granted, artificially lowering the total for that clearly seems to support the observation that year). the level of patenting for information technology in the United States is vigorous and may As shown by table 8, the share of U.S.-origin be indicative of extensive R&D in this field. patents decreased from 79 to 58 percent of the total during 1968-81, while the share of Japa- Foreign-Origin U.S. Patents nese-origin information technology patents granted in the United States increased from In 1973 the U.S. Patent and Trademark Of- 3 to 19 percent. fice (USPTO), began detailed documentation of foreign-origin patent activity through its The two “top 50” electrical category lists Office of Technology Assessment and Fore- noted earlier reveal a significant proportion of cast (OTAF). One of OTAF’s reports,63 which foreign-origin U.S. patents in the high patent- is cited extensively in this section, provides growth categories. Fifty-four of the entries in useful background and many important find- both lists show the percentage of foreign ori- ings on foreign patenting in the United States. gin to exceed the average of 38.5 percent for Among the findings are: all technologies for 1978-80. This is not surprising, since the high-growth patent subclasses 1. Because patents obtained in the United are pursued by companies in all of the devel- States convey no protection in other coun- oped countries. tries and vice versa, inventors tend to patent in more than one country, and espe- Table 9 shows the percentage of foreign-origin U.S. patenting in patent category group- 621 bid., p. 22-26. ings dealing with some components of infor- etIbid., SW for ex~ple, Section I, Part IV— “Most Foreign mation technology. Three of the five category Active Patent Technologies, ” pp. 27-32, and Section II, Pat- ent Trends: Foreign Multinational Corporations Patenting groupings examined (two in fiber optics and Trends in the United States, pp. 33-46. one in television) show foreign-origin patenting

48 . /nformation Technology R&D: Critica/ Trends and Issues

Figure 5.–U.S. Patents in Information Technology, SIC Codes 357, 365, 366, 367

14,000

12,000

11,000

10,000

9,000

8,000

7,000

6,000

5,000

4,000

3,000

2,000

1,000

0 1968 69 70 71 72 73 74 75 76 77 78 79 80 81 Years

SOURCE: U.S. Patent and Trademark Office, Office of Technology Assessment and Forecast. Ch. 2—The Environment for R&D in Information Technology in the United States ● 49

Table 8.—Percentages of U.S. Patents Granted in Information Technology (IT) and in All Technologies (ALL), 1981

United States Japan West Germany United Kingdom France Year IT ALL IT ALL IT ALL IT ALL IT ALL 1968 ...... 79 77 3 2 5 6 4 4 3 2 1981 ...... 58 60 19 13 8 10 4 4 4 3 1982 ...... NA 59 NA 14 NA 9 NA 4 NA 3 NA—Not available alnformation Technology (IT) here in~l”de~ Slc Codes 357, Office Computing and Accounting Machines; and 365-367, communication Equipment and Electronic Components. SOURCES The Office of Technology Assessment and Forecast, US Patent and Trademark Office, All Technologies Report, 1963-June 1983, and Indicators of the Patent Output of US. Industries (1963-81). IT numbers were calculated from data developed with assistance from the National Science Foundation, Science Indicators Unit.

Table 9.— Foreign-Origin U.S. Patents in cent of the foreign origin U.S. patents from Some Components of Information Technology 1980-83.64 Figure 6 depicts the shares of U.S. communications equipment and Percent patents for foreign origin electronic components among Japan, West Title 1/81-6/83 Germany, and the United Kingdom. Light transmitting fiber, waveguide, or rod . . . 48.2°/0 Laser light sources and detectors ...... 50.6°/0 These findings are consistent with comments Color and pseudo color television ...... 52.8°/0 from OTA workshop participants concerning Active solid-state devices, e.g., transistors, solid-state diodes ...... 40.80/o the growing intensity of foreign competition General-purpose programmable digital in R&D. The statistics no doubt understate computer systems and miscellaneous the level of foreign ownership of U.S. patents, digital data processing systems ...... 34.9°/0 since they do not take into account patents of NOTE The percent foreign origin IS determined by dlwding the total number of U S patents granted between January 1981 and June 1983 to foreign. U.S. origin that are controlled by foreign in- resident Inventors by the total patents granted in the same time period, and multlply!ng by 100 terests, e.g., patents issued to U.S. residents SOURCE Off Ice of Technology Assessment and Forecast, Patent and Trademark or companies that are foreign-owned or for- Off Ice, U S Department of Commerce eign-controlled, or the inclination of foreign multinational corporations to patent in other countries (see section below on Foreign Multi- national Companies). Even understated, how- to be significantly higher than the average of ever, the intensity of foreign influence over 41 percent. U.S.-patented technology” is clearly signifi- Table 10 shows that in selected telecom- cant. By way of providing perspective, it munications categories, the Japanese share of foreign-owned U.S. patents ranges from 38 to ‘Patent Profiles: Tekcomnnmications, Patent and Trademark Office, Office of Technology Assessment and Forecast, U.S. De- 56 percent. For all categories of telecommu- partment of Commerce, August 1984, p. 15. nications, Japanese residents received 45 per- b51bid., p. 4.

Table 10.—Foreign-Owned U.S. Patents in Selected Telecommunications Classes, 1982

Percent foreign Percent Japanese Percent Japanese of total of total of foreign Telephony...... 43% 17 ”/0 390/0 Light wave communication ...... 53 ”/0 20 ”/0 380/o Analog carrier wave communications ...... 43 ”/0 240/o 560/o Digital and pulse communications ...... 40 ”/0 160/0 38% Telephony–Class 17911 R.l AA: IAT.1 FS, 1 H-1 MF, 1 MN.l SS, 1SW.106, 108 R-I9O Light wave communications–372/43-59 & 75, 357/17 & 19, 455/600-619, 370/1.4, 350/96.1-96 34 Analog carrier wave communlcat!ons—455/ l.355 Dlgltal and pulse communication (excludes I!ght wave, Includes error detection and A/D & D/A conversion) —375/all subclasses; 371/1-6 & 30-71, 178/all subclasses, 340/347, AD347, AD.347 SY, 332/9 R-15, 329/104.109 SOURCE Reports prepared by Off Ice of Technology Assessment and Forecast for publication In late 1964

50 ● Information Technology R&D: Critical Trends and Issues

Figure 6.–Share of Foreign Patenting in the United Foreign Multinational Corporations States for the Three Most Active Countries in Selected Product Fields: 1981 Another important OTAF finding concerns Communication equipment the role of foreign multinational corporations and electronic components (FMNCs) in patenting. Taking into account the relative annual sales of the 10 major FMNCs and their ranking among the “Fortune 500” companies, OTAF has found a strong correlation between ranking by patents and sales level.67 A comparison of the 10 FMNC's patenting with total U.S. patenting for 1969-80 is shown in figure 7. In addition to noting that the FMNC's ownership of U.S. patents has recently (1980) leveled off to about 5.5 percent, the OTAF study observes that: ● The 10 FMNCs own or control, on the av- SOURCE: National Science Board, National Science Foundation, Science indicators—19S2, 1983. erage, 4.7 percent of all U.S. patents granted each year. should be noted that some other countries ● The extent of the 10 FMNC's ownership have an even higher percentage of foreign- of U.S. patents doubled from 1969 to origin patents, e.g., Canada, 93.4 percent; the 1976–although the rate of increase had United Kingdom, 84.2 percent; France, 67.6 diminished to near zero by 1980. percent. Japan has 16.6 Percent.66 Patent Statistics, Op.

Figure Patent Activity of 10 Foreign Multinational Corporations, 1969-80

7.0% 4,000

3,500 6.00/0

0 3,000 5.0%

2,500 4.0%

2,000 3.0%

1969 70 71 72 73 74 75 76 77 78 79 80

Years SOURCE: Tenth Report, Technology Assessment and Forecast, Patent and Trademark Office, U.S. Department of Commerce, November p. Ch. 2—The Environment for R&D in Information Technology in the United States ● 51

● More recently, (1979-80) the percentage an indication of the concentration of foreign- of U.S. patents granted to the 10 FMNCs owned U.S. patents by a few multinational has begun to decline in proportion to total firms. foreign origin U.S. patents suggesting a These statistics further confirm the testi- diminished role in ownership of U.S. pat- mony of OTA workshop participants concer- ents for these particular FMNCs. ning growing foreign competition in informa- The trend, while changing, shows that about tion technology R&D, and the observation one in every eight U.S. patents of foreign origin that other countries have developed national is owned or controlled by only 10 FMNCs68— policies and programs that target information —.—. —— technology. 6aIbid. ((_J’I’AF Study), P. 40.

A Synthesis: The Changing U.S. R&D Environment

Some measures, such as investment in R&D Universities are encouraging new institu- and growth in profits, indicate that R&D in tional arrangements with industry, and the im- U.S. information technology is vigorous. Other portance of their role in the R&D process (par- measures, such as competition in advanced- ticularly in performing basic research) maybe technology products and foreign ownership of growing. There are widespread problems re- U.S. patents, indicate a less robust situation. lating to both the quantity and quality of uni- Thus, although information technology re- versity equipment and facilities for conduct- search and development is making marked ad- ing information technology R&D, although vances, it is—at the same time-undergoing these conditions may be improving. Some pressure from foreign competition. In response State Governments have become active in to the pressure, the participants in the R&D helping their universities to improve research process are initiating a variety of changes. capabilities and in encouraging university- These changes are discussed in later chapters industry pairings (see ch. 6). of this report. Finally, the Federal Government has adopted Industry continues to invest heavily in in- policies intended to encourage private sector formation technology R&D–an indication of investment in R&D and to facilitate the trans- its belief in R&D’s importance to competitive fer of technology from Government to indus- ness. The increasing costs of R&D are mak- try. The Federal Government’s (especially ing new institutional arrangements such as DOD’s) expenditures for information technol- joint research ventures and closer ties with ogy R&D are growing rapidly and continue to universities more attractive. However, indus- have a strong influence on the direction of try experts recognize that although R&D is technological development in some informa- necessary to competitiveness it is not suffi- tion technology areas. cient to ensure it; other components of the innovation process are also important to maintaining competitiveness in international trade. 52 ● information Technology R&D: Critical Trends and Issues

Observations

Some useful observations can be made based ing trade balances and job losses—associated on the changes taking place in the U.S. R&D with falling by the wayside in the competitive environment. First, the growth in foreign race. As a consequence, they have established competition-whatever its effect on U.S. jobs national policies and programs to enhance and trade balances for the long term—is stim- their domestic industrial position. ulating R&D investment, as shown in chapter Third as foreign competition has inexorably 9. There has probably not been a time since strengthened in the post-World War II era, the the turn of the century when new products and product improvements have been marketed in broad margin of error that the United States such rapid succession as has been the case once enjoyed has essentially vanished. with information technology, nor a peacetime The long-term effects of several factors—na- era when R&D had such a central role in the tional industrial policies, nontariff trade bar- affairs of nations. riers (e.g., prohibiting the import of certain products or services, incentives for industrial Second, the U.S. information technology in- innovation, interest rates, export controls— dustry is facing a “new world” of foreign com- will determine the winners and losers, as na- petition. The intensive level of targeted and tions maneuver to remain competitive. The well-funded foreign competition is not likely to decline in the foreseeable future. Each of United States will need to find ways to monitor its position relative to international com- our major competitors’ governments believe in the central importance of information tech- petitors and to refine its policies as needed to keep in step with the changing global R&D en- nology as an essential ingredient for achiev- vironment. ing economic, social, or national security goals, as well as the penalties-in terms of worsen- Chapter 3 Selected Case Studies in Information Technology Research and Development Contents

Page Table No. Page Introduction ...... 55 12. Milestones in the History of Computer - Architecture ...... 57 Case Study 1: Advanced Computer 13. 1982 Federal Spending for Computer Architecture ...... 55 Architecture R&D...... 58 Findings ...... 55 14. Federal Open Access Supercomputer Changing Computer Architecture ...... 56 Facilities ...... 58 Computer Architecture R&D ...... 57 15. Summary occurrent and Planned Critical Areas of Research ...... 63 Government Open Access Advanced Manpower ...... 65 Computer Architecture Software International Efforts...... 65 Development Facilities...... 59 Case Study 2: Fiberoptic Communications 67 16. NSF Plans for Computer Research .,.. 60 Findings ...... 67 17. Summary of New Commercial Advantages ...... 67 Supercomputer Systems Under Commercialization Trends ...... 69 Development ...... 61 United States R&D...... 71 18. 1983 DARPA, DOE, and NSF University Directions of U.S. Research ...... 73 Funding of Advanced Computer Research in Japan ...... 74 Architecture ...... 62 19. Major Government Sources of Artificial Case Study 3: Software Engineering . . . . . 75 Intelligence R&D Funding...... 94 Findings ...... 75 20. NSF Grant Awards in Artificial Introduction ...... 75 Intelligence ...... 94 Software R&D Environments ...... 76 21. Strategic Computing Cost Summary . . 94 Content and Conductor Software 22. Representative Expert Systems ...... 104 Engineering R&D...... 79 International Efforts in Software Engineering R&D...... 85 Case Study 4: Artificial Intelligence ...... 87 Findings ...... 87 Figures Introduction ...... 87 Figure No. Page Artificial Intelligence R&D 8. Computer Architecture Functional Environments ...... 89 Elements ...... 56 Content and Conduct of Artificial 9. Optical Fibers Are Small, Lightweight Intelligence R&D ...... 96 and Versatile ...... 68 Foreign National Efforts in Artificial 10. The Software Life Cycle...... 80

Intelligence R&D ...... 105 110 Artificial Intelligence ...... 90 12. Artificial Intelligence Science and Engineering ...... 92 Tables 13. Strategic Computing Program Structure Table No. Page and Goals ...... 95 11. Major Federal Advanced Computer 14. A Semantic Network...... 98 Architecture R&D Projects ...... 57 15. Markets for Artificial Intelligence . . . . 101 Chapter 3 Selected Case Studies in Information Technology Research and Development

Introduction

These case studies present a microcosm of formation technology R&D. These fields were the R&D process in information technology. chosen to include both hardware and software The diversity of research and development ef- and both computer and communications tech- forts in information technology make it vir- nologies. They also illustrate the mix of long- tually impossible to examine in detail all of the term goals and near-term capabilities, thus re- many fields and disciplines. Therefore, four flecting the importance of these different fields have been selected for detailed analysis. perspectives in the R&D process. These four They are: Advanced Computer Architecture areas, moreover, are among those considered (ACA), Fiber Optics (FO), Software Engineer- to be critical in determining the direction and ing (SE), and Artificial Intelligence (AI). pace of advance of information technology as a whole. The importance of advances in the These fields were selected for several rea- four fields is exemplified by the development sons. They depict the wide range, diversity of government funded national R&D pro- and inter-relatedness of the fields and applica- grams in Japan, Britain and the European tions of information technology. An analysis Economic Community. of them provides a broad overview of the scientific, technical and institutional issues in in-

Case Study 1: Advanced Computer Architecture

Findings processing architectures, which are inherently more complex to design and to use. ● The technology of advanced computer de- VLSI (Very-Large-Scale Integrated Circuit) sign is critically important for the expan- design facilities, based on powerful com- sion of information technology in many puters, are now being used to develop and fields. There is extensive R&D activity test computer architectural designs, includ- underway in universities, in industry, and ing parallel processors and special designs in the National Laboratories aimed at pro- for certain dedicated operations, such as ducing and exploiting new computer de- communications signal processing, image signs; but there is considerable uncertainty processing, and graphics. over which new designs will be viable. Software has been difficult to produce for ● Since their invention, electronic computers computers of advanced, high-performance have been based on one architectural model, design. As the variety and complexity of ar- the von Neumann sequential processing architectural types increases, the difficulty of chitecture. The limits of computational developing and integrating software will in- speed achievable with this design are being crease. Therefore, research in software de- reached; significant further increases in velopment for novel computer designs will computer performance will require parallel be critical.

38-802 0 - 85 - 5 56 . Information Technology R&D: Critical Trends and Issues

● The Federal Government has been a major Figure 8.—Computer Architecture Functional driver of advanced computer architecture Elements because of its scientific and national security applications. The Government has influenced the evolution of computer design through the funding of R&D and the procurement of state-of-the-art systems. This leverage, though still important, is diminishing as commercial applications for advanced computers grow and as Federal requirements become a smaller fraction of sales. ● National programs in Japan, Great Britain, France, Germany, and the European Com- munity have been established to pursue ad- SOURCE: Office of Technology Assessment. vanced computer R&D. The Japanese have recently demonstrated an ability to produce advanced architecture computers of com- First, the limits imposed by physical laws petitive performance to American products. on the computational speed attainable with ● American companies and universities pur- traditional computer design are being ap- suing R&D in- computer design face-dif- proached. Science and engineering demand ficulties: continual advances in computational speed to increase the precision of calculations and to Universities cannot afford design and 1 testing facilities for developing an architec- improve accuracy in models and simulations. tural idea to the point where its performance Sequential processing constitutes a severe re- can be assessed. striction on the precision and completeness Companies face large, risky investments that these calculations and models can achieve in the design of new high-performance com- in a reasonable amount of time. Therefore, puter systems. Markets for novel machines computer designers are studying architectures are initially small and expand only slowly that can make possible decomposition of large as new applications are exploited and soft- calculations into pieces for simultaneous proc- ware becomes available. essing by a number of computational units in parallel. Changing Computer Architecture Second, as information technology is applied in more and more areas, special problems are Computer architecture is the internal struc- encountered that impose unique demands on ture of a computer, the arrangement of the computer capabilities. Until recently, system functional elements that carry out calculations designers have relied on software to apply the and information manipulations. (see fig. 8). capability of von Neumann processors to prob- Since the early 1950s, all electronic comput- lems. Now, it is possible to create special in- ers (with a few exceptions) have been designed tegrated circuits to address specific problems. around one basic architectural model, the von Neumann machine, invented by mathemati- IThese are the applications normally associated with so-called “supercomputers.” Major current applications are in, for ex- cian John von Neumann. In this architecture, ample, weather modeling and the simulation of nuclear weap- instructions and data are stored in memory, ons explosions. The reader is referred, for a discussion of the fetched one by one in sequential fashion, and applications of and policy issues surrounding supercomputers, to Supercomputers: Foreign Competition and FederaJ Fund”ng acted on by the processor. Computer design by Nancy Miller, Congressional Research Service, Issue Brief is now changing, encouraged by two factors. 83102, latest update July 12, 1984. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 57

Computer architecture R&D is making possi- former of work. Major elements of the soft- ble the economical design of custom computer ware development work for each generation of architectures for specialized applications in- these systems have been done by the National cluding telecommunications and data acqui- Laboratories, especially Los Alamos (LANL) sition signal processors, image and graphics and Lawrence Livermore (LLNL).2 Moreover, processors, and symbol processors for the the impetus for the development of each suc- manipulation of nonnumerical information. cessive generation of advanced architecture The impending changes in computer archi- scientific computers has come predominantly tecture promise cost-effective solutions to from government demand for faster, higher ca- many problems, but they also challenge the pacity, and more sophisticated systems for designers, suppliers and buyers of computer weapons, intelligence, energy, and aerospace systems. Designers will need to have more applications (see table 12). The National Labs still constitute the greatest concentration of detailed appreciation of applications; computer vendors will be faced with more com- users of supercomputers (see table 15). plexly segmented markets; and buyers will The Federal Government has provided be- need to be more sophisticated in defining their tween $15 and $20 million in annual funding needs and in choosing among a wider offering for advanced computer architecture R&D in of products. recent years (see table 13). In addition to sponsoring research in universities and in indus- Computer Architecture R&D try, the Government has performed computer research at the National Labs. Federal Government Involvement

The Government has had considerable in- ———— volvement in advanced computer architecture The first Cray-1 computer was placed in Los Alarnos National R&D (see table 11), both as a funder and a per- Lab without any software.

Table 11 .—Major Federal Advanced Computer Architecture R&D Projects

Machine Year delivered Agency Contractor Major use ENIAC ...... 1945 Army University of Pennsylvania Ballistics calculations NORC ...... 1950 Navy IBM Ordinance research — 1950 NSA Sperry Classified CDC 1604 ...... 1959 NSA Control Data Classified LARC ...... 1961 AEC Sperry Nuclear weapons design STRETCH ...... 1961 AEC IBM Nuclear weapons design CDC 6600 ...... 1964 LLNL Control Data Nuclear weapons ILLIAC IV ...... 1972 DARPA/NASA Burroughs/University of Illinois Aerodynamics MPP ...... 1983 NASA Goodyear Aerospace Image processing s-1 ...... – LLN L/Navy — Signal processing SOURCE Office of Technology Assessment

Table 12.—Milestones in the History of Computer Architecture

Class Date Typical machines Major innovation I ...... 1953 IBM 701 Vaccuum tubes II ...... 1960 CDC 1604 Transistors Ill ...... 1964 IBM 360 1/0 processing CDC 6600 Freon cooling Iv , , ... , ., . . . 1970 IBM 370 Integrated circuits CDC 7600 v ...... 1972 Illiac IV Parallel processing TI ASC Pipeline architecture VI ...... 1976 Cray-1 Vector processing SOURCE Office of Technology Assessment 58 • Information Technology R&D: Critical Trends and Issues

Table 13.—1982 Federal Spending for Computer searchers in science and engineering fields to Architecture R&D (millions of dollars) support the missions of several Government Department of Defense ...... $12.0a departments (DOD, DOE, NASA), and to fur- Department of Energy , ...... 2.9a ther basic research (NSF). Seven Federal fa- National Aeronautics and Space Administration . . . 1 .5a cilities provide limited open access to certain National Science Foundation ...... 1.2 Total ...... $17.6 groups of researchers (see table 14). aEstimates. The National Laboratories plan to add more SOURCE: Department of Defense and Department of Energy numbers from the Office of Technology Assessment Workshop on Advanced Computer supercomputers over the next few years, so it Architecture; NASA number from personal communication with Paul Schneck; NSF number from Summary of A wards, FY 1982, NSF Direc- can be expected that Government scientists torate for Mathematical and Physical Sciences, Computer Sciences and engineers and contractors on mission Section, agency work will have access to state-of-the- art large-scale computing facilities (see table DOE spent $645,000 at Los Alamos and 15). Academic researchers will have limited ac- $500,000 at Lawrence Livermore National cess to these facilities for work in fields related Labs in fiscal year 1983 on two experimental to agency missions (e.g., fusion energy, atmos- research projects on advanced computer archi- pheric and ocean sciences, and aerodynamics). tecture hardware.3 In addition, Los Alamos is These facilities also provide support for soft- leasing a Denelcor HEP-1 to experiment with ware development. parallel processing software concepts. The Navy, in conjunction with Lawrence Liver- The Department of Defense, through the more, has been involved with the design and Defense Advanced Research Projects Agency construction of an advanced architecture com- (DARPA), has formulated ambitious plans for puter, termed the S-1 Project. The design is research and development in advanced com- intended to handle signal processing tasks for puter-based systems. Included will be efforts Navy missions. Approximately $20 million to develop high speed signal processor archi- has been spent over the last 4 years on the S- tectures and to integrate numeric and symbol- 1 Project.4 ic processing in advanced computer architectures for use in intelligent weapons systems.5 Facilities and help in advanced computer applications development are provided to re- In April 1983, the National Science Foun- dation organized a working group to study

9Edward Oliver at the OTA workshop on Advanced Comput- ———— er Architecture, July 14, 1983. ‘This program, called “Strategic Computing, ” is covered in ‘Personal communication from Sidney Fembach, Consultant, some detail in the Artificial Intelligence Case Study later in Control Data Corp. this chapter.

Table 14.—Federal Open Access Supercomputer Facilities

Facility Major system Research users Charges NCAR ...... 2 CRAY 1-As Atmospheric and ocean sciences No charge to NSF users. $2,200 per prime CPU hour for others MFECC ...... 2 CRAY 1 Magnetic fusion energy No charge 1 CDC 7600 community LANL ...... Open 1 CRAY Government agencies, labs, and $636 per prime CPU hour 3 CDC 7600s nonprofit institutions 1 CYBER 825 NASA-Ames ...... CRAY 1-S Computational fluid and No charges to NASA grantees. CDC 7600 aerodynamics $2,000 per CPU hour for CRAY NASA-Goddard ...... CYBER 205 NASA-funded and NASA-project No charge to NASA grantees. $1,000 related per CPU for others NASA-Langley...... CYBER 203 NASA and NASA-funded $1,300 per CPU hour scientists NASA-Lewis ...... CRAY 1-S Principally aerodynamics related No charge for NASA supported SOURCE: A Nafioml cornwtiw ~rrtirortmerrt for Academic l?esearcl? National Science Foundation, October 1983

Ch. 3—Selected Case Studies in Information Technology Research and Development ● 59

Photo credit: U.S. Department of Energy, Los Alamos National Laboratory

View of a part of the main computing facility at Los Alamos National Laboratory. CRAY 1 in foreground

Table 15.—Summary of Current and Planned Government Open Access Advanced Computer Architecture Software Development Facilities

Number Planned additions Agency Current systems of users fiscal years 1984-88 National Science Foundation. ...2 Class VI 850 1 Class Vll Department of Energy ...... 3 Class VI 6,400a 1 Class Vl, 5 Class Vll National Aeronautics and Space Administration ...... 3 Class Vl, 1 special 3,000 1 Vl, 1 Vll, 1 special Totals ...... 8 Class Vl, 1 special 10,250 2 Vl, 7 Vll, 1 special performing classified work. SOURCE: A Computing Environment for Academic Research, National Science Foundation, 60 ● Information Technology R&D: Critical Trends and Issues what was recognized as a critical scientific im- ment of science and technology. Accordingly, perative. The report of that group stated:6 they approved a budget of $40 million in fiscal year 1985 for NSF’s Advanced Scientific Com- Computing facilities have a decisive effect puting initiative, thus doubling the adminis- on the kind of research which is done by aca- 7 demic scientists and engineers. During the tration’s request for this program. 1950s and 1960s the Government encouraged the growth of computing in research, re- Industry’s Role search methods were transformed in disci- Three U.S. companies are developing next pline after discipline, and the United States generation (Class VII) supercomputer sys- enjoyed a large, ever-widening lead in quantitative research and modeling complex phe- tems. In addition, two Japanese companies nomena. In the 1970s Government support (Fujitsu and Hitachi) have introduced new sys- slackened and academic computing facilities tems and a third (Nippon Electric-NEC) is de- no longer kept pace with advancing technol- veloping a new supercomputer, planned for de- ogy . . . Science has passed a watershed in livery in 1985 (see table 17). using computers for research. Computers are Cray will introduce the Cray-2 in 1985. This no longer just tools for measurement and 8 analysis but have become the means for mak- will be a four processor vector machine. The ing new discoveries . . . Academic research in Cray-3 is scheduled for introduction in 1986. computer architecture, computational math- It will be an 8 to 16 processor vector machine ematics, algorithms, and software for paral- with Galium Arsenide (GaAs) (see ch. 9) cir- lel computers should be encouraged to in- cuitry. Cray sees integrated circuit technology crease computing capability. as critical. The Japanese are the major sup- In response to this imperative, the NSF pliers of state-of-the-art fast bi-polar memory chips, and one-half of the integrated circuits working group recommended an expansion in 9 spending for academic research in advanced in current Cray machines are Japanese made. computers, and improved access to computing Control Data (CDC) spun off development facilities including 10 new supercomputer fa- work for its next generation advanced archi- cilities and special networks to make these sys- tecture machine to a new company, ETA Sys- tems widely available (see table 16). tems, which CDC capitalizes with $40 million The House Committee on Science and Tech- for 40 percent ownership. This approach is be- nology considered R&D in advanced coming taken by CDC because small groups with puters and access to powerful computer sys- dedication, entrepreneurial spirit, and a personal stake in the success of the project are tems by scientists and engineers in many 10 fields to be crucial elements in the advance- considered important. ETA Systems will spend $4 million to $6 million the first year 6 Natiomd Computing Environment for Acadenu”c Comput- A on direct R&D costs. Plans are for the first ing, prepared under the direction of Marcel Bardon by the Work- ing Group on Computers for Research, Kent K. Curtis, Chair- demonstration machines to be available late man, July 1983, pp. 1-2.

Table 16.—NSF Plans for Computer Research 7Authorizing Appropriations to the National Science Foun- (million of dollars) dation, House Committee on Science and Technology, Report 98-642, Mar. 30, 1984, pp. 8-9. Fiscal years 8Vector computers have specialized architectures that achieve 1984 1985 1986 high speed calculation of mathematical formulas by treating entire arrays of data (vectors) as processable by single instruc- Local facilities ...... $45.5 $ 90.9 $106.7 tions, saving time on certain calculations that can be arranged Supercomputers ...... 14.4 70.0 110.0 as a series of vectors. Networks ...... 2.1 7.3 11.5 ‘L. T. Davis, “Advanced Computer Projects, ” presentation Advanced computer systems and at the Frontiers of Supercomputing Conference, Los Alamos, computational mathematics. . . 8.0 20.0 33.0 Aug. 15, 1983. Total ...... $69.9 $188.1 $261.2 ‘OW. Norris, “A Conducive Environment for Supercomput- SOURCE: A National Computing Environment for Academic Research, National ers, ” banquet address at the Frontiers of Supercomputing Con- Science Foundation. ference, Los Alamos, Aug. 18, 1983. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 61

Table 17.—Summary of New Commercial Supercomputer Systems Under Development

Maximum speed Companv. Model (M FLOPS’) Available Cray Research ...... Cray-2 1,000 mid 1985 Cray-3 NA 1985-86 ETA Systems ...... GF-10 10,000 1986-87b GF-30 30,000 NA Denelcor ...... HEP-2 4,000 1985-86 NEC ...... SX-2 1,300 March 1985 NA—Not announced aM FLOpS (million floatlng ~olnt ~Peration~ per second) a measure of computer performance on high preci SiOn Calculations bL M Thorndyke at the ~ro”tier~ of s~pe~~o~p”t~~g conference, LOS Alamos, August 1983. SOURCE The IEEE Committee orI Super Scierrt/f/c Computers. in 1986, and volume production of two machines processing, data-flow or multiprocessor ar- per month is planned for 1987. The design will chitectures. 13 employ two to eight vector processors with the Industry representatives characterize the maximum eight processor version to sell in the 11 advanced computer architecture business as range of $20 million. risky. The market is small: approximately 100 Denelcor, a former maker of analog com- Class VI supercomputers have been installed puters, developed a parallel processing com- worldwide as compared to tens of thousands puter design (HEP-Heterogeneous Element of less powerful computers. Development Processor). Since 1982, this design has been costs are high: design tools include other ad- available for sale or lease. Considered to be an vanced architecture machines for hardware experimental machine by users at facilities simulation and software development. such as Los Alamos, this design is a step toward a new generation of computer archi- The Role of Universities tectures. Work is currently underway on the Although as many as 50 U.S. universities HEP-2, which should be competitive with Cray are involved in advance computer architecture and CDC machines if component and software (ACA) research,14 significant funding levels are problems can be overcome. Moreover, the via- available in only a few major schools. (See bility of Denelcor efforts will require higher table 18.) sales than have so far occurred with the HE P-1 .12 University research in advanced computer In addition, other U.S. firms including com- architecture is characterized by a series of puter companies (Digital Equipment, Hewlett- stages of elaboration of a concept including: Packard, Honeywell, IBM, NCR and Sperry), 1) theoretical paper and pencil work; 2) simu- telecommunications companies (Harris), semi- lation of ideas on existing computer systems; conductor companies (Advanced Micro De- 3) “breadboard” wiring of designs with off-the vices, Intel, Monolithic Memories, Mostek, shelf components; and 4) full-scale engineer- Motorola and National Semiconductor), elec- ing and construction of prototype machines tronics companies (Allied, Eaton, General in which state-of-the-art components, software Electric, RCA and Westinghouse) and aero- and peripheral devices can be integrated to space companies (Martin-Marietta) are in- test the design on full-scale problems. OTA volved to some extent in research on parallel found that few if any projects currently under- —.— “L. M. Thorndyke, “The Cyber 2XX Design Process, ” presentation at the Frontiers of Supercomputing Conference, Los “’’Next-Generation Computing: Research in the United Alamos, Aug. 15, 1983. States,” IEEE Spectrum, November 1983, pp. 62-63. “B. Smith, “Latency and HEP, ” presentation at the Fron- “The OTA workshop on Advanced Computer Architecture tiers of Supercomputing Conference, Los Alamos, Aug. 15, concluded that every major Computer Science and Electrical 1983. Engineering department has some interest. 62 • Information Technology R&D: Critical Trends and Issues

Table 18.—1983 DARPA, DOE, and NSF University Funding of Advanced Computer Architecture

Institution DARPA DOE NSF a California Institute of Technology ...... 1 ,000 550 Massachusetts Institute of Technology...... 250 102 2,000a University of California, Berkeley ...... 2,000a Stanford University ...... 2,000 New York University ...... 600 150 University of Illinois ...... 155 200 University of Texas ...... 107 120 University of Wisconsin ...... 95 Duke University ...... 197 Other ...... 570 a Totals...... 7,000 1,757 1,339 a Estimate SOURCE DARPA numbers from a personal communication with Duane Adams; DOE numbers from Edward Oliver at the OTA Workshop on Advanced Computer Architecture; NSF numbers from Summary of Awards: Fisca/ Year 1983, National Science Foundation, Division of Computer Research.

way in universities have funding to carry a de- pose and custom application architectures are sign concept through to the final, systems implementable in VLSI, opening opportunities engineering stage. Several projects will pro- for the testing and evaluation of many more duce prototypes, but the elaboration of an idea computer architecture ideas. However, the ini- into a system with software and supporting tial investment required for VLSI design and peripherals to demonstrate the performance fabrication equipment is very costly and will and utility of the concept on real problems re- probably remain so. It is unlikely that most quires funding on the order of twice what the universities will be able to afford this equipment. currently best funded projects receive. Other expenses associated with ACA re- The major distinction between efforts pur- search include computer-based simulation fa- sued in industry and in universities on ACA, cilities. There may be a need for current gen- aside from the commercial and product devel- eration supercomputers at universities to opment orientation of industry work, is that facilitate and test the design of new architec- more radical and advanced designs are being tures. Bell Laboratories currently devotes pursued in universities, whereas evolutionary most of its Cray computer to VLSI circuit designs are sought by industry. This is a re- design. sult of the stake that industry has in the ex- Supercomputers are also used by Cray and isting base of software and users and the need Control Data for software development, so for upward compatibility of systems. Univer- that software is available when a new hard- sity researchers have a greater ability to pur- ware design is completed. Universities would sue revolutionary designs that could require benefit from access to these software devel- completely new programming approaches and opment tools, giving researchers the chance techniques. to test ideas experimentally. But here again, the costs associated with the procurement and Facilities Requirements operation of these design and computing re- The increasing availability and capability of sources are beyond the means of university VLSI circuitry and computer-aided design project budgets. Some universities are form- tools are expected to have significant impact ing consortia to spread the cost of microelec- on computer design.15 Prototype production tronics design and fabrication facilities across time and cost will decrease. Both general pur- several institutions (see ch. 6). Shared super- ——- -- - — ‘sS. Trimberger, ‘‘Reaching for the Million-Transistor Chip, computer facilities are a key element to NSF IEEE Spectrum, November 1983, p. 100. plans for Advanced Scientific Computing. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 63

Annual operating costs for government There are currently more than 50 concepts supercomputer facilities average more than for parallel processing architectures under con- $10 million.” Only three U.S. universities cur- sideration in academic and industrial institu- rently operate such facilities and these are uti- tions. However, there are no standard metrics lized at less than 50 percent of capacity. The for comparing the performance of different ar- reason for this low usage is the high cost of chitectural designs or the software to be used computer time on these systems, ranging from on parallel machines. Nor is it likely that any $2,000 to $3,000 per hour. University ACA re- one metric could fully measure differences in search is therefore usually done on minicom- performance, since different applications place puters which lack the capability of generating different demands on systems. The develop- sophisticated, high-resolution graphics of im- ment of such metrics and the establishment portance to the design of integrated circuits. of suitable test facilities for implementing standard design evaluations are critical issues Critical Areas of Research in advanced computer architecture. The Fed- eral Government may have a role in this area Currently, the goal of most advanced com- by setting voluntary standards for computer puter architecture research and development performance measurement, and by providing is parallel computation. facilities for testing. The two major U.S. industrial developers of Thus far little attention has been devoted supercomputers, ETA Systems and Cray Re- to the problems of symbolic, as opposed to search, and the Japanese manufacturers, numeric processing architectures. In the past, Hitachi and Fujitsu, are pursuing parallel ar- the von Neumann architecture has been used chitectures in a conservative incremental fash- for both kinds of computations; the focus of ion, contemplating the production of machines advanced computer architecture R&D has with up to 16 parallel processors by the late been on computers for “number crunching” 1980s. Vector architecture will remain the applications, or high precision calculation, dominant method for achieving fast numeri- simulation and modeling for science and engi- cal processing in these systems. neering. The increasing importance of artifi- Universities, by contrast, are pursuing a cial intelligence is encouraging the design of number of methods of achieving “massively special machine architectures, both to ease the parallel” computation with upwards of 1,000 programming of artificial intelligence applica- processors working in concert. One of the basic tions, and to speed the processing of symbolic problems of computing in parallel is the re- computations. Several companies are now pro- quirement for communication and coordina- ducing machines for artificial intelligence, and tion among the individual processing elements the list is expected to grow. ’7 when they are working on pieces of a single Three other areas of technology are critical problem. Often the results of one process are to the development of advanced computer ar- required for another process to go forward. chitecture systems: integrated circuits, circuit Several architectural solutions to these dif- packaging, and algorithm and software design. ficulties are under study, and extensive evaluation of different approaches must be done Integrated Circuits (IC) before their viability in real-world problems can be assessed. Detailed simulation of con- An order of magnitude (1 OX) increase in com- cepts and testing of prototypes is required, puter speed is expected from improvements and present university funding is inadequate in IC materials and manufacturing techniques. to support such work. Silicon will remain the dominant IC substrate material through 1990 because the technology

‘6A National Computing Environment for Academ”c Re- search, op. cit., p. 22. “See the case study on Artificial Intelligence in this chapter. 64 Ž Information Technology R&D: Critical Trends and Issues

is well understood and the practical limits of Packaging is also a critical factor in super- device density, speed and power consumption computer manufacturing costs. Cray machines have yet to be reached. Silicon will be the basis are currently hand wired. In an effort to re- of a growing set of special purpose VLSI ar- duce costs, the Japanese are developing de- chitectures. Gallium Arsenide (GaAs) digital signs that lend themselves to automated man- logic and memory circuits are growing in im- ufacturing procedures. portance, and will be used in the Cray-3.18 University based research in chemistry, phys- Software and Algorithms ics and microelectronics are expected to make The lack of applications software for super- significant contributions to increased inte- computers has been a significant barrier to grated circuit capability. their adoption and use. Packaging One half of recent Cray Research R&D funds have reportedly been devoted to software de- Interconnections among the logic elements velopment, 21 and the vectorizing FORTRAN on some complex chips occupy over half of the compiler, a software program that helps pre- useable chip area, and affect the performance pare standard FORTRAN code for execution speed of chip functions. Currently, three sand- on the Cray vector architecture, has been a ma- wiched layers of interconnection within a chip jor factor in the commercial success of the are typical, and it is expected that as many 19 Cray-1 line.” Users of vector computers, in- as 12 layers will be common in a decade. cluding the Cray, are quite pleased to obtain As chips have become more complex, con- 20 percent of the maximum rated speed of taining greater numbers of logic elements, the these machines on typical problems.23 Work number of “pins,” or inputs and outputs, re- is continuing in industry and universities to quired for communication with them has develop software to make vector machines grown. The connection of sets of chips has more effective and easier to use, and this work thus become more complex, and the difficulty will be of critical importance through this of simultaneously housing and powering chips, decade. and dissipating the waste heat from chip sets The introduction of parallel processing de- is forcing advanced computer designers to find signs and the proliferation of special purpose more sophisticated methods of packaging 20 computer architectures will make software them. production and design more complicated.24 The Cryogenic liquid cooling equipment is re- creation of new high-level languages that are quired for most existing and planned super- more easily understood by users, and other computers. Facilities must be provided for the tools and programming support environments refrigeration and storage of the coolant. And that facilitate the expression of logical, sym- the size, weight, and reliability of the cooling bolic, mathematical, scientific, and engineer- equipment must be considered in the purchase ing concepts in computable form, could greatly and use of these systems. — 2’Rollwagen, op. cit. —.——— *’Nippon Telephone and Telegraph, the Japanese state tele- “L. T. Da~is, op. cit. communications monopoly, has chosen a Cray-XMP over re- “J. A. Armstrong, “High Performance Technology: Direc- cently introduced Japanese supercomputers of comparable or tions and Issues, ” presentation at the Frontiers of Supercom- superior speed, reportedly because of the software, and the ex- puting Conference, Los Alamos, Aug. 15, 1983. perienced team of field representatives, available from Cray. 20MCC is devoting some of their initial efforts to packaging “NTI’ Picks Cray Super CPU,” Ektronz”cAfews, Oct. 10, 1983, technology; and ETA Systems sources estimate that 60 per- p. 87. cent of the R&D effort for the GF-10 will be in packaging. ETA Z~David Ku*, at the OTA workshop on Advanced Computer is planning to use liquid nitrogen cooling to obtain a doubling Architecture, July 14, 1983. in speed from CMOS silicon integrated circuits (L.M. Thorn- *’Paul Schneck, at the OTA workshop on Software Engineer- dyke at the Los Alamos Conference). ing, Nov. 17, 1983. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 65 expand the utility and lower the costs of oper- from the uncompetitive salaries and low job ating advanced computer systems.25 mobility offered by universities. The problem is expected to become acute as demand for In order for the speed potential of advanced computer architectures employing symbolic computers to be used, problems must either processing and artificial intelligence capabil- be programmed to take advantage of the com- ities increase. (See Artificial Intelligence case puter design, or the architecture must be de- study.) signed to handle the unique characteristics of the problem. The mediating factor between the International Efforts problem and the architecture is the algorithm, Japan or the structured procedure for solving the problem. In the future, computer designers Two Japanese firms, Fujitsu and Hitachi, will have to be more cognizant of computa- have introduced advanced architecture com- tional algorithms and the effects of computer puters whose performance is competitive with architecture on programming and problem the fastest available American supercom- solution, and thus they will need greater knowl- puters. Early copies of these machines have edge of applications. Similarly, designers of been installed in three Japanese universities. complex programs, especially scientists, mathe- A third company, Nippon Electric, has an- maticians, and engineers, will need to have a nounced plans to introduce a supercomputer greater appreciation of the inherent capabil- in 1985. ities and limitations of particular computer ar- The Japanese Government is funding two chitectures as they become more dependent on national efforts in advanced architecture re- advanced computers in their work. Collabora- search and development: “High Speed Com- tion between the users and designers of future puting Systems for Science and Technology’ computer systems is critical to both the util- and the “Fifth Generation Computer System” ity and the commercial success of advanced program. architecture computers.26 The Electrotechnical Laboratory of the Manpower Agency of Industrial Science and Technology (AIST), an arm of the Ministry of Interna- There is a shortage of people capable of de- tional Trade and Industry (MITI), is manag- signing software for advanced architecture ing a 10 year project (January 1981—March computer systems. In particular, people skilled 1990) called “High Speed Computing Systems in the design of software and software tools For Science and Technology. ” It is focused on for use in sophisticated scientific and mathe- microelectronics research and development matical applications are scarce.27 There is a (GaAs, Josephson Junctions and High Elec- need for people who understand scientific tron Mobility Transistor devices), parallel problems in a range of disciplines, and who can processing systems with 100 to 1,000 com- design and implement computer systems to puting elements, and systems components, in- solve those problems. cluding mass storage and data transfer devices, to support high-speed scientific and Attracting talented faculty to train the next engineering calculations. Total government generation of computer researchers is a problem. The difficulty results, in large measure, funding will be on the order of $100 million. A consortium of six major Japanese computer —- . —- — — . companies (the Technology Research Associ- “M. B. Wells, “General Purpose Languages of the Nineties, ” ation) has been formed to conduct much of this presentation at the Frontiers of Supercomputing Conference, Los Alamos, Aug. 17, 1983. work in industrial laboratories on “consign- 28 Z60TA Workshop on Advanced Computer Architecture, JUIY ment” from MITI. 14, 1983. ———. - — --— Z7This point was emphasized in one applications area in Par- “’’Super Computer-High Speed Computing Systems for Sci- ticular, telecommunications (Paul Ritt at the OTA workshop ence and Technology, ” Science & Technology in Japan, October- on Advanced Computer Architecture, July 14, 1983). November, 1982, p. 16. 66 ● /formation Technology R&D: Critical Trends and Issues

The “Fifth Generation Computer System” suing work in this area.32 Three “supercomput- program is managed by the Institute for New er” projects are expected to produce machines Generation Computer Technology (ICOT). The by 1985-88, but these systems are not likely program is described in more detail on page to be speed rivals of American and Japanese 105, as part of the artificial intelligence case systems of similar vintage.33 study. West Germany Great Britain West Germany has four universities and six Great Britain has reacted to the Japanese industrial companies working on advanced efforts (in particular the Fifth Generation proj- computer architecture R&D.34 The govern- ect) by establishing a national plan for re- ment is providing about $4 million per year search in advanced computer systems, known for research at the universities on parallel proc- as the Alvey Programme for Advanced Infor- essing. A project at the Technical University mation Technology. It is discussed below in in West Berlin has received a grant from the the Software Engineering and in Artificial In- Ministry of Research and Technology to de- telligence Case Studies. Researchers in Brit- velop a full-scale prototype.35 ish universities have made significant contributions to advanced computer architecture The European Community research. For example, the University of Man- The Commission of the European Commu- chester has an advanced prototype of a “data- 29 nities (EC) has initiated ESPRIT (European flow” architecture machine. Work is also be- 30 Strategic Program for Research and Develop- ing pursued in industry . Inmos, Ltd. has ment in Information Technology) to pursue in- developed the Transputer, a device which com- formation technology R&D on a cooperative bines processing and communications func- basis with industry, universities, and the tions on a single chip. This device has been governments of the EC countries pooling their specifically designed for the connection of a efforts. Four institutions (three Belgian and number of units to achieve concurrent proc- 31 one French) have thus far announced plans to essing. study parallel processing with ESPRIT funding.36 The proposed ESPRIT plan calls for ‘he France development and use of computerized facilities France has considerable interest in advanced to study new computer designs.37 (For more computer architecture research and develop- information on ESPRIT, see ch. 7.) ment. Two industrial companies, Cii-Bull and CGE, as well as seven government funded in- ‘z’’ Next–Generation Computing: Research in Europe, ” IEEE Spectrum, November 1983, pp. 64-65. stitutions including five universities are pur- 39Report of the IEEE Super Scientific Computer Comm.f”ttee$ . Oct. 11, 1983. ‘9A. L. Davis, “Computer Architecture, ” IEEE Spectrum, “IEEE Spectrum, November 1983, op. cit., pp. 67-68. November 1983, pp. 98-99. “’’Western Europe Looks to Parallel Processing for Future ‘°Five have been identified, see “Next-Generation Comput- Computers, ” Ekctrom”cs, June 16, 1983. p. 111. ing: Research in Europe, ” IEEE Spectrum, November 1983, “A. L. Davis, op. cit. pp. 65-66. “Proposal for a Council Decision adopting the first European ““lhnsputer Does Five or More MIPS Even When Not Used Strategic Programme for Research and Development in Infor- in Parallel, ” Iann Barron, Peter Cavill, David May and Pete mation Technologies (ESPRIT), Commission of the European Wilson, Electronics, Nov. 17, 1983, p. 109. Communities, June 2, 1983, p. 24. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 67

Case Study 2: Fiber Optic Communications

Findings Advantages

● Fiber optic communications technology is Fiber optic communication is the transmis- an important export (e.g., a $3 billion world sion of light signals through transparent glass market projected for 1989), and is a key ele- or plastic fibers, where the signals are gener- ment in improved productivity and reduced ated by lasers or light emitting diodes (LEDs) costs in telecommunications systems. and received by photodetectors, which convert ● Fiber optic communications technology is light signals to electrical signals. The major developing rapidly and the potential bene- components of fiber optics technology are the fits from continued research are extensive. fiber cables and connectors, transmitters and ● The most significant research is concen- receivers, and repeaters, or regenerators,39 trated in only a few large firms and univer- which amplify and reconstitute the signals pe- sities, in part because of the expense and riodically along the fiber. the required long term commitment. There are several properties of fiber optic ● Increased Government funding in this tech- communications that make them attractive nology-now at a low level—would enlarge for telecommunications applications: university participation and accelerate technological advances. ● large bandwidth, meaning that large ● The scarcity of trained research and devel- amounts of information (voice conversa- opment personnel is a continuing handicap. tions, computer data, graphics) can be A large proportion of university research- transmitted rapidly. For example, a quar- ers in fiber optics are foreign nationals and ter-inch diameter optical cable with two many return to their native land after com- fibers carries as much data as a 3-inch pleting graduate studies. copper cable with 20,000 wires.40 (See fig. ● Research conducted in Japan and Europe 9.) Conservatively, the capacity of a single is among the world’s most advanced and ex- pair of fibers currently available commer- change of research information internation- cially is about 4,000 voice grade circuits ally among colleagues is critical to scientific in field applications and about 400,000 advancement. voice grade circuits under controlled lab- ● Japan is the world’s leader in several key oratory conditions; aspects of the technology, and is a strong ● less susceptibility than copper wire to competitor to American firms. radio frequency interference, providing less cross-talk, higher quality signal The first successful transmission of voice transmission, and immunity from electro- signals using energy from the Sun was accom- magnetic pulse (EMP) effects-character- plished in 1880 by Alexander Graham Bell and istics of value for both civilian and mili- Sumner Tainter using a device called the photo- tary uses; phone, but was abandoned because weather ● 38 lower loss of signal strength, meaning made the system unreliable. Since 1970, in- that fewer repeaters are needed; terest has resumed in using light energy for ● resistance to “noninvasive” or covert telecommunications, in the form of fiber op- wiretaps; tic communications because of two technological advances: the laser and light transmission . . .— through low loss silica glass fibers. sg]n Stmdwd copper telephone wires, Signal regeneration is required at about onemile intervals. Repeaters add significantly —. —..— to the installation and maintenance costs of transmission 38 Report on Research at the University of Arizona, vol. 1, No. systems. 1, fall 1983, p. 11, published by the Research Office, Universi- 4~High Technology, “Fiber Optics: Light at The End of the ty of Arizona. Tunnel, ” March 1983, p. 63.

68 ● Information Technology R&D: Critical Trends and Issues

Figure 9.—Optical Fibers Are Small, Lightweight, and Versatile

Photo credits: AT& T Sell Laboratories Ch. 3—Selected Case Studies in Information Technology Research and Development ● 69

small size and low weight, factors that tance applications are currently the most cost contribute to ease of transport and less effective for fiber optic communications, and need for underground and building duct comprise the vast majority of current use in space; the United States. declining cost compared to other terres- 41 Interoffice trunking provides links between trial telecommunications technologies. intracity telephone facilities. Feeder lines include intracity transmission links between car- Commercialization Trends rier facilities and subscriber distribution points, The technology has progressed rapidly into while local feeder lines extend to subscriber the commercial marketplace since 1966, when locations. Local area networks (LANs) serve researchers 42 first proposed the possibility of limited communities, for example, within a purifying the glass used in optical fibers to re- building or a building complex. They are just duce losses in signal strength. In 1970, the emerging and will become an important mar- first low loss fiber was produced by Corning ket for fiber optics. Glass Works, and by 1977 prototype systems There are a growing number of installations were being installed by AT&T and General of fiber optic systems in long distance telecom- Telephone and Electronics Corp. in the United munications. Among these are AT&T’s North- States. Today, many developed countries have east Corridor route. In 1983 a line connecting fiber optic communication systems in opera- Washington, DC and New York City was in- tion or plan to install them. augurated. In 1984 this line was extended to The United States, Canada, Western Eur- Cambridge, MA, and Richmond, VA. The to- ope, and Japan accounted for an estimated 96 tal Northeast Corridor line will use 45,000 percent of a $550 million world market in fi- miles of fiber over the 750 mile route. The first ber optics communications equipment in 1983. phase of a west coast route, which will even- The world market was approaching $1 billion tually extend from Los Angeles to Oakland for 1984,43 and is projected to expand to $3 bil- and Sacramento, has been completed. A trans- lion by 1989,44 as countries satisfy their tele- atlantic cable (TAT 8), engineered to carry communications system expansion and re- 40,000 voice circuits at a cost of less than half placement needs with the increasingly cost- that of its predecessor, is scheduled for in- competitive fiber optic communication sys- stallation in 1988. By March 1983, AT&T had tems rather than with microwave radio, cop- already installed over 100,000 miles of fiber per twisted wire pair, and coaxial cable. and projections are for another 300,000 miles by 1990.4’ Applications in the United States Other firms planning major systems include Telecommunications, the major market (85 United Telecommunications, Inc., with its 45 23,000 mile, $2 billion lightwave network to percent) for fiber optics, can be described in 47 four segments: long distance; interoffice trunks be completed by 1987. Southern New Eng- that connect telephone central offices; local land Telephone Co., in a joint venture, will feeder lines; and local area networks. Long dis- route its system through 20 States along rail- —— road rights of way; MCI, with a 4,000 mile sys- tlThiS Compuison is based on the relative cost per chmnel- tem intended to serve the east coast;48 and mile, which is the number of voice circuit equivalents (chan- Cable and Wireless, a British company, with nels) multiplied by the distance of the transmission link. 42Kao and Hockrnan, I(IT Standard Telecommunications Lab- oratories, England. 43D. G. Thomas, “Optical Communications, ” Research and 4Qptoelectronics Supplement to JZkctrom”c News, “Fiber Op- Development, June 1984, p. 203. tics Market Still Baffles Suppliers, ” p. 5, Apr. 4, 1983. “Signal, September 1983. 47’’ 23,000 Mile Fiber Network Planned by United Telecom, ” “The remaining 15 percent of the noncommon carrier applica- The Journal of Fiber Optics, June 1984. tions are said to be in vehicular,, industrial control systems, 4aFiber Optics Industry Service: Competitive Environment, and in CATV. High Technology, op. cit. Gnostic Concepts, Inc., 1983, pp. 1-2. 70 ● Information Technology R&D: Critical Trends and Issues

a 560 mile network that will link major cities In applications other than communications in Texas. carrier uses, such as aerospace and military systems, fiber optic technology is already be- Fiber optics is already an attractive replace- ing exploited. These are primarily in comment for copper cable in some local telecom- mand, control, and communications applica- munications applications: 1) in trunk lines be- tions including guidance and control systems tween telephone central (switching) offices for aircraft, spacecraft, and missiles; optically where the average 10-mile distance can be multiplexed data bus transmission systems; spanned by optical cables without repeaters, electronic warfare and sonar applications; and thus installation and maintenance costs are advanced instrumentation systems. The im- lower, and 2) in local feeder lines in large cit- perviousness of fiber optics to electromagnetic ies where crowding in underground utility interference, along with light weight, small ducts is a growing problem. size, and high information rates, make it of Local area network applications are expected special value in aerospace and military appli- to grow rapidly according to some industry cations. projections, particularly as user requirements for bandwidth increase to the 10 to 100 mil- Foreign Applications lion bits per second (Mbs) range. The past slow Installation of fiber optic communications growth of fiber optic applications in LAN sys- systems have been completed, or are planned tems is due to the variety of technical needs in a number of countries. A small sample of among different customers, and a lack of uni- these includes: form technical standards. In contrast with long distance communications systems, which ● West Germany. By late 1984, 10 broad- have traditionally paid extensive attention to band integrated fiber optic local area net- technical standards development, a Federal works will be built in Berlin, Hamburg, telecommunications standards committee only Hanover, Dusseldorf, Nuremburg, and recently (July 1984) held its first meeting to Munich, called the BIGFON (Broadband develop Federal guidelines for LANs. Once Integrated Fiber Optic Local Network) standards issues are settled, growth in the use network. In addition to having access to of fiber optics within buildings and building the public switched telephone network, complexes is expected to be rapid.49 subscribers will also be able to access the integrated telex and data network and re- Other factors limit the adoption of fiber op- ceive radio, television, telephone, and full tic communications technology. The large base motion picturephone. The long-term goal of installed copper wire in the AT&T plant is to include all telephone subscribers in (some 827 million miles) is likely to be replaced the nation. slowly. The large capacity (up to 100 televi- ● France. Several fiber optic systems are be- sion channels) of some CATV systems and the ing installed by the French. The govern- fact that this expensive investment in coax- ment has decided to upgrade the nation’s ial cable has been made quite recently in many antiquated telecommunications network cities, suggests that fiber optics will not be by leapfrogging toward the most advanced widely used for cable television for some time. technology available-especially fiber op- Advances in nonoptical transmission tech- tics. They also plan to install a submarine niques are enabling considerable increases in cable between France and Corsica in 1985, the information carrying capacity of copper providing over 7,600 channels operating wire pairs. at 280 Mbs. ● Japan. One of several projects being undertaken is an 80 km fiber optic route ‘gAviation Week and Space Technology, “Promising Future in the suburbs of Tokyo. The system oper- Seen for Optical Fibers,” pt 1, Oct. 12, 1983, pp. 44-77, ates at 400 Mbs, providing video confer- Ch. 3—Selected Case Studies in Information Technology Research and Development ● 71

encing and facsimile services. In addition, research in 1960 and related laser research in the Japanese have installed a broadband 1958. network for data communications in There is concern that many U.S. companies Tsukuba to facilitate scientific and tech- have been inclined to undertake research only nical communications. where the prospective payoff is likely to oc- An Intelligent Network System (INS) cur within a very few years. This attitude has is being developed by Nippon Telephone the effect of shifting investigation away from and Telegraph’s Yokosuka Laboratory promising areas such as research on infrared near Tokyo. Services include voice, data, systems, where another 5 to 10 years of work CATV, still and motion pictures, facsim- may be required before commercial products ile, TV conferencing and high resolution become available. Thus, the importance of TV. The INS is expected to make exten- stable Federal funds for basic research is sive use of fiber optic technology. Another underscored by short-term planning within in- project involves a 45 km undersea fiber dustry. optic cable south of Tokyo, operating at 400 Mbs. The system will be extended The expense of research in fiber optics over a 1,000 km route between islands. makes it difficult for small firms to play a role in R&D, except in some areas of product com- ● British Telecom, the na- United Kingdom. mercialization. In addition, there is a tendency tion’s telecommunications authority, has for equipment purchasers to prefer vendors of committed itself to using only fiber optics complete lines of components, which works to in the trunk network from 1984 on. By the detriment of small firms. As a result, small 1990, half the trunk network will be fiber specialty firms often must rely on DOD for re- optic systems. search funds, on takeovers by larger firms, or Mercury Communications, a communi- on venture capital in order to remain competi- cations carrier in limited competition with tive. Regardless, these firms play an impor- the government telecommunications au- tant role in the technology by providing in- thority, has begun installing an intercity novative ideas and products, by serving as fiber optic network using the British rail conduits for the commercialization of univer- rights of way. sity-based research, and by filling niches that The British plan to construct a nation- might not be attractive to larger firms. wide cable telecommunications system for the delivery of television programming, Because of the considerable expense asso- FM radio, pay television, and text. An ex- ciated with research in fiber optics technology, periment with a small number of homes and the low level of available funding, few is being conducted using fiber optics tech- universities have major research programs. nology. This is considered to be the pro- The importance of cost is illustrated by the totype for the national system. $500,000 or higher cost of Molecular Beam Epitaxy equipment (needed for growing alter- United States R&D nating epitaxial layers on semiconductor light sources and detectors) and $200,000 for fiber Much fiber optics R&D in the United States drawing equipment (needed to produce fiber is being conducted by a few large companies: and to experiment with different fiber designs) AT&T Bell Laboratories; Corning Glass Works; required to perform research. Very little uni- ITT; and, to a lesser extent, GTE; a few university research is focused on glass fibers, but versities; and some smaller companies. Many instead is directed toward light sources and of the commercial products now available are detectors. The principal universities with ma- a result of R&D performed by Bell Labs. jor research programs are the California In- AT&T began funding optic communications stitute of Technology, the University of 11-

38–802 O - 85 - 6 72 • Information Technology R&D: Critical Trends and lssues

linois, Stanford, Cornell, Princeton, and the Cooperative Research Massachusetts Institute of Technology. NSF has several activities directed at encouraging cooperative research between uni- Government Funding versities and industry, and transferring tech- University research has been supported for nology into industry. One of these noted in decades through NSF funding for the support chapter 2, the Industry/University Coopera- of research and, more recently, training and tive Research Centers program, provides plan- laboratory equipment. NSF supports several ning grants to aid universities in establishing universities, such as the University of Arizona industry affiliations and support for specific in Tucson, Northeastern University, and Cor- scientific or engineering technologies. nell University’s Submicron Facility. Although NSF funding levels in this technology are not NSF has awarded a $75,000 grant for plan- large $1.75 million in 1983–they represent ning purposes to the University of Arizona at a consistent source of funds, often supporting Tucson, Optical Sciences Center. The Univer- sity held its first meeting with industry in fundamental research with very long term potential payoff. Additional levels of funding early 1984 to begin determining mutual inter- would likely accelerate the rate of technologi- est in specific areas of cooperation and indus- cal advance. try support. Most of this research is expected to be directed toward long term projects in Between 18 and 20 projects are being funded physics and materials science with potential by NSF in about a dozen universities. Many applications in optical logic circuitry and op- of these are concerned with advancing theo- tical computers, with limited attention to fi- retical knowledge in areas such as laser tech- ber optics. nology, the development of pioneering optic Another NSF activity funds specific proj- and optoelectronic integrated systems and ects where research is undertaken cooperative bistable optical switching devices, research into infrared lasers and detectors, and the ap- ly by universities and company investigators. Funding is at levels of about $100,000 per year plication of integrated optical interface circuits in local area networks at gigabit (billions of over a 2 to 3 year period, on the average. One of the funded projects is being undertaken bits) per second data rates. Another $300,000 of NSF funding is available for upgrading jointly by Bell Laboratories and the Univer- sity of Arizona, Optical Sciences Center. This university laboratory equipment. project’s long-term research is in high speed The DOD funds fiber optics research through optical, bistable switching devices operating mission-oriented procurements. Approximately at picosecond (1 trillionth of a second) rates. 95 percent of the research is carried out by in- The current level of cooperation in research dustry. DOD has some $32.4 million commit- is not extensive, but holds promise for broad- ted to the development of cables and connectors, light sources and detectors, radiation ening the base of research. Among the problems and issues to be worked out are finding effects exploration, and to sensor and communications applications. An estimated $12.6 ways for competing firms to share research data and establishing a balance between uni- million of this is committed to research, prin- versity investigators’ interest in long-term re- cipally applied research. In addition, the mil- search and companies’ desire for short-term itary departments allocated about $22 million payoffs. for fiscal year 1984 among seven procurement programs for applications ranging from sur- Manpower and Industry Support veillance, shipboard and long distance communications, and helicopter flight control Fiber optics research requires training in systems. both physics and in electrical engineering. Be- Ch. 3—Selected Case Studies in Information Technology Research and Development ● 73 cause universities, in general, do not provide Today’s single mode fiber systems are able to this cross disciplinary training at the bache- transmit, without repeaters, up to 200 Mbs lor and masters degree levels, companies hir- (million bits per second) over 80 to 100 km. ing recent graduates provide supplementary (This information rate is sufficient to carry in-house training. Some companies also estab- simultaneously a video channel, high fidelity lish an affiliation with universities. An exam- audio, data, and many telephone calls.) In lab- ple is the affiliation between Corning Glass oratory tests, this performance has been ex- Works and the University of Rochester Insti- ceeded by about 10 times, suggesting far tute of Optics, in which the company provides greater potential gains from research. Im- some faculty and funding. Another example provements from research are expected to con- is the support from United Technology Re- tinue in both multimode and single mode search Center to the Rensselaer Polytechnic fibers. Institute which is providing real estate for new Improvements have been made in lowering facilities and adjunct faculty to teach specialized engineering courses. attenuation (losses in signal strength) in both types of fibers by a factor of 100 since 1970, The University of Rochester is the only primarily due to development of methods to school in the United States to offer an under- reduce impurities in the fibers. graduate degree in optics. Only three schools Activities are being directed toward further in the nation offer graduate degrees in optics: improvements in optical fibers. These include The University of Arizona in Tucson, the research into different cross sections for fiber University of Rochester, and Northeastern University. cores, such as circular, triangular, and oval, as investigation into new materials such as A factor aggravating the availability of plastics, and improvements in fiber splicing trained Ph.D. graduates in this field is that techniques. Research into plastic materials is a large proportion of the students are foreign about at the stage of 1975 era research in glass nationals. Industry argues that immigration fibers, and promises even lower cost, more laws make it difficult for the student-graduate durable fiber materials for some applications. to remain in the United States after gradua- Longer wavelength (1.7 and 4.0 micron) mation, although many would prefer to stay and terial for fibers is also receiving attention, as perform research. these show promise of decreased attenuation by a factor of 10 to 100 over that of currently Directions of US. Research available fibers, with long-term prospects for Among the areas of R&D focus identified transcontinental or transoceanic transmission during the course of this study are: without the need for repeaters.

Fibers Light Generators and Detectors The early fiber optic cables put in use were Light generator and detector technologies of the multimode type, in which lightrays en- are important areas of research. Research is ter the fiber at a variety of angles and travel continuing to improve the lifetimes of these through the core of the fiber reflecting from devices, their spectral stability, the narrow- its inner refractive surfaces. However, single ness of spectral emissions, switching speeds, mode fiber technology, where lightrays follow current threshholds, and receiver sensitivity. a single direct path along the fiber core, has The most recent advance is the cleave-coupled important advantages. Single mode fibers cavity laser—a device notable for its wave- have greater information carrying capacity length stability and capability of changing fre- and allow a tenfold increase in the distance be- quencies rapidly, making it attractive as a tween repeaters for regenerating signals. multisource generator. It has been demon- 74 . /formation Technology R&D: Critical Trends and Issues strated at 274 Mbs over 100 km of fiber optic for reducing the number of conversions re- cable by Bell Laboratories, and at 1.6 Gbs quired from optical to electronic and vice (billion bits per second) over 40 km by the versa. Current switches are expensive, limited Japanese. in applications, and of unpredicted reliability. Coherent detection, a technique to improve Optical switching research is being con- receiver sensitivity and to increase the infor- ducted at Bell Laboratories and the Univer- mation carrying capacity of fibers, is being sity of Arizona, where experimental, room pursued in many research laboratories, and temperature switching rates, for “turn on, ” of may become important if a variety of obstacles 50 picosecond (50 trillionths of a second) have can be overcome. been measured. Research is also continuing into ways of in- Storage tegrating light sources, detectors, and the associated circuitry into single chips thus Research is continuing into methods for lowering cost and increasing reliability. storing optical signals on fixed and volatile memory devices. Improvements in storage will Optical Multiplexer make possible store and forward and electronic mail features for optical networks. Optical multiplexer (and demultiplexers) are devices that combine (or separate) different Amplifiers and Repeaters signals so they can be sent through the same optical fiber. Wavelength multiplexing is al- Regenerative repeaters detect a signal, then ready being used in AT&T’s east coast and amplify, reshape, and retime it into a replica west coast systems, and a few experimental of the original signal. The regenerated signal systems in Japan, Canada, and Europe. Re- then modulates a laser or light emitting diode search in wavelength multiplexing techniques for transmission along the next span of optical should lead to important cost savings for wide- fiber. Decreasing the number of repeaters re- band systems. quired along a line depends in part on amplifi- cation capabilities. Connectors and Splicing Techniques Integrated Circuits Research is continuing to simplify techniques for splicing together separate segments Research is continuing to improve capabil- of optical fiber and to achieve lower losses due ities for putting optical and optoelectronic to the splice. light generators and detectors onto single in- Bell Labs recently announced the develop- tegrated circuits, and to increase the operational bit rates. Recent breakthroughs hold ment of an ultraviolet splicing system that promise for reducing the number of discrete contributes only 0.03 decibels to signal loss, components required in fiber optic systems using an optical test signal to assist in the and expanding bit rate capabilities. alignment of fibers.

Switches Research in Japan Switching permits a signal to be routed While research in fiber optics is being ac- through specific paths to subscribers. Switch- tively pursued in the United Kingdom, France, ing is a bottleneck in optical communications and West Germany, the most advanced for- systems because the conversion of signals eign research has been undertaken in Japan. from light to electronic (current switch technology is electrical) causes delays in moving Japan’s research is being conducted, at least the signals through the system. Improve- in part, to support the development of a new ments in switching capabilities hold promise nationwide broadband telecommunications Ch. 3—Selected Case Studies in Information Technology Research and Development ● 75 network that will make extensive use of fiber tors. The budget is approximately $100 million optic technology. Research is also being sup- for the 1979-87 time period. Half of this amount ported for future commercialization and inter- is to be devoted to a coordinated research fa- national markets. Fiber optic research sup- cility, information exchange among research- porting this network is performed mainly in ers, and the remainder on projects in six or private companies, and is supported by the seven companies, including Hitachi, Nippon Ministry of International Trade and Industry Electric Corp., Toshiba, Mitsubishi, Fujitsu, (MITI) through the Optical Measurement and and Matsushita. Japanese investment in in- Control System (OE) project. OE plans to de- frared laser research is estimated at between velop optoelectronic integrated circuits, tran- $3 million and $4 million. sistors, and GaAs/GaAIAs lasers and detec-

Case Study 3: Software Engineering

Findings ● The Federal Government, through the De- partment of Defense, is making some efforts Software is an important factor in informa- to create a software engineering applied tion technology, exceeding four times the research base for national defense purposes, cost of hardware in large systems. The rela- but the applicability of this research base tive decline in hardware costs is shifting the to the general problem of software produc- focus of R&D to software. The complexity tivity in the American economy is un- of new applications and information sys- certain. tems is also forcing the focus of information ● It is difficult to differentiate software pro- technology R&D toward software issues. duction activities from R&D, especially de- R&D in software engineering has produced velopment. Much software production is a prototypes of software design tools and pro- creative design effort. There are some as- gramming environments (integrated sets of pects and types of programming that are tools) that promise significant productivity clearly not R&D, but the dividing line is dif- increases. But the cost of retooling, includ- ficult to define. ing retraining software manpower, and the uncertainty associated with innovation Introduction in software development are retarding adoption of innovative tools and techniques. The term software refers both to the instruc- In order to speed the acceptance of software tions that direct the operation of computer engineering innovations, an applied research systems, and the information content, or data, base needs to be established to scientifically that computer systems manipulate. Software test and validate new software development is thus a logical rather than a physical prod- techniques, and to disseminate information uct. An adequate organizing formalism or on their performance in specific applications calculus for software creation has not yet been environments. Such an applied research discovered. 50 Therefore, the development of base would link basic research in univer- large, complex software systems depends sities to applied research and development heavily on the insight and creativity of sys- efforts in industry and government. Foreign efforts, particularly those in Japan 50 but also national targeted efforts in Europe, ” We are in a business that is 35 years old . . . and I invite you to think where civil engineering was when it was 35 years show signs of movement toward such an ap- old . . . they had not discovered the right angle yet. ” Harlan plied software engineering research base. Mills at the OTA workshop, Nov. 17, 1983. 76 ● /formation Technology R&D: Critical Trends and Issues

terns designers and programmers. The meth- fort among researchers, educators, data proc- ods presently employed to develop and test essing managers, systems designers, and pro- software are ad hoc, without a strong scien- grammers, and support from corporate and tific basis. Thus, software systems are expen- Government management is required for this sive to build and maintain, and can be unreli- transformation to occur. able in operation. The problems associated with inadequate software methods are grow- Software R&D Environments ing as information technology uses spread and are relied on for larger, more complex, and Varieties of Software and Characteristics more critical applications. of Software Development Research and development efforts have pro- Software is classified as being of two gen- duced new methods that show promise for im- eral types: applications software that is de- provement in the productivity and reliability signed to apply computer power to a specific of software creation, testing and maintenance. task or tasks, such as computer-aided design These include Advanced program editors and of automobiles, or payroll or inventory man- debuggers, new languages, design methodol- agement in a department store; and systems ogies and integrated programming environ- software that is used to manage the compo- ments. However, the introduction of innova- nents of an information system itself, such as tive techniques into software production is computer operating systems that control in- proving difficult. The adoption of new meth- put and output operations. ods often requires the conversion of large ex- In general, systems software is an integral isting program inventories. This is expensive component of the hardware because its job is and risky because evidence that one method- to control the hardware, including the peri- ology is better then another is not systemati- pheral equipment–e.g., disk, printer, and cally collected. As well, most software devel- memory usage, and to schedule and accommo- opment is oriented toward the single project date the creation and execution of applications at hand and often relies on antiquated pro- programs. A recent trend, encouraged by the gramming habits and attitudes. The high job spread of personal computers, has been toward mobility of programmers and software sys- the use of standard systems software so that tems designers perpetuates individualism and a large number of applications programs can fragmentation in software methodology. be made comparable with hardware from dif- The establishment of software engineering ferent suppliers. The manufacturers of hard- then, involves not only the development of ware or specialized software vendors write superior methodologies, but a transformation most systems programs, and more of the sys- of the programming process—from an art to tems programs are being embedded in hard- a science and from a labor intensive to a capi- ware-in programmable integrated circuit tal intensive effort. The pressures on software memory called ROM (Read Only Memory). production for larger, more complex, more cost Thus end-users now generally do not create or effective, and more reliable systems, make the alter systems programs. present situation untenable.” A concerted ef- Much applications programming is done by ——— ... —- —-—— users. For personal computers in homes and ‘l’’ According to a projection made at an early 1981 data proc- for supercomputers at the National Weather essing managers conference, Department of Defense software Service, programs must be written to tell the costs would increase nearly three times as fast as the department’s budget, and nearly 20 percent faster than expenditures machines how to solve problems and organize for computers during the 1980s, ” B. M. Elson, “Software Up- date Aids Defense Program, ” Aviation Week and Space Tech- nology, Mar. 14, 1983, p. 209. Just as the explosion in num- ming seems to be increasing the pressure for the introduction bers of telephone operators in the 1930s and of bank clerks for of less labor intensive software development techniques; see check processing in the 1960s forced a move to automated sys- T. C. Jones, “Demographic and Technical Trends in the Com- tems, the sheer demand for manpower in computer program- puting Industry, ” DSSD User’s Conference.

information. Much of this work is done by software as not eligible for tax credits, unless highly trained professionals knowledgeable the software is “new or significantly improved, ” about computer systems and the problems to or “if the programming itself involves a sig- be solved. nificant risk that it cannot be written. ”52

Not all applications programming is soft- Software R&D in Industry ware R&D. For example, an economist writing The information processing industry is a short program to calculate a unique set of devoting a large and growing amount of re- statistics that may never be used again or by sources to software development. Purchases others is not engaged in software R&D. Con- alone, currently some 12 percent of total versely, a physicist developing a program for spending for software, are expected to exceed a supercomputer to calculate formulas used in $10 billion in 1984. Approximately 10,000 nuclear reactor design certainly may be in- companies of various sizes, from one man oper- volved in R&D. There is a large area in be- ations to divisions of major corporations, are tween these examples that is ambiguous, and developing software for sale. Estimates of the no hard data is available concerning the time total number of programmers in industry spent creating different categories of pro- range from 500,000 to nearly a million,53 and grams. The Internal Revenue Service has ——-—— faced this difficulty in defining the types of ‘*’’Credit for increasing Research Activity, ” Federal Regis- software work that are eligible for the R&D ter, Jan. 21, 1983, pp. 2799-2800. There has been considerable controversy over this issue. See W. Schatz “A Taxing Issue, ” tax credit. Their proposed solution has been June 1983, pp. 58-60. to consider the costs of developing computer 53ADAPS0 estimates, and T. C. Jones, op. cit. p. 83. 78 ● Information Technology R&D: Critical Trends and IssueS the distinction between hardware and soft- Systems Design, Theoretical Computer Science ware developers is becoming increasingly and Special Projects Programs) bringing the blurred.54 total funding to $5 million to $10 million per year.59 Every major computer and telecommunications equipment manufacturer and service pro- The Federal Government is the world’s larg- vider develops programs and software tools est user of data processing resources. A recent to make their products suited to the needs of GAO study found that 95 to 98 percent of the users. Some firms such as Hewlett-Packard, Government’s applications software is custom reportedly spend nearly tw0thirds of their developed. 6o The cost of software for the De- R&D budgets on software.66 More than 40 per- partment of Defense is estimated to be $4 bil- cent of the technical people at Bell Labs are lion to $8 billion per year.61 DOD operates a involved with software development.56 patchwork of incomparable systems and computer languages.62 The incomparability of soft- Interest in the improvement of software pro- ware contributes to increased software devel- ductivity has led some large corporations to opment costs, through schedule slippages, establish formal programs for the develop- lengthy testing programs, and problems in ment and evaluation of software practices—to contracting for hardware and software services. introduce scientific and engineering techniques into the evaluation of software devel- DOD has moved toward the development opment and use. AT&T currently employs ap- and use of a single standard computer lan- proximately 300 Ph.D.s in software research, guage. The rationale for this is the potential while IBM has approximately 150 and ITT for saving several hundred million dollars a has about 20 of these researchers developing year through lower personnel training costs, and using formal experimental methods of increased programmer productivity, and sub- evaluating software engineering techniques. stantial reuse of standard code modules. After Other large companies, including Control competitive development of four separate lan- Data, Xerox, and Honeywell, are also begin- guages and extensive design evaluation, the ning to use experimental R&D studies as the Pentagon chose a language developed by the basis for improvements in software produc- French company Cii Honeywell Bull. The tion.57 name of the language, Ada, is trademarked, and compilers using the Ada name are strictly The Federal Role controlled and validated to assure that the language remains standard. Ada is expected to The National Science Foundation funds uni- be the primary DOD computer language by versity and some corporate research in soft- 1987. ware engineering. The NSF Software Engineer- ing Program within the Division of Computer There is some resistance to use of Ada. Sev- Research awarded $2.2 million in grants in eral years ago, the Air Force developed its own fiscal year 1983.56 Additional research related quasi-standard language, Jovial. Comparison to software engineering is funded by other pro- tests between Ada and Jovial will continue for grams in the Computer Research Division (e.g., the Software Systems Science, Computer — — . —— 59Estimates by OTA. 54See, for example, S. B. Newell, A. J. De Geus, and R. A. ‘“’’ Federal Agencies Could Save Time and Money With Bet- Rohrer, “Design for Integrated Circuits, ” Scieme, Apr. 29, ter Computer Software Alternatives, ” General Accounting Of- 1983, pp. 465-471. See also “The Changing Face of Engineer- fice, GAO/AFMD-83-29, May 20, 1983, p. 1. Even standard ap- ing, ” Electrom”cs, May 31, 1983, pp. 125-148. plications such as payroll are largely custom designed. This ‘SW. P. Patterson, “Software Sparks a Gold Rush, ” indus- GAO report found that there are at least 78 different Federal try Week, Oct. 17, 1983, pp. 67, 69-71. civilian payroll systems. 56AT&T: 1982 Annual Report, p. 19. elElson, Op. Cit., p. 209. ‘70TA workshop on Software Engineering, Nov. 17, 1983. ’21. Peterson, “Superweapon Software Woes, ” Science News, 58Summary of Awards, Fiscal Year 1983, National Science May 14, 1983, pp. 312-313. Testing of software alone is esti- Foundation, Division of Computer Research. mated to account for nearly half of this cost. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 79 the rest of the decade.63 The Army is more en- Content and Conduct of Software thusiastic about Ada. The Navy, which has Engineering R&D more software already written than the Air Force and Army combined, is interested, but Software engineering ideally is a set of con- the task of switching the Navy to Ada will be cepts and tools for transforming descriptions enormous. The Navy has over 450 different of tasks to be performed by computer systems systems and subsystems with embedded com- into digital code that machines can under- puters, and the number of Navy computers stand. Software engineering research involves has been doubling every 2 years.64 the study of methods to understand, improve, implement, and evaluate these concepts and A limitation of Ada is that a new program- tools, and to embody them in software devel- ming language only addresses about 20 per- opment systems or programming environ- cent of the total software problem. As detailed ments to facilitate the entire software lifecycle. below, coding of computer programs is 20 percent or less of the effort of developing and The Software Lifecycle maintaining software. Government computer systems in particular require an enormous As can be seen in figure 10, there are sev- amount of documentation, and Ada has no fa- eral stages in the life-cycle of a piece of soft- cilities for automated documentation.65 ware. Each of these stages is characterized by its particular set of objectives and methods To deal with these problems of software de- that influence each stage of research and ex- sign, production and maintenance, DOD has pected improvements. Testing occurs continu- proposed a new initiative called Advanced ously throughout the life of a software and is Software Technology.66 This is envisioned as 67 an integral part of software production and a 10 year program costing $250 million. The maintenance. Documentation, an activity not administration requested a funding level of depicted in figure 9, is of preeminent impor- $19.3 million in fiscal year 1984, but the Sen- tance in every phase of the software lifecycle. ate Armed Services Committee cut the pro- Comprehensive records of every activity and gram to $10.5 million because, “The Com- software characteristic must be created to aid mittee is not convinced that the necessary designers, programmers, maintainers, and planning has been done to justify a budget of 68 users in understanding the structure and oper- nearly $20 million in the first year. ” Included ation of the software system. in the plan, as it has thus far been developed, is a provision for a Software Engineering In- Requirements Specification stitute to help formulate and standardize software engineering techniques and practices.69 This initial phase of software development involves the description of the system objectives and the tasks that the end users want performed. This description requires a thor- —.— ough knowledge of the application by the soft- “J. Fawcette, “Ada Tackles Software Bottleneck, ” High ware design project team. It is accomplished Technology, February 1983, pp. 49-54. 64petergon, op. cit., p. 313. It has been estimated that there by rigorous and continuing communication be- are some 50 million unique lines of Navy software code in a va- tween the project team and the end users. riety of languages currently in use, It would cost, it has been reported, some $85 billion and take several years to rewrite this OTA advisors and published sources empha- mass of code. size that requirements specification is the e5The OTA workshop on Software Engineering, NOV. 17, 1983. “This initiative is known within DOD as STARS, Software most critical phase of project development, be- Technology for Adaptability, Reliability and Serviceability. cause all of the later stages must build on the eTPeterson, Op. cit. eaomm”bus Defense Authorization, 1984, Report to Accom- foundation laid in this activity. At the same pany S.675, Committee on Armed Services, U.S. Senate, p. 131. time, it is the most difficult activity in the soft- ‘BOTA Workshop on Software Engineering, NOV. 17, 1983. ware lifecycle to develop a rigorous method- 80 ● Information Technology R&D: Critical Trends and Issues

Figure 10.—The Software Life Cycle

SOURCE: Data Communications. ology for. Requirements specification involves converted into machine code. Research is pro- the understanding, explicit organization, and ceeding in knowledge representation and integration of what are often idiosyncratic knowledge engineering (see Artificial Intelli- practices of information usage. Often it is not gence Case Study). Near term prospects are possible to completely specify requirements at uncertain for identifying core concepts for the beginning of a project; also, requirements organizing the different kinds of knowledge can be expected to change over the course of found in the variety of environments in which a long software development project. computers are applied, or designing broadly Computerized tools now available or under applicable very-high-level languages which can development may offer assistance in the speci- automatically render abstract specifications in machine-executable procedures.71 fication process. 70 They are designed to help users describe and specify system properties, The experience now being gained in research functions and performance requirements. and development of “expert systems, ” and the Some of these tools allow the specifications to more fundamental research efforts in knowl- be checked by computer for consistency and edge representation, machine learning and correctness; some of them can generate simu- human cognition should eventually contribute lations to help the user analyze the operation to the process of computer systems require- of a specified system. Thus far these tools have ments specification. But for the near term, this worked well only on a limited range of ap- phase of software development will remain la- plications. bor intensive and will require special skills (including human relations) and specific knowl- A difficult problem is specifying how the edge of applications among its practitioners. knowledge, experience and habits of workers in particular environments can be described and how these abstract specifications can be ——.— “R. Yeh, “Software Engineering, ” IEEE Spectrum, Nov. 7WTA workshop on Software Engineering, Nov. 17,1983. 1983, p. 92. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 81

Prototyping fications to a set of procedures that can be programmed for the computer. The software Once the project team has a working knowl- design team analyzes the application and seg- edge of the requirements of the software, a pre- ments the design problem into subproblems liminary design of the final system is made so or that can be programmed by in- that the team can obtain feedback from the modules dividuals or small groups. users. Thus, fine adjustments can be made at an early stage. The segmentation or modularization of the A software prototype can be as simple as a software design breaks the problem into man- ageable pieces to maximize the ease of pro- paper and pencil sketch that runs through the gramming and testing the segments, and to workings of the system, or as complex as a facilitate the interchange or replacement of the large and intricate computer simulation. The modules to simplify maintenance. Proper choice of prototype complexity is based on the segmentation is of particular importance be- nature of the application, the level of training cause it has been demonstrated that as the size and the sophistication of the end users, the of the programming project team grows, more size of the final software system, and the 72 and more time is spent communicating among criticality of the application. team members and coordinating communica- Research and development of the prototyp- tion among program segments and less in ac- ing phase involves the design, development tually writing code. Thus segmentation has a and testing of methods and tools to make pro- crucial impact on the overall productivity of totyping more understandable to systems de- the software development team. signers and users, and more automated so that The design phase of software production it is a logical and accepted part of software de- relys heavily on the experience and intuition velopment. Research efforts in universities of the design team and is considered to be best and industry are focusing on languages that learned by apprenticeship. Understanding of make the quick production of prototypes pos- fundamental mathematical principles is also sible, and on the design of man-machine inter- 73 important. faces to enhance and verify the effectiveness of communication between the system and its Research on the design phase of software users. engineering is embedded within the larger framework of R&Don “programming environ- The major issue in prototyping involves the merits. ” The objective of software engineer- acceptance of this activity and the awareness ing is to provide as much support as possible among system designers of its importance. for the creative designer. Facilities should be Prototyping is commonly ignored in current made available to ease the rendering of ab- software development practice because of its stract principles and problems into workable, expense and difficulty and the lack of auto- testable, reliable, and cost-effective solutions. mated support. But experts contend that, in Software design can be made more productive the long run, prototyping can save time and in much the same way that design of complex effort because changes to requirements speci- integrated circuits has improved: by the ap- fications become more costly as each phase of plication of computer-aided-design (CAD) tools software development proceeds. based on sophisticated graphics workstations. A number of companies are making significant Design productivity gains using CAD tools for soft- The design stage of software development ware design. involves the reduction of the software speci-

72 For example, air traffic control software requires extensive .— simulation, whereas an inventory system might need only a simp- “Data Communications, op. cit. p. 81, and AZosa”c, op. cit. le sketch and some text to check against user needs. pp. 4-5. 82 ● Information Technology R&D: Critical Trends and Issues

In the longer term, software design produc- grammers control the operation of computer tivity could improve significantly through the systems. expanded development and use of software Many tools are already available to pro- Libraries, or systems to store and access stand- grammers to help produce efficient and reli- ard, reusable software modules that perform able code. The most familiar of these are the functions required in many programs. As now high-level languages such as FORTRAN, practiced, software design generally segments COBOL, and Basic that were designed to aid each program problem in a unique fashion, both to accommodate the unique features of programmers in coding certain broad types of new applications and to make the most effi- applications. Compilers translate high-level cient use of hardware. Program modules are language into machine executable binary dig- usually not designed for reuse, and there is no ital code. Editors and debuggers facilitate the strong incentive to do so because facilities are entry, testing, and correction of code lines. not provided for storing separate modules or Research and development in the improve- for classifying and finding appropriate segment of computer code writing continues. In- ments for new programs, and because the proj- novations over the last decade include struc- ect development team is generally unable to tured programming which greatly simplifies spread the cost of creating reusable modules the tasks of testing and deciphering code dur- across a large organization. The result is that ing maintenance, and new high-level languages designers and programmers are constantly designed to be easy to learn and use or tech- “reinventing the wheel” in software devel- nically sophisticated (no languages yet intro- opment. duced appear to be both simple and highly flexible). Researchers in computer and information science have identified basic techniques and A recent trend, that has proceeded hand in have designed tools to provide generic pro- hand with the introduction of personal com- gramming modules, data structures and algo- puters, has been the development of coding rithms, and to classify and organize them for systems or languages that can give end users storage and access. But the software commu- of computer systems more control—and re- nity has been slow in adopting the facilities move the need for the help of a professional necessary to make software libraries work be- programmer for each application. An example cause of high initial cost, poor management is Visicalc. understanding of such systems, and reluctance Large companies, including manufacturers among programmers to modify long-established of mainframe computers, are interested in design practices. decentralizing the control of computer programming and use by providing software sys- coding tems that are responsive to the needs of non- Coding is the stage of software production computer professionals. A proposed solution that makes a procedural description of the sys- to the software bottleneck dilemma is to farm tem executable by computer. More emphasis out some of the work now done by professional has traditionally been placed on this activity programmers to end users. IBM, for example, than on any other, even though it is estimated asserts that up to one-half of computer ap- to comprise only 15 to 20 percent of the total plications development can and should be han- software development effort.74 The reason for dled by the end-users with personal com- the traditional emphasis on coding is that it puters. 75 represents the basic mechanism by which pro-

— .————. .— 75AppZication Development in Practice, Tech Tran User Sur- “Data Communications, op. cit. p. 58, and E. B. Altman, vey, Xephon Technology Transfer, Ltd., 1983, pp. 53-54. Some “Software Engineering, ” Mim”-Micro Systems, December 1982, experts consider this solution to be dubious, and believe that p.184. it may in fact merely spread software problems. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 83

The proliferation of personal computers in software engineering as the complexity of the business world has some other possible im- computer-based systems increases and as plications, both for the process of software de- more of society’s functions are entrusted to velopment and for employment patterns of computerized systems. professional software development personnel. End-users programming on personal com- Testing and Validation puters could allow central data processing de- An important activity in software produc- partments to concentrate their efforts on the tion and maintenance is the testing of the larger, more complex, and more critical ap- products of the various phases of the software plications. A reduction in the backlog of small lifecycle. Until recently software development projects could also encourage the development budgets included sufficient funds for adequate of new large-scale applications that heretofor testing in only a few highly critical projects. languished because of long waiting times. It is conceivable that the demand for profes- It is impossible, no matter how extensive sional programmers will slacken in such indus- the testing activities, to guarantee large soft- tries as financial services which have tradi- ware systems to be error free. A risk assess- tionally employed a large part of the data ment based on the cost associated with soft- processing workforce in coding and maintain- ware defects or breakdown in a given application ing applications programs. This could result must be made to determine the level of effort from trends toward: 1) small-scale applications in testing that is justified. A rigorous testing development by end-users with personal com- regime can double the cost of software devel- puters using commercially available software opment. packages, and 2) increased use in data proc- There are many concepts, tools and tech- essing shops of standardized and automated niques available and undergoing research or software development tools, also commercially development for testing and validating soft- obtained, for large-scale projects. But increas- ware. They range from traditional methods ing demand for innovative, off-the-shelf ap- such as manual “desk checking” or “walk plications programs and software develop- throughs, ” to statistical methods to determine ment tools suggests that demand for the most the probability of errors in a set of code, to talented software designers will increase as some as yet highly experimental automated competition intensifies in these expanding program provers that verify that properly markets. structured programs perform as specified. The There is a secondary impact of the introduc- broad applicability of program provers will de tion of personal computers on software devel- penal on some fundamental breakthroughs in opment. Upper management is becoming fa- program theory.76 Until the requirements miliar with the problems and possibilities of specifications and design stages of software software and computer use through experience development are better understood and sup- with personal computers. Management aware ported, manual techniques will be the key ness of and commitment to change in the soft- testing methods available. Although some ware development process may be a key fac- automated tools are in use, few exist as in- tor in introducing and accepting software engi- tegrated packages. They must generally be neering concepts and innovations. pieced together by individual projects; and Large and complex applications will con- — —. “Some experts contend that the pursuit of formal verifica- tinue to depend on the efforts of professional tion methods such as program provers may prove fruitless be- software designers and programmers. Thus, cause of the increasing complexity of software and the rapid the introduction of automated tools and tech- changes in software development practices. See R. A. DeMillo, R. J. Lipton and A. J. Perlis, “Social Processes and Proofs of niques such as CAD are imperative. Pressure Theorems and Programs, ” Communications of the ACM, May will continue for more productive and reliable 1979, pp. 271-280. 84 ● /formation Technology R&D: Critical Trends and Issues

they themselves can become sources of high problems of making changes and testing. A cost and errors.77 change in one part of a poorly designed program will require changes in other parts of the Maintenance program, multiplying the cost and time for maintenance. Novel coding of standard func- This is the phase of the software engineer- tions increases the difficulty of predicting the ing lifecycle that is concerned with the revi- effects of changes. Large and cumbersome, sion of software that is in use. Software main- poorly organized and incomplete documenta- tenance consists of two kinds of activities: tion make it difficult for maintenance pro- correction of errors that went undetected in grammers to know what the system is sup- the course of software development and test- 79 posed to do. ing but that crop up in the use of the software; and changes in programs resulting from al- Junior programmers with little experience tered or additional requirements specifica- are generally assigned to maintenance tasks, tions. It has been estimated that 50 to 90 thus increasing the delays in bringing software percent of current software costs involve main- back into use, and also perpetuating the tenance. Some software systems may cost 25 knowledge and acceptance of poor program- times as much to maintain as to develop.78 Be- ming styles and habits. cause of poorly designed, badly written, and inadequately documented code, much of the The Scientific Basis for Software Engineering ammers’ time is spent try- maintenance progr There are some well understood scientific ing to understand programs rather than chang- methods that can contribute significantly to ing them. the improvement of software engineering tools A solution to the maintenance problem and practices. The application of scientifically adopted in many cases is to discard old code. designed, experimentally based statistical This is because experience indicates that the methods for the evaluation of the effectiveness “patches” made in programs when they are of software tools may produce the concepts fixed or revised often increase the complexity and methods for at least an order of magni- and difficulty of maintaining code and make tude increase in software productivity. it inefficient to run. Therefore an average of In particular, in the critical area of human 10 months worth of programming develop- interaction with computers, a body of research ment effort is thrown away each year by main- methods have been developed in universities frame based data processing organizations. that, if widely and systematically applied, Unfortunately, this old code is often replaced could identify, quantitatively assess, and pro- by poorly designed new code which again is duce improvements in programs for making difficult and expensive to maintain. computers more responsive and easier for peo- Poor requirements specification and design ple to program and use. Research in behavioral practices, idiosyncratic coding styles, and psychology and psychophysics is being ap- poorly organized documentation increase plied to the design of computer systems and maintenance costs. Because requirements software engineering tools on a limited basis were vaguely specified or have changed, the in a number of university programs. design proceeded on assumptions that were false or later rendered invalid. Bad design The limited scale of commitment to applied assumptions lead to poor segmentation and research in software engineering maybe a ma- structuring of programs which exacerbate the jor reason for lack of dissemination and use of these well understood techniques. Acad-

77w. R. Adrion, hf. A. Branstad, and J. C. Cherniavsky, “Validation, Verification, and Testing for Computer Software,” 7%3TA advisors related that program documentation captures, ACM Computing Survey, June 1982, p. 183. on average, only 60 percent of the information needed to fully 7sOlsen, op. cit. p. 58. understand a program. Ch. 3—Selected Case Studies in Information Technology Research and Development • 85 emics are generally involved in small-scale, International Efforts in Software fundamental research oriented projects that Engineering R&D cannot test large size programs or a large number of possible software development aids. Japan Some researchers suggest that applied re- The Japanese have had a software engineer- search on a massive scale following the model ing effort since 1970.81 The Information Tech- of medical clinical trials is necessary to dis- nology Promotion Association (IPA), a consor- seminate and make use of the knowledge em- tium of private companies and the govern- bodied in available techniques. ment, together with long-term credit banks, Industry participation is an important in- provided $25 million in 1983 for software R&D gredient in increasing the scope of experimen- and technology transfer. Current projects in- tation because of its large and diverse software clude: purchase or development of software efforts. As a user of innovative tools and sys- packages for rent to users; a Software Main- tems, industry must also be involved in the tenance Technology Project, to be completed research so that essential feedback is provided in 1985, which is developing work stations for to tool designers and evaluators. Inadequate program analysis, documentation, testing, and appreciation of the gains that can be achieved production management; and a cooperative ef- in software productivity and a concentration fort among users, suppliers, and researchers on short-term results have made industry to develop computer-aided design, computer- management reluctant, until recently, to make aided instruction, and software engineering in- a commitment to large scale or long-term ex- novations. perimental software development studies. A noteworthy aspect of Japanese software As mentioned earlier, some large companies engineering efforts is the establishment of are becoming interested in applied software re- software factories. These are consortia, staffed search and are hiring computer and human by people from participating companies, which factors scientists to develop research pro- are creating integrated environments for soft- grams. However, competitive and proprietary ware production, testing, and maintenance. considerations and the uncertainty of software Impressive programmer productivity gains protection by copyright or patent appear to are reported to have resulted from the aver- have inhibited an integrated applied research age reuse of 30 percent of computer code in effort among companies. Antitrust laws may new applications.82 Japanese software fac- also have a chilling effect on industry based tories appear to owe their success in raising joint applied research.80 productivity levels, in part, to the fact that only restricted kinds of applications with fixed An alternative approach is for the Federal and well understood requirements are at- Government to use its software development tempted. It is uncertain how effective these efforts as a test-bed for software engineering software factories would be on more difficult research. Some moves in this direction are be- applications. Nevertheless, these concerted, ing discussed within the defense community, cooperative efforts provide a test-bed for new and the National Bureau of Standards has a software engineering concepts and tools in a research effort in software engineering in its well capitalized development and production Institute for Computer Science and Tech- environment. As well, they train people in the nology. use of advanced software development techniques, serve as a dissemination mechanism SOEleven major defense contractor companies recently announced plans to form a consortium for the pooling of efforts in software engineering R&D. The Justice Department has 81 Summary of Major Projects in Japan for R&D of Informa- reportedly given the go-ahead for the development of a formal tion Processing !fechnology, prepared by Arthur D. Little (Ja- business plan patterned after the Microelectronics and Com- pan), Inc., under contract to OTA, July 1983. puter Technologies Corp. (MCC). M. Schrage, “Software Re- ‘*Raymond Yeh at the OTA workshop on Software Engineer- search Group Set, ” Washington Post, Oct. 10, 1984, pp. C1-C2. ing, Nov. 17, 1983. 86 • Information Technology R&D: Critical Trends and Issues when those people return to their companies, 2. Integration of tools and methods for the and raise management awareness of the im- improvement of hardware and software portance and efficacy of software engineering. development which will be focused in an Thus software factories have potentially sig- Information Systems Factory. nificant implications for future Japanese soft- 3. Innovation software engineering through ware development efforts on a broad scale, re- increased levels of R&D aimed at devel- gardless of their impact on individual software oping new tools and evaluating their ef- systems. fectiveness. The Information Systems Factory, to be es- Britain tablished by 1989, is based on the premise that As a part of the Alvey Programme for R&D information systems’ functional requirements in information technologies (see ch. 7), Brit- may be written independently of whether a ain plans to spend approximately $100 million given function is to be implemented in hard- over 5 years (1984-89) on software engineer- ware or software. Therefore, it is reasoned, the ing R&D, which is approximately 20 percent trend should be toward the design of modules of the total Alvey Programme effort.83 Bri- whose uses are known in relation to other mod- tain’s traditional strength in software will be ules, such that the decision of whether to im- built upon, expanded and modernized to min- plement a function in hardware or software imize dependence on software imports.84 An may be made on the basis of economic, time- initial focus of the Alvey effort will be the scale and other cost-benefit criteria. The In- measurement of the cost effectiveness of soft- formation Systems Factory concept will evolve ware engineering through analysis of software into the 1990s, by which time advances in fun- imports and exports. Also, the program will damental research are expected to make pos- track the capitalization of software develop- sible a truly intelligent software engineering ment and analyze the relationship of capital production environment with fully automated intensity to programmer productivity. Efforts and integrated tools and a large library of will be made to establish formal output meas- standard function modules.87 ures for software cost and quality to help The British consider the understanding of evaluate new tools and methods and to estab- 85 software engineering concepts to be too prim- lish a system for software warranties. itive at this point to make a final determina- The software engineering effort will consist tion of a single direction to pursue to produce of three main thrusts:86 a fully integrated software engineering environment. Like those in the United States, cur- 1. Exploitation of existing tools and meth- rent British research projects are too small in ods, the transfer of technology from uni- scale to adequately test software in diverse ap- versities to industry, and increased in- plications to determine the efficacy of new vestment in software production capital tools and methods. The Alvey Programme will and in training. fund several approaches for full-scale development, both to select useful techniques and to —.. .. —-. .— build technology transfer bridges between re- 83 A fiO/@UIMIl e for Advanced Information Technology: The search, development, and production. In this Report of The Alvey Co-”ttee, Department of Industry, Her Majesty’s Stationery Office, p. 47. context, the software engineering effort is ex- 84A]vey Softwwe En@”nMring—A Strategy Overview, Pre- pected to contribute to, and to benefit from, pared by Alvey Software Engineering, Director D. E. Talbot, the parallel Alvey Programme efforts in com- Department of Trade and Industry, p. 1. as~ftwm &h”abih”ty and Metn”cs ~gramm e: Ovem”ew, puter-aided design for VLSI circuits and prf+ 88 pared for the Software Engineering Directorate by the Center knowledge-based systems. for Software Reliability, Alvey Directorate, Department of Trade and Industry, April 1984, p. 4. ——..————— ‘Alvey Software En@”neen”ng-A Strategy Overview, op. cit., “Ibid., pp. 5-7. p. 3. *aIbid., pp. 7-8. Ch. 3—Selected Case Studies in Information Technology Research and Development Ž 87

France Telecommunications (CNET). Together, these laboratories employ approximately 100 re- Like the British, the French have been noted 89 searchers in software engineering and related 90 in the past for the quality of their software. studies. In addition, 11 other French Gov- The major French R&D centers for software ernment and industry labs are pursuing R&D engineering are run by L’Institute National in software engineering.91 de Recherche en Informatique et en Automati- que (INRIA) and the Centre National d’Etudes wRe~ewch and De\~e]opnlent in h’lectrom”cs uSA-France 1982/1983, French Telecommunications and Electronics Coun- “yThree significant computer languages, Algol, Prolog, and cil, March 1984. Ada, were originated in France. glI~~E Spectrum, November 1983, PP. 64-65.

Case Study 4: Artificial Intelligence

Findings several national efforts to push AI technology forward. ● Artificial intelligence (AI) research seeks to In the United States, AI research has been make computers perform in ways that dem- supported principally by the Department of onstrate human-like cognitive abilities: per- Defense. Plans for putting artificial intelli- ception and action in complex situations gence concepts to work in defense applica- and environments; interaction with humans tions have recently been announced. These in natural language; and common sense and applications are far beyond the capabilities the ability to learn from experience. of any present systems. ● The capabilities of artificial intelligence are Basic AI research is conducted at a rela- often subject to exaggeration. AI has gone tively small number of the nation’s top through several waves of optimism and dis- universities by a small number of research- appointment as new fundamental concepts ers. These researchers are under conflicting have been discovered by research, as new pressures from companies wishing to cap- computational techniques and equipment italize on the production of highly valuable have allowed these concepts to be tested, systems, from students demanding training and as the limits of these concepts have in AI, and from personal desire to pursue been realized when applied to real-world or their own research interests. large-scale problems. Only a few universities have the facilities ● In the last several years, the first real com- and equipment to compete with industrial mercial products of about 25 years of AI re- AI labs. Therefore some highly motivated search have become available. These sys- researchers are drawn out of academia tems are of three types: where they would be available to perform — expert systems that aid human experts needed fundamental research and to train in analyzing complex situations and in future generations of researchers. making decisions, — natural language processing programs Introduction that serve to make the interaction of humans and computers more natural, and The goal of artificial intelligence (AI) re- — image or vision processing systems that search is to create systems that demonstrate can make robots and other automated some of the following characteristics: the processes more flexible in operation. ability to assimilate unstructured information ● Presently available AI technology is of and to act independently in complex situa- value in only a limited range of applications. tions; the capability of natural language in- But the promise and potential have led to teraction (e.g., in English) with humans; com-

38-802 0 - 85 - 7 88 • Information Technology R&D: Critical Trends and Issues

mon sense, and the ability to learn from facilities are needed that present information experience. 92 The possibility that machines in forms that are tailored to the unique needs may be made capable of such activities has of individuals, or that can control operations burst into public consciousness as a result of in complex and perhaps unpredictable circum- the increasing power of microelectronics-based stances. computers and their application in many new The complexity of the environment in which environments, the establishment of national the system is to operate is perhaps the most efforts with the aim of creating intelligent important dimension to be considered in the computer systems, and the sudden emergence design of AI computer systems. As the num- of some commercial products that embody ber of environmental variables increases, and characteristics of artificial intelligence. the number of possible responses multiplies, Some experts contend that the ultimate aim automated systems must exhibit increasing of computer science is to produce intelligent degrees of intelligence. Currently, computer machines. Just as the industrial revolution systems are being designed for environments was driven by the urge to amplify human and in which they will be required to resist confu- animal muscle power with more efficient me- sion and error from ambiguous or contradic- chanical devices, the computer revolution is tory input signals, automatically coordinate driven by the desire to amplify human intel- diverse activities under changing conditions, lectual power with electronic information and provide logical and understandable ex- machines. 93 planations of their actions to operators. An expanding user population and the in- The highest degree of machine intelligence creasing diversity and complexity of the ap- might be the ability to automatically adapt to plications of current and planned information conditions that were not specifically antici- systems, are compelling designers to find ways pated by the designers. Such ability would re- to build intelligence into computers. Com- quire a machine to have “common sense, ” puters are increasingly being required to ex- broad world knowledge from which to infer change information among people with diverse reasonable courses of action, and also the interests, backgrounds, and levels of computer ability to assimilate new knowledge by learn- sophistication, so they are being designed with ing, through self-organization of experience. software that permits people to interact with Machines with this degree of intelligence are them in forms more closely resembling natu- speculative, and attempts to push present AI ral language. New varieties of sensors and in- concepts toward such capabilities illustrate creasingly diverse sources of information are both the difficulty of the problems, and the in- providing input, thus computer systems must adequacy of present concepts as a foundation be more flexible to make use of information for machine intelligence on a scale approaching with varying levels of importance and con- the cognitive abilities of humans.94 fidence attached to it. Processing and output There have been some notable recent successes in applying AI concepts to real-world problems. The introduction of a number of AI 9~D. w~tz, “Artifici~ Intelligence: An Assessment of The State-of-theArt and Recommendation for Future Directions, ” systems into commercial use has elicited a de-

AI Magazine, fall 1983, pp. 55-66. - mand in industry for AI software that can 93 For exmple, the Japanese 5th Generation project is moti make computer systems more responsive and vated by a belief that intelligent computer systems will be one of the cornerstones of a healthy economy and society in the later productive. Three types of artificial intelli- years of the 20th century and beyond. See T. Moto-oka, “Chal- gence products in particular are generating in- lenge of Knowledge Information Processing Systems (Prelimi- nary Report on Fifth Generation Computer Systems), ” pp. 3- “Some tasks requiring “intelligence” are already performed 89, and H. Karatsu, “What is Required of The 5th Generation better by machines than humans, f~r example long division; Computer–Social Needs and its Impact,” pp. 93-106 in Fifth other tasks may never be performed by machines. See Waltz, Generation Computer Systems, T. Moto-oka (cd.), JIPDEC- p. 55. See also M. Michael Waldrop, “The Necessity of Knowl- North Holland, 1982, 287 pages. edge, ” Science, Mar. 23, 1984, pp. 1279-1282. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 89 tense interest among industrial companies. The sudden discovery of promise in artifi- These are: natural language data base inter- cial intelligence by society at large raises sev- faces that allow users to access information eral questions. Of immediate concern are the stored in a computer data-bank with English questions of whether AI, as it is now under- language queries; expert systems that aid peo- stood, can meet industry and military expec- ple in complex tasks that require experience tations for performance, or whether there will and detailed knowledge to perform; and robot be expensive failures in applying this technol- vision systems that promise increased flex- ogy, and by implication a casting of doubt on ibility in manufacturing and other robot ap- the entire AI enterprise, because present con- plications. Industry sees these systems as cepts are immature and limited; and whether enhancements or potential replacements for the present educational structure, or some rea- rare and/or expensive human skill. sonable extension of it, can meet the growing demand for trained AI R&D talent. For the Similarly, the Department of Defense is in- longer term, there is a question whether the terested in artificial intelligence to enhance the progress of AI research can be sustained in power, efficiency and reliability of increasingly the face of feverish commercial development, automated computer-based weapons and com- or whether the limited number of researchers mand and control systems. DOD’s new “Stra- will be drawn away from the study of funda- tegic Computing” program seeks to push the mental and difficult problems and the teaching frontiers of AI science and technology for use of new AI people. There are also profound in battlefield management systems, autonomous (unmanned) tanks and submarines, and questions concerning how artificial intelligence technology might be used, and how these uses automated expert assistants for airplane might affect society in general, especially in pilots. terms of employment and the potential the There is concern, among both AI research- technology offers to enhance or compete with ers and observers of information technology human intellect. R&D, that enthusiasm about recent successes and fear of foreign competition are encourag- Artificial Intelligence R&D Environments ing irrational expectations for artificial intelligence. These forces are drawing limited AI The Roots of AI R&D resources toward risky, short-term de- The idea of automated intelligence has in- velopment work, and away from some of the trigued mankind for more than a century.% In tough questions that AI research should and the 1950s, two important figures in the early could answer. As Arno Penzias, Nobel laureate history of electronic computers, John von Neu- and Vice President for Research at Bell Lab- mann and Alan Turing, both expressed confi- oratories has written about artificial intelli- dence that within a short time computing gence R&D: machines would equal or surpass human intellectual capabilities.97 A crash effort in any one area would inevi- tably pull talented people away from other ‘E. Charniak, “Artificial Intelligence-An Introduction, ” pre sentation at the 1983 Conference of The American Association areas . . . whose payoffs to society might be for Artificial Intelligence, Washington, DC. One can conceive even greater in the long run. Our challenge is of artificial intelligence as the culmination of the mecha”stic to improve computers and extend their exper- paradigm of scientific thought that has been dominant since tise into all the areas where people will need the time of Newton. See Science and Change, 1500 to 1700 by Hugh Kearney, McGraw Hill, New York, 1971, for a discus- them and want them. We can best do that sion of the emergence of the mechanistic paradigm. Some with a balanced program of research and de- historians suggest that Man has always sought to represent velopment. . . using knowledge to help create and embody life and intellect in his artistic and useful creations. a society that provides a meaningful life for See J. David Boulton, Turing’s Man: Western Culture in the all the people in it.95 Computer Age, (Chapel Hill) University of North Carolina Press, 1984, ‘6A. A. Penzias, “Let’s Not Outsmart Ourselves in Thinking- “See J. von Neumann, “The General and Logical Theory of Computer Rush, ” Wall Street Journal, Sept. 13, 1983. Automata, ” pp. 99 and 109, and A. M. Turing, “Computing 90 • Information Technology R&D: Critical Trends and Issues

Improvements in the understanding of com- Figure 11 .—Artificial Intelligence putation and knowledge organization coupled with advances in computer speed and memory capacity made possible through improvements in electronics, have encouraged inter- mittent waves of optimism that super electronic “brains” were just a few years in the future. Time and time again, these waves of optimism have ebbed in the face of a common problem. As the systems that researchers contemplate and build are required to deal with more information, and as the situations in which they are required to operate become more realistic (more perceptually complex and unpredictable), the fundamental computational principles and methods they have to work with become 98 less efficient and impossibly sl0W. The term “artificial intelligence” was coined by John McCarthy in a grant application in 1956 to describe the subject of a conference that he was organizing.99 This meeting, held at Dartmouth College, brought together researchers in different fields whose common concern was the study of human and machine cognition. The conference established AI as a SOURCE: Randy Davis and Chuck Rich. distinct discipline, and also served to define the major AI research goals: 1) to design ma- As an esoteric study melding some aspects chines that think, and 2) to understand and of computer science, linguistics, psychology, model the thought processes of humans. Thus and philosophy, artificial intelligence research AI research is grounded in computer science has grown up, for the most part, in an exclu- and electrical engineering and also has its sive set of universities. Some industry-based roots in cognitive psychology, linguistics, and research efforts at AT&T Bell Laboratories, philosophy. There is an appreciation among Xerox Palo Alto Research Center, Bolt Beranek many researchers that AI is an interdiscipli- and Newman, and SRI International span nary study born of the interaction of these two more than a decade, but most of the work has goals, and that the intersection of the sepa- been concentrated at Massachusetts Institute rate traditions constitute what has come to be of Technology, Stanford University, and Car- a core of concerns that distinguish artificial negie-Mellon University, the leading research intelligence from other fields of science and and academic centers offering a full range of engineering. loo (See fig. 11.) AI subdiscipline. The concentration of effort and expertise in these top tier universities has Machines and Intelligence,” p. 245, in Perspectives on the Com- tended to restrict entry into the field, and to puter Revolution, Z. Pylyshyn, (cd.), Prentice-Hall, Englewood foster a very tightly knit community of re- Cliffs, NJ, 1970. ‘aJ. T. Schwartz, “Research in Computer Science: Influences, searchers. Accomplishments, Goals,” Report of The Information Z’echnology Workshop, R. Cotellessa, Chairman, National Science Foun- Some years ago, AI research and training dation, Oct. 10, 1983, p. 18. programs began to spread to a wider set of ~gcharni~, op. Cit., p. 5- universities, and this expansion of academic ‘OONils Nilsson, “Artificial Intelligence Prepares for 2001, ” Presidential Address given at the Amual Meeting of the Amer- programs continues. These schools have pro- ican Association for Artificial Intelligence, Aug. 11, 1983. grams built around one or two graduates of Ch. 3—Selected Case Studies in Information Technology Research and Development . 91

the established programs, and thus they of- and the establishment of new academic pro- fer only restricted AI curricula. Also, a num- grams for the training of applied AI scientists ber of new industry-based R&D programs are and engineers. being formed. The newer industrial programs The growing commercial and military impor- in AI R&D have also tended to center on the tance of AI is forcing researchers to make dif- ideas of one or two researchers from the top ficult choices in how they spend their time. schools. The lifespan of these industrial efforts They must balance a number of commitments: may be limited if they are unable to draw in to the study of research issues that have long- new talent from the small available pool and term significance; to the teaching and super- are unable to obtain infusions of new ideas. 101 vision of students;103 to consulting for indus- The difficulty of establishing and maintain- try and government; and to startup companies ing new AI R&D programs results from the in which they develop personal interests. To fact that there is only a small base of existing be sure, not all of these commitments are researchers. Second, the nature of AI work is mutually exclusive. Long-term research and such that an interdisciplinary team is required student supervision can reinforce one another. in most cases to produce programs of useful Consulting and entreprenuerialism can be size and complexity. This “critical mass” of complimentary. But the wearing of many hats workers knowledgeable in AI concepts exists by AI faculty may pose conflicts apart from only in a few universities and companies and the overcommitment of time. requires a number of years to pull together l02 First, there is a danger that students may before fruitful work can be expected. be judged, consciously or unconsciously, on OTA found that a major transformation is their contribution to the business interests of occurring in AI, resulting in two major direc- faculty, rather than on their progress in learn- tions for the R&D community. On one hand, ing the science of AI.104 Second, the progress fundamental research will advance theory. The of AI, as well as that of other sciences, is de- other direction is taking existing AI concepts pendent on the free exchange and public dis- and developing their applications. As in other semination of research results. Yet proprietary fields, the two directions reflect the divergence interests may inhibit that flow of informa- of interest between basic scientists and ap- tion.105 Third, many researchers face a conflict plied scientists or engineers (see fig. 12). How- between the objectives of the sponsors of re- ever, AI is distinct from other fields that have search and their own conception of the best undergone this transformation in at least four direction for the research. (See ch. 6 of this re- significant ways: 1) AI is a very young (scien- port on New Roles for Universities.) tifically immature) field, 2) there are few This latter conflict may be particularly acute trained AI practitioners, 3) the expectations among those in AI research because of the fact for the technology are high, and 4) the pace that the social implications of artificial intel- of transformation to an applied study is fast. The divergence of interests is expected to eventually result in the concentration of the IOWne OTA advisor said that at his university, one quarter of the students now entering computer science and electrical traditional “top-tier” university centers of AI engineering want to study AI. AI faculty across all schools are research on the fundamental scientific ques- obliged to supervise, on average, some ten graduate students; tions, an expansion of industrial R&D efforts in other fields the average ranges from less than one to about four. As well as teaching and advising, faculty must spend time to deal with the development of applications, writing proposals and seeking funding and equipment for graduate thesis work. ‘04 MIT has developed a policy that prohibits faculty from ‘“’ The OTA workshop on Artificial Intelligence, held Oct. 31, supervising students that work for companies in which the 1983. There is some controversy on this point, but it is gener faculty member has substantial interest. Patrick Winston at ally agreed that some “critical mass” of AX trained people, work- the OTA Workshop on Artificial Intelligence, October 31, 1984. ing in close proximity where they can exchange ideas, is a crit- 105This point was made by advisors at the OTA workshop on ical factor in performing advanced AI work. z Artificial Intelligence, and is discussed by Jordan J. Baruch “ Waltz, op. cit., pp. 64-65. in an editorial in Science, Apr. 6, 1984.

92 ● Information Technology R&D: Critical Trends and Issues

Figure 12.—Artificiai Intelligence Science and Engineering Research Topics

Motivations

I Requirements

The evolution of the field of artificial intelligence is producing new relationships and new distinctions within the Al science and engineering community.

SOURCE. Office of Technology Assessment. ligence are so profound. They include the Until recently, artificial intelligence research replacement of humans with machines in was supported by four Federal agencies, pri- skilled work, the increased reliance on auto- marily the Defense Advanced Research Proj- mated systems and the risk of system failure, ects Agency (DARPA) plus the National Sci- and affects on man’s perception of himself, and ence Foundation, the Air Force Office of his institutions. These conflicts raise ethical Scientific Research, the Office of Naval Re- questions such as: how much responsibility do search, and a handful of companies. Now, sig- the creators of powerful technologies have for nificant levels of work are beginning to be the social impacts of their creations? or, should funded in several corporations as the profit po- scientists consider the public interest in de- tential of AI grows. The Department of De- ciding what research should be done?l06 fense plans to expand AI research in universities and defense contractor companies. reader is referred to Relations of Science, Government and Industry: The Case of Recombinant DNA, ” by Charles Industry Efforts Weiner, ch. 4 in Science, Technology, and the Issues of The An increasing number of industrial firms Eighties: Outlook, A. H. and R. Thornton (eds.), Westview Press, Boulder CO, 1980. have begun AI R&D efforts in the past few Ch. 3—Selected Case Studies in Information Technology Research and Development • 93

years. Market research sources report that in- Government Funding dustry spent a total of $66 million to $75 million in 1983 on AI products.107 As many as 15 The Department of Defense is the lead agen- or more of the largest U.S. corporations are cy funding AI research. The Defense Advanced on the threshold of expanding the develop- Research Projects Agency (DARPA) estab- ment of expert systems.l08 The Microelectron- lished the three university “centers of ex- ics and Computer Technologies Corp. (MCC) cellence” at MIT, Carnegie-Mellon, and Stan- is devoting a fair amount of its early R&D ef- ford, and has awarded contracts since the early forts to AI related questions. 109 Three U.S. 1960s. The Office of Naval Research (ONR) companies, Xerox, Symbolics, and Lisp Ma- and the Air Force Office of Scientific Research chine, Inc., currently offer computer systems (AFOSR) have been funding AI work since the especially designed for AI program develop- 1970s. The Air Force in particular is interested ment. 110 Four other computer companies, Sper- in AI techniques for image understanding. ry, Apollo Computer, Data General, Hewlett- DARPA has proposed a significant expansion Packard and Digital Equipment, are also de- in its AI efforts (see “Strategic Computing” 111 below). ONR and AFOSR funding is expected veloping AI computers. Of most significance 113 is the number of startup companies develop- to remain stable in the near term. (See table ing AI products, particularly expert sys- 19.) tems.112 At least a dozen firms have been The National Science Foundation has awarded founded in the past few years and they are ob- grants for university and some industrial AI taining people, techniques, and seed ideas from projects since the late 1950s. In recent years the top universities and from nonuniversity the NSF AI research budget has remained sta- centers such as SRI International. ble, between $5 million to $6 million annually. (See table 20.) Two NSF directorates, Comput- er Science and Information Science and Tech- nology, provide the bulk of these funds, and a number of the awards (13 of 80 in 1983) are made jointly between these two divisions, or ““T. Manuel, and S. Evanczuk, “Commercial Products Begin with the NSF Office of Interdisciplinary Re- to Emerge From Decades of Research, ” Electrom”cs, Nov. 3, 1983, pp. 127-129. It should be noted that these numbers research, or with other programs within NSF. flect a very broad definition of AI with which many experts would disagree. “STRATEGIC COMPUTING” ‘“*J. Johnson, “Expert Systems: For You?, ” Datanation, February 1984, pp. 82, 84, 88. DARPA has embarked on a major effort to ‘“’See ch. 6, in this report. push the frontiers of computer technology and ‘‘“These so-called “Lisp Machines’ (because they use the AI programming language Lisp) represent an estimated market artificial intelligence for application in future value of $50 million. military systems. The “Strategic Computing” “’These machines are useful in many types of complex pro- program plans to spend $600 million over the gram development; nearly one-half of the sales of these machines, by at least one of the current vendors, are to users that next 5 years (see table 21), over and above past are not necessarily developing AI applications. J. M. Verity, levels of funding, and spending will accelerate “LISP Markets Grow, ” Datamdion, October 1983, pp. 92-94, to unspecified levels as prototype development 98,100. “*See Verity, op. cit. p. 94, and Kinnucan, P., “Computers gets underway in the late 1980s. The project That Think Like Experts, ” High Technology, January 1984, calls for the demonstration of significant mil- p. 42. One OTA advisor believes that venture capital will be a significant source of funding for AI R&D. At least $16 mill- I I ‘Personal commumcatmns from Dr. David Fox, Director of ion has thus far been invested in startup companies develop- !tIathematics and Information Science, Air Force Office of Sci- ing expert systems, J. W. Verity, “Endowing Computers With entific Research, and Paul Schneck, Head, Information Sciences Expertise, ” Venture, November 1983, p. 49. Division, Office of Naval Research. 94 ● /formation Technology R&D: Critical Trends and Issues

Table 19.—Major Government Sources of Artificial Intelligence R&D Fundinga

Estimates Low High National Science Foundation ...... 5.0 6.0 National Institutes of Health ...... 3.5 4.2 Office of Naval Research ...... 1.2 1 .4b Air Force Office of Scientific Research ...... 2.5 3.0 Defense Advanced Research Projects Agencyc ...... 15.0 20.0 Total ...... 27.2 34.6 asources indicate that these levels have obtained for the past several years, and with the exception Of DARPA, are eXrx?C@d to remain stable or increase modestly in the near term. bThiS does rro~ include $.goo,@o in robotics research, some of which iS in artificial irIt011i9encf3 CTlliS represents the base level Of C)ARPA funding before the initiation of the “strategic computing”Pr09rWI (began in fiscal year 1984). SOURCES: NSF numbers from Swnrnary of Awards publications of the Division of Computer Research, FY 1982-83, and the Division of Information Science and Technology, FY 1981-82 and 1983, National Science Foundation. NIH numbers from Susan Stimler, Director, Biomedical Research Technology Program, Division of Research Resources, National Institutes of Health. ONR numbers from Paul Schneck, Leader, Information Sciences Division, Office of Naval Research. AFOSR numbers from David Fox, Director, Division of Mathematics and Information Science, Air Force Office of Scientific Research, DARPA numbers from Ronald Ohlander, Program Manager, Defense Advanced Re- search Projects Agency.

Table 20.—NSF Grant Awards in Artificial Intelligence

Fiscal year 1982 Fiscal year 1983– Number–of Institutions ...... 41 38 Number of awards ...... Total awards granted ($OOO)a ...... $6,077 $5,150 Top 12 institutions: (Dollars in thousands) MIT ...... 946 240 Stanford University...... 485 587 Rutgers University ...... 245 442 University of Maryland ...... 351 317 University of Illinois ...... 478 141 University of Pennsylvania ...... 302 288 SRI International ...... 302 202 New York University...... 360 130 Yale University ...... 170 301 Carnegie-Mellon University ...... 185 227 University of Massachusetts ...... 87 229 University of Michigan...... 40 337 Total ...... 3,951 3,441 aThese numbers reflect the awards In each fiscal year and not I he amount Of funding dur!ng those Years sixteen Of the 67 awards made In 1982 were mult!-year grants (most were 2 year grants), 5 of the 80 awards made I n 1983 were multl.year qrants Therefore, total funding probably Increased somewhat between 1982 and 1983 SOURCE Office of Technology Assessment, and the National Science Foundation, Divisions of Information Science ancl Tech. nology and of Computer Research

Table 21.—Strategic Computing Cost Summary

Fiscal year 1984 1985 1986 1987 a 1988a (Dollars in millions) Total military applications ...... $ 6 $15 $27 TBD TBD Total technology base ...... 26 50 83 TBD TBD Total infrastructure ...... 16 27 36 TBD TBD Total program support ...... 2 3 4 TBD TBD Total ...... $50 $95 $150 TBD TBD %ut-year funding levels to be determined by program progress. SOURCE: Defense Advanced Research Projects Agency. Ch. 3—Selected Case Studies in Information Technology Research and Development • 95 itary capabilities within 10 years (see fig. 13).114 efforts on research and development of ad- The major areas of focus are: vanced computer architectures to meet the computational speed, and physical size and ● autonomous (unmanned) vehicles, weight requirements of mobile systems (see ● pilot’s associate systems to aid combat the case study of Advanced Computer Archi- pilots in coping with the complexity of tecture in this chapter), as well as artificial in- current and planned aircraft, and telligence R&D in vision, speech, natural lan- ● battle management systems to help com- 116 guage, and expert systems. manders make decisions under conditions of uncertainty. Some of the goals of the “Strategic Comput- ing” program, particularly the vision system To lay the groundwork for the development requirements for autonomous vehicles, have of such systems, DARPA plans to fund the been described as “extremely ambitious.’’’” building of Galium Arsenide integrated circuit Some describe this program as unprecedented pilot production lines for the manufacture of in the history of U.S. Government funding of low power and radiation resistant microelec- 117 science and technology. Unlike the Manhat- tronics. In addition, DARPA is focusing ——. ‘14Strate@”c Computing, “New-Generation Computing Tech- ‘]51bid., app. IV, p. 2. nology: A Strategic Plan for its Development and Application ‘]’ Ronald Ohlander, at the OTA workshop, Oct. 31, 1983; see to Critical Problems in Defense, ” Defense Advanced Research also Strate&”c Computing, pp. 22-23. Projects Agency, Oct. 28, 1983, p. vii. “’Mitch Marcus at the OTA workshop, Oct. 31, 1983.

Figure 13.—Strategic Computing Program Structure and Goals

Develop a broad base of Major machine intelligence technology goals to increase our national security and economic strength

Military Autonomous Pilot’s Battle applications systems associate management

● &

Inteligent functional Natural language Vision Expert systems capabilities Navigation Speech Planning and reasoning

High-speed signal-processing General purpose systems Technology Hardware/software base system Symbolic processors Multi-processor programming architecture and operating systems

Microelectronics Silicon and GaAs technology VLSI systems

1 Networks Research machines Rapid machine prototyping infrastructure I implementation systems and foundries Interoperability protocols Design tools

SOURCE’ DARPA. 96 • Information Technology R&D: Critical Trends and Issues tan Project or the Manned Moon Landing Mis- military requirements, additional pressures on sion, which were principally engineering prob- the limited AI manpower resources are ex- lems, the success of the DARPA program re- pected. 121 Although the expansion of training quires basic scientific breakthroughs, neither programs should eventually increase the sup- the timing nor nature of which can be pre- ply of manpower, the rate of that expansion dicted. is limited by the small existing faculty base.122 The current imbalance in the supply and de- The kind of work that is being done now [in AI research] that would support, for in- mand of AI manpower will continue, and likely stance, an autonomous vehicle system, is increase, and may intensify competition among primitive in relation to the problems that are commercial, military and academic AI R&D to be addressed [by the DARPA project]. agendas. Thus some applied research, such as Operation in a complex combat environment work toward “intelligent library systems” and that may have multiple targets with camou- computer assisted education, and fundamen- flage, different kinds of obstacles with vary- tal research that is unlikely to have high im- ing degrees of threat and impedance associated mediate commercial or military value, may be with them, and the integration of various neglected. 123 kinds of sensors, for example touch and vision, and the appropriate knowledge representation to deal with them, is an enormous Content and Conduct of Artificial problem that will be solved only by signif- Intelligence R&D icant strides in basic AI research, and not just development in narrow vehicle applica- Artificial intelligence as a field is a set of tions.118 somewhat loosely related R&D activities that range from the study and implementation of It is expected that the Strategic Computing machine sensing (e.g., vision and speech) and Program will approximately quadruple the an- pattern recognition algorithms and systems, nual Federal funding of R&Din AI and related ll9 to theoretical and practical work on automatic hardware. This major increase in R&D fund- problem solving, inferencing and reasoning ing will have significant and far reaching ef- strategies and programs. Of central concern fects on the artificial intelligence community is the concept of knowledge, a set of informa- and research environment. Undoubtedly, the tion (facts, procedures, patterns) that is in- increased flow of money will have some posi- tegrated and processable as a whole, and thus tive effects on AI research. In particular the constitutes a useful representation of a part availability of funds for modern equipment or domain of the world. AI deals with how should make university laboratory facilities knowledge is built up and used in computer- comparable with increasingly well equipped in- 120 based systems: how it is collected, stored, ac- dustry labs. The increased availability of cessed, manipulated, and transferred. AI R&D graduate student financial aid in the form of seeks methods of formalizing and represent- research assistantships should draw in and ing knowledge in consistent and unambiguous, help hold more qualified post-graduate train- yet flexible ways so that these tasks can be ees, thus expanding the potential faculty base, performed by machines. and relieving some of the pressure on current faculty to obtain funding for graduate re- “’Ronald Ohlander, at the OTA workshop, Oct. 31, 1983. search. However, as industry scales up to meet ‘*zIn the three AI subfields from which most commercial and military applications are expected, expert systems, natural language understanding, and vision, there are approximately 60 faculty at the top schools turning out some 30 Ph.D. graduates “’Marvin Denicoff, at the OTA workshop, Oct. 31, 1983. per year, half of whom take industry jobs. D. Waltz, “Artifi- 1‘gFrom about $30 million to approximately $120 million (aver- cial Intelligence: An Assessment of the State-of-the-Art and aging the $600 million program over its 5 year projection). One Recommendations for Future Directions,” prepared for the NSF DARPA source estimated that one-fourth of the program funds Information Technology workshop, Jan. 5-7, 1983; and a per- would be spent in universities. (Duane Adams at the OTA Work- sonal communication from Azriel Rosenfeld, Apr. 25, 1984. shop on Advanced Computer Architecture, held July 14, 1983.) ‘z’ David Waltz at the OTA workshop, Oct. 31, 1983. A pri- ““This may be an inducement for researchers to accept re- vate funding organization, the Systems Development Founda- search appointments. tion, is sponsoring research directed at “intelligent libraries. ” Ch. 3—Selected Case Studies in Information Technology Research and Development Ž 97

Although investigators in a number of fields Pattern Recognition perform work that contributes to and overlaps Of particular importance to the branches of with the subject matter of AI, a number of AI concerned with sense perception, such as areas are considered the core of AI research. machine vision and speech understanding, is the idea of matching input from the environ- Symbolic Computation ment with symbolic representations of pat- Since AI is concerned with the processing terns (e.g., visual objects or speech utterances) of knowledge, intelligent machines are re- stored in the system. Raw sensory input, in quired to manipulate symbols that represent the form of electrical signals from a television objects, concepts, and qualities, as well as camera or a microphone, must proceed through symbols that represent numbers or quantities, several levels of processing to produce pat- which are the focus of traditional computation. terns that are comparable with stored symbol- These qualitative symbols may be represented ic representations and recognizable as having within a computer as logical relationships or unique meaning. as lists of symbols with pointers linking vari- In general, these processing levels include: ous objects, concepts, and qualities. The ma- ● nipulation of symbols provides a mechanical Formation: The sound or light signals are means of achieving inference, in particular digitized and stored as a set of simple deductive inference, in which a problem solu- physical parameters such as frequency tion or a conclusion is tested against facts in and amplitude. ● a knowledge base to determine its truth or Analysis: Patterns of variations in the pa- validity. rameters, for example areas of light and dark in pictures or variations in pitch and A branch of AI research and development, intensity in an utterance, are detected and which forms a link with R&D in advanced measured to produce a detailed physical computer architectures, is the design of com- description of the input. puting machines that optimize efficient man- ● Interpretation: The patterns, now repre- ipulation of symbols. Some in computer ar- sented by sets of measurements, can be chitecture research maintain that this R&D either directly compared with templates, 24 effort has not received sufficient attention. 1 stored descriptions of entire patterns, or further analyzed to extract features, which are parts of patterns that are of particular value in defining and identify- ‘*’Sidney Fernbach at the OTA workshop on Advanced Com- puter Architecture, July 17, 1983. ing the patterns.

Knowledge Representation This area of research deals with ways of expressing knowledge in computable form and making and exploiting connections among the facts or propositions in a knowledge base. A m basic concept in knowledge representation is the propositional formalism, which is simply Photo credit, AT&T Be// Laboratories the structured way that facts are expressed Specialized microprocessors promise increasing levels so that ambiguities are avoided and process- of “intelligence” in computer systems ing operations are as efficient as possible. 98 . Information Technology R&D: Critical Trends and Issues

As well as a formal structure or syntax, knowledge domain. But instead of represent- knowledge representation requires a method ing concepts, scripts describe situations and of capturing meaning, or semantics, in a actions that are expected in those situations. knowledge base. Meaning in a qualitative A classic example is a restaurant script in sense is expressed by the relationships that which one expects people to order from a are stated to exist among concepts. One meth- waiter (or waitress), to eat, and to pay the bill. od of representing relationships is the Seman- Thus, invoking the restaraunt script would in- tic Network (see fig. 14). voke the entire set of expectations surrounding restaurants that have been programmed. A knowledge representation scheme that is similar to semantic networks has been devel- Another knowledge representation scheme, oped in which the attributes of concepts are one that was developed in the mid-1960s, is expressed and related. Frames provide slots the production system. This essentially con- to fill to structure knowledge about things, sists of a set of if-then rules that specify a pat- thus they represent expectations about what tern, for example, “if the temperature is less attributes members of a given frame will pos- than 65 degrees” and an action to be taken, sess. These “expectations” can be exploited “then turn the furnace on. ” The importance to imply procedures. For example, a computer of this scheme is its ability to express proce- can be programmed to inquire about the at- dural knowledge and to initiate operations tributes of a newly introduced object. depending on prevailing conditions. Its major Similarly, knowledge representations called weakness is that a fixed rule must be stated scripts can represent expectations about a for every condition that is to be encountered.

Figure 14.– A Semantic Network

1 J

mI I r Parakeets I I Chickens , I “

Cows

SOURCE: Randy Davis and Chuck Rich. Ch. 3—Selected Case Studies in Information Technology Research and Development ● 99

All of the knowledge representation schemes binatorics explosion”: as the size of a knowl- thus far adopted by AI researchers use consid- edge base increases, the number of possible erable computer memory if they contain a siz- paths to a sought-after piece of information able or complex domain of knowledge. Tradi- increases exponentially. AI interest was orig- tionally, researchers and builders of programs inally in developing general knowledge AI sys- have had to “shoe-horn” knowledge into com- tems, but the combinatorics explosion sets a puter systems with limited and expensive limit on how broad or useful such systems can memory capacity. The precipitous decline in be. More recent work has focused on specific the cost of computer memory is alleviating problem domains where a detailed represen- some traditional AI problems, making possi- tation of the world can be built and exploited, ble for the first time the production of cost- and where the search is controlled by knowl- effective systems with enough knowledge to edge about the problem domain, or heuristics. tackle “real-world” problems. Advances in computer architecture design, especially the Machine Learning ability to make customized VLSI processors dedicated to symbolic computation and capa- The study of heuristics (“rules of thumb” ble of parallel processing, should have a great or knowledge acquired from experience), how impact on the kinds of problems that may be they may be used to process knowledge, and addressed by knowledge-based systems. how they may be generated automatically by computer systems are major pursuits of AI Other difficulties inherent in the application research. Ideally, a computer could modify its of existing knowledge representation schemes repertoire of procedures as new facts are added will not necessarily be alleviated by lower to its knowledge base. Likewise, new proce- memory costs and advances in hardware ca- dures should suggest new relations among pability because they involve the limitations facts, but current knowledge representation of the schemes themselves. Existing knowl- and programming techniques are incapable of edge representations deal only with surface supporting changing knowledge and rule knowledge, which is either explicitly stated in bases. relationships or deductively inferred by the chaining of propositions. Deep knowledge, Commonsense Knowledge knowledge gained from inductive inference Human knowledge is based on a wide range (reasoning from facts to general principles), of experience acquired through sense organs. and commonsense knowledge that is routinely This experience is fed into a cognitive system learned by children through experience, are that can integrate that knowledge in order to not expressable using current knowledge rep- cope with a complex, dynamically changing resentation systems. Other problems involve world. Much of what people know involves the inflexibility of procedural knowledge and relationships that are obvious but that depend the difficulties of programming machines to on explanations beyond the comprehension of reason about unforseen occurances. most. Representing commonsense knowledge in computer systems is a major challenge to Given the large collection of facts, relations researchers, and is essential for certain kinds and patterns in a useful size knowledge base, of intelligent machine behavior, such as rea- even the fastest computers can bog down. soning about the chronology of events in the Therefore, an AI program must concentrate course of a disease or reasoning about the the search of its knowledge base on those por- movement and characteristics of robots. tions that are most likely to hold the facts, patterns, or solutions that are needed in a given Pragmatic Knowledge situation. The problems associated with the search of knowledge bases are ubiquitous in Current artificial intelligence systems em- AI. They result from what is termed the “com- ploy syntax (structure) and semantics (mean- 100 . Information Technology R&D: Critical Trends and Issues

ing) to represent knowledge about the domains they will be used is depicted in figure 15, which in which they operate. Systems of higher in- shows AI emerging from the laboratory into telligence, ones capable of choosing among a the home, the factory, and into schools. Three range of responses in complex situations, of types of AI-based systems in particular are ex- adapting to changing (and unanticipated) con- pected to experience rapid growth-vision sys- ditions, and of interacting with humans and tems, natural language understanding sys- perhaps other intelligent systems on a high tems, and expert systems. level of understanding, must employ ways of representing and reasoning about the needs COMPUTER VISION SYSTEMS of users and about knowledge itself. Truly in- Vision systems are being designed as input telligent machines would respond to people of subsystems to enhance the utility and flexi- different backgrounds and needs in ways that bility of computer-based automation. Vision are appropriately tailored to individuals. They systems are increasingly used in industry for would also possess an understanding of the quality control inspection, for product iden- uses and limits of knowledge-a mechanism tification, and for robot guidance and con- for judging the appropriateness of information trol.127 Machine vision will be a crucial func- to a given situation. For example, a robot tank tion in many future automated systems, would need to be able to distinguish between including aspects of the DARPA “Strategic trees it could run over and those that it would Computing” initiative. For example, an ad- have to avoid. vanced vision system will be an integral com- Current syntax- and semantics-based arti- ponent of autonomous vehicles. ficial intelligence systems employ knowledge Some sources estimate a current annual representations that are inadequate to support market for machine vision systems of $35 mil- pragmatic knowledge. Frame- and script- lion. This market is projected to double each based representations can, in a rudimentary year for the next 5 years, and should reach way, deal with expectations about situations, about $1 billion by the end of the decade.128 but those situations must be tightly circum- More than 250 companies, most of them in the scribed and clearly defined at the time of pro- United States, are designing and selling ma- gram design. Research has yet to identify chine vision systems.129 knowledge representation schemes that can support the processing of pragmatic knowl- Vision systems consist of cameras and com- edge in complex and dynamically changing puter processors to analyze and interpret the domains. information collected by the cameras. The cameras provide the processor with frames, Commercially Available Artificial Intelligence typically 60 per second, collected by scanning the camera’s visual field. A frame forms an im- Notwithstanding the fundamental limita- age consisting of a matrix of dots, or pixels tions of present artificial intelligence concepts, (picture elements), numbering in current sys- programs are being developed that are prov- tems 64 X 64 or 256 X 256 pixels. Higher ing useful in a widening array of applications. 125 resolution cameras are being developed, but Two market forecasts describe the commer- as the resolution (number of pixels per frame) cial potential of AI systems as exploding over increases, higher speed processors are needed the next decade. From the small 1983 sales to handle the increased information within rea- base of $66 million to $75 million, the market for AI products is expected to rise to $2.5 bil- 126 IZIThe reader is referred to the OTA report, cornpu~er~z~ lion to $8.5 billion by the early 1990s. The Manufacturing Automation: Employment, Education, and the changing mix of products and sectors in which Workplace, April 1984, pp. 89-92, for a discussion of current .—--—— —. --- industrial applications of machine vision systems. ‘2sInternational Resource Development, Inc., and DM Data, ‘*s’ ’Machines That Can See: Here Comes a New Generation, ” Inc. Business Week, Jan. 9, 1984, pp. 118-120. ‘z’ Manuel and Evanczuk, op. cit., pp. 127 and 129. 12gIbid., p. 118.

Ch. 3—Selected Case Studies in Information Technology Research and Development ● 101

Figure 15.— Markets for Artificial Intelligence terns allow a pixel to represent only black or white, though systems are being developed and are in use in which pixels may be one of as many as 64 shades of gray, and color vision systems will soon be available. Gray scale and color systems will be capable of higher acuity, but such systems require much higher processor speed and more sophisticated algorithms to interpret the increased information flow.

The processor analyses the image to extract patterns that may represent edges or textures that can be used to characterize and recognize objects in the visual field. Simpler systems have stored “templates,” or representations of entire objects, in memory to which the extracted patterns are compared. More sophisticated systems store “features,” or characteristic parts of objects. In these systems, sets of features are extracted from the image and 1993 total $8,536 million • are combined to represent constraints on the list of possible objects that may be present; recognition involves the identification of an object (or objects) that could satisfy those constraints. A major problem in machine recognition is inferring three-dimensional information from two-dimensional information in an image of the scene.131 One method that has been adopted to solve this problem is to project “light stripes” on objects in the scene; depth cues and contours can be inferred from distortions in the stripes. Current vision systems are primitive in comparison to human visual recognition capability. The human retina can perform at least 10 billion operations per second, and the brain is undoubtedly capable of much higher processing speeds.132 As important, human experience with the visual world provides a store of knowledge with which the eye and brain may

SOURCE International Resource Development Inc. make inferences and interpretations to resolve

130 sonable time limits. Each pixel represents naviagate at walking speed would require a computer capable an area of light or dark in the image. Most sys- of from 1010 instructions per second. T. and R. —— — “Computer Vision: The of Imperfect Inputs, ” Stanford University researcher has developed a robot cart Spectrum, November 1983, pp. 88-91. Computers of this that can navigate in a simple environment at a speed of 3 to speed are being developed, but may require facilities as large 5 meters per hour while analyzing one image of its surroundings as a room just to provide cooling for the hardware. for each meter of progress using a computer capable of proc- ‘3’ personal communication, May 25, 1984. essing a million instructions per second. For such a system to ‘32 and op. cit., p. 88. 102 ● Information Technology R&D: Critical Trends and Issues ambiguities caused by shadows and variations humans remains far off, awaiting break- in lighting, the orientation of objects, and the throughs in basic research, improved infor- presence and often overlap of multiple objects mation processing algorithms, and perhaps in a scene. Also, subtle variations and move- alternative computer architectures. How- ments can be detected and used by the human ever, the significant progress experienced in the last decade demonstrates the feasibility eye and brain to distinguish superficially simi- of dealing with natural language in restricted lar objects (e.g., the faces of identical twins). contexts, employing today’s computers. In order to make vision systems based on Automatic natural language understanding current processor architectures and algo- work began with the goal of producing sys- rithms work, the number of environmental tems that could translate one natural language variables that may cause ambiguities must be into another. Such systems are in use now, but reduced. For example, objects on an assembly are proving to be limited to rough translation, line may be arranged so that only one face is requiring a human translator to produce final presented to the camera or only one object at fluent text. a time is in the visual field. Lighting may be controlled so that perceived edges can be in- The goal of another class of natural lan- terpreted as facets of an object and will not guage systems is to understand documents: be the result of shadows or variations in the to produce an abstract of a piece, perhaps alert reflectance of surfaces of an object; or struc- people who might be interested, and answer tured lighting may be used to infer distance questions on its content. Such systems might and contour. be paired with document generators that could produce instruction manuals.134 Document If vision systems are to operate in less con- understanding and generation systems are trollable circumstances, such as out-of-doors still largely experimental, but they have the or in the home, much higher speed processors potential to augment human knowledge through and more sophisticated algorithms for analyz- the production of intelligent library systems, ing and interpreting visual scenes will be aiding people in searching and evaluating large needed. Custom designed, massively parallel bodies of information. VLSI-based processors dedicated to visual analysis are being looked to to provide more More immediate-term applications are in powerful hardware. Research in vision algor- using natural language systems as “front- ithms is proceeding to provide more powerful ends” or interfaces to computer systems. Nat- concepts and more sophisticated software. The ural language systems are in operation that parallel advance of hardware and software serve as interfaces to data bases, for example, solutions, and fundamental advances in AI re- to help store and find personnel or inventory search, will be required to produce systems ca- records. One can expect that model systems pable of the interpretation of complex and will soon be available that demonstrate the unstructured visual scenes. feasibility of controlling complex systems such as power generation facilities and weapons NATURAL LANGUAGE UNDERSTANDING systems with natural language interfaces. In SYSTEMS conjunction with expert systems, natural lan- These systems are intended to allow com- guage systems are being developed for communication with computers in English or other puterized medical advisor, trouble-shooting human languages, freeing users from the need and repair, mineral exploration, and invest- to learn a computer language. In the words of ment analysis applications. Future applica- one researcher:133 tions, requiring advanced work to produce, might include use in the creation of graphical The ultimate goal of creating machines displays and in computer-aided education.135 that can interact in a facile manner with — ——. . .-— “’Ibid., p.-57. “’Waltz, Al Magazine, op. cit., p. 56. “’Ibid. Ch. 3—Selected Case Studies in Information Technology Research and Development • 103

There is also current work in producing systems consist of a set of “if-then” rules speech understanding systems, which would which express the knowledge and experience replace keyboards with voice input devices. of an expert, and the actions one would take Primitive systems are available, but true when faced with a set of conditions in the do- speaker independent systems with large vo- main of his expertise. They also generally have cabularies that can handle continuous speech, a separate knowledge base which states facts as opposed to isolated words separated by about the domain, to which the program can pauses, are as yet unattainable with present refer to make inferences and deductions about technology. Advanced integrated circuit sig- given situations and conditions. nal processors should produce some progress 136 Expert system programs are developed from over the next decade. extensive interviews with recognized experts In part, the capabilities of natural language in the field of application. Such interviews systems have been limited by the cost of com- often reveal many unwritten (and even uncon- puter memory. Natural language systems scious) rules and criteria for judgment that the must have large vocabularies to be “natural’ expert applies in solving problems.138 The in- to users. This problem is diminishing because terviewer, called a knowledge engineer, then of the fall in computer memory prices. codes the rules and facts in a way that is efficiently processable by computer. Extensive Two other problems are replacing memory testing and validation must be conducted be- cost as an upper limit to system size and func- tionality. First is the “combinatorics explo- fore a system is considered complete enough to use in actual practice. sion” problem: as the size of the system vocabulary increases, the time required for This area of AI application is receiving con- presently available computer architectures to siderable industry and military interest be- process a sentence increases exponentially. cause expert systems may be capable of reliev- Novel machine architectures based on paral- ing some of the demand for high-priced experts lel processing promise some relief for this prob- in the fields in which they are applied. Ironi- lem. Second, as noted in the previous discus- cally, because they require man-years of effort sions of commonsense reasoning and pragmatic from scarce knowledge engineers to develop, representation, researchers are finding dif- expert systems are quite expensive. Therefore ficulties in designing systems with broad and they are currently being attempted commer- deep knowledge of the world, and with prag- cially only in applications that promise a par- matic understanding of situations. In lan- ticularly high payoff.139 guage, this kind of knowledge is crucial to the Research in this field has demonstrated that solution of ambiguities. expert systems can rival human expert judg- EXPERT SYSTEMS ment if the domain of application is properly These AI-based programs are designed to serve as consultants in decisionmaking and ists than to write programs that approach a child’s ability to problem solving tasks that require the applica- perceive, to understand language, or to make ‘commonsense’ tion of experience and judgment. ’37 Expert deductions, ” R. O. Duda, and E. H. Shortliffe, “Expert Sys- tems Research, ” Science, Apr. 15, 1983, pp. 261-268. —— —.. ————— 138This fleshing out of unwritten rules is a possible side ben- “’For discussions of the state-of-the-art in voice input com- efit of the development of an expert system; it can lead to pro- puter systems, the reader is referred to: A. Pollack, “Computers gress in the field of application itself. Conversely, some fear Mastering Speech Recognition,” New York ~“mes, Sept. 6, 1983, that this codification of knowledge in automated systems could pp. Cl and C7; R. D. Preuss, and D. J. Jurenko, “Digital Voice “ossify” and constrain progress in certain fields by foreclos- Processing,” Astronautics and Aeronautics, January 1983, pp. ing the possibility of the unexpected discovery of new solutions 44-46; and R. J. Godin, “Voice Input Output, ” Ehxtrom”cs, Apr. through serendipity. 21, 1983, pp. 126-143. ‘3gSome of the enthusiasm recently derponstrated by compa- “’Expert systems illustrate an interesting point about human nies for expert systems may be unfounded; OTA advisors were intelligence, “Paradoxically, it has proven much easier to wary of overstated claims for the usefulness and applicability emulate the problem-solving methods of some kinds of special- of current generation expert systems. 104 . Information Technology R&D: Critical Trends and Issues

chosen.140 This means that the application Current systems, because of the limitations must have certain characteristics for an expert of the concepts on which they are based, all system to be successfully applied. These in- suffer from several serious weaknesses:144 4 clude: 1 1 ● They are highly customized to specific ap- ● There must be recognized experts in the plications and are useless in other, even field; closely related fields; ● ● The task normally takes the expert a few Since the knowledge on which they are minutes to a few hours to accomplish; based is collected over a period of time, ● The task is primarily cognitive (as op- often from a number of experts, there can posed to manual); be inconsistencies in the programs that ● The skill is teachable to neophytes; are difficult to detect and repair; ● ● The task requires no commonsense. The systems are necessarily based on narrow sets of rules and facts, therefore their More generally, the application domain judgments and recommendations can be must be restricted enough to be expressable myopic and naive; in a finite set of rules, numbering less than ● Since all of the current systems contain 2,000 at present, so that available computers only surface knowledge from which to can solve problems in a reasonable amount of make inferences, if knowledge that is crit- time. Even so, most current expert systems ical for a given judgment is missing, their take longer to perform a task than a compe- performance is poor; tent human expert.142 ● Few current systems have natural lan- Table 22 shows a representative sample of guage interfaces, therefore they can be current expert systems and their uses. Several difficult for the uninitiated to use. companies are forming or expanding efforts Although many systems can provide explain developing expert systems, and venture nation for the chain of reasoning that led to capital interest in startup companies is re- 143 a given conclusion, these explanations are of- portedly brisk. ten unsatisfactory because they are not tailored to the needs or understanding of individ- ———. ual users. Neither do the explanations usually ‘401n fact, one OTA advisor said that the success of current applications depends more on the choice of domain than it does refer to underlying principles, such as physi- on how well the program is written. Patrick Winston at the OTA ology or geology, so human experts find them workshop, Oct. 31, 1983. unconvincing. 145 ‘“R. Davis, and C. Rich, Expert Systems: Fundamentals, —— Tutorial at the AAAI Conference, Aug. 22, 1983, p. 31. “’Waltz, op. cit. “’Waltz, AI Magazine, op. cit., p. 61. 14’Duda and Shortliffe, op. cit., p. 266. See also G. O. Barnett, “’see l%ctrom”cs, Nov. 3, 1983, pp.127-131, and Venture, No- “The Computer and Clinical Judgement, ” editorial in The New vember 1983, pp. 48-53. England Journal of Medicine, Aug. 19, 1982, pp. 493-494.

Table 22.—Representative Expert Systems

Type of Routine Expert system Domain evaluation use DENDRAL ...... Mass spectroscopy interpretation Case studies Yes MYCIN ...... Antimicrobial therapy Randomized trials No INTERNIST-1 ...... Internal medicine diagnosis Case studies No CASNET ...... Glaucoma assessment and therapy Case studies No PROSPECTOR ...... Geological exploration Case studies No RI ...... Computer layout and configuration Case studies Yes Digitalis Advisor ...... Digitalis dosing advice Randomized trials No PUFF ...... Pulmonary function test interpretation Randomized trials Yes Microprocessor EXPERT ...... Protein electrophoresis interpretation Case studies Yes HASP and SIAP ...... Ocean surveillance (signal processing) Case studies No SOURCE’ R. 0. Duda and E. H. Shortliffe, “Expert Systems Research,” Science, Apr. 15, 19s3. Ch. 3—Selected Case Studies in information Technology Research and Development ● 105

Research is underway to help alleviate some The Fifth Generation of these shortcomings, including the study of The Japanese plan to build, from a base of improved knowledge representation schemes research that is almost exclusively borrowed and inference methods. Novel computer ar- from the United States and Western Europe, chitectures should push back some of the significant artificial intelligence systems with limitations now imposed on system size by the stated purpose of enhancing productivity computer speed. The most immediate and in areas of the Japanese economy that thus probably far-reaching problem is the difficulty 149 far have proven resistant to automation. The and expense of building and testing expert total funding for the life of the project is not systems, including the acquiring and vali- publicly stated, but some sources estimate dating of rules and facts. Applied research that up to $1 billion to $1.5 billion may be work is progressing in the development of pro- 150 spent over the next 8 to 10 years. gramming aids (tools, languages, environments) to lower the time and cost of building The Fifth Generation Program is centrally expert systems. Some current basic research managed by the Institute for New Generation is concerned with the automatic acquisition Computer Technology (ICOT), formed by the and structuring of knowledge. Incremental ad- Ministry of International Trade and Industry vances can be expected over the next decade, (MITI) in 1982. ’61 The program will be funded, but fundamental breakthroughs will be re- for the most part, by eight industrial compa- quired to expand the usefulness of expert sys- nies. The government has provided seed mon- tems significantly beyond current limits. ey and staffing; 42 of the total staff of 52 have come from MITI’s Electrotechnical Labora- Foreign National Efforts in Artificial tory. 152 Spending thus far has been $2 million Intelligence R&D in 1982, $12 million in 1983, and $27 million in 1984.153 Artificial intelligence is a key pursuit of Overall, the functions that the Japanese see recently initiated national, government fi- their Fifth Generation system performing in- nanced R&D programs in Japan and Great 154 clude: Britain. Britain’s effort is in fact a reaction to the Japanese Fifth Generation Project 146 as it 1. problem-solving and inference, was detailed in a conference held in October 2. knowledge-base management, and 1981. ’47 The achievements of the objectives of 3. intelligent interface (with users), these targeted national programs will require These goals correspond roughly to previous- unprecedented fundamental advances in basic ly discussed research in: research. Their ultimate success is much less certain than were previous national technical 1. expert systems, efforts, such as the Manhattan Projector the 2. knowledge representation, and Manned Moon Landing. However, these national AI research commitments are likely to ——— lt~hese include a=iculture and fisheries, the office, and the produce advances in AI science, technology, 148 service industries; applications are expected, as well, in areas and commercial application. that are already highly automated such as electronics and automoble design and manufacturing. Fifth Generation Com- —.—. puter Systems, T. Moto-oka (cd.), op. cit., see p. 3. ‘“A British delegation reported that the extent and cohesive- IWP. Marsh, “The Race for the Thinking Machine,” iVew Su”en- ness of the Japanese plan, and the reaction to it that could be tist, July 8, 1982, p. 85-87. Note, this is the highest estimate expected from American industry, were a “major competitive encountered. threat. ” A Programme for Advanced Information Technology, 15’’ ’Fifth Generation Computer System Under Development, ” The Report of The Alvey Cou”ttee, Department of Industry, Science & Technology in Japan, January/March 1983, p. 24-26. Her Majesty’s Stationery Office, London, October 1982, see p. 152’’ ICOT: Japan Mobilizes for the New Generation, ” IEEE 5. Spectrum, November 1983, p. 48. ~d~The In~mation~ Conferene on Fifth Generation COmpUt- ‘s’Personal communication from John Riganati, National Bu- er Systems, Tokyo, Japan, Oct. 19-22, 1981. reau of Standards, June 1984. “’OTA workshop, Oct. 31, 1983. “’Science & Technology in Japan, op. cit., p. 24. 106 ● Information Technology R&D: Critical Trends and Issues

3. natural language understanding. ’55 aspects of basic AI research, and will later focus on specific demonstration systems to be The Fifth Generation R&D plan calls for 156 developed in collaboration between the strength- three phases of 3 to 4 years each: ened university programs and industries in- ● Phase I: Development of “basic technol- terested in particular applications.158 Expected ogies” consisting of computer architec- milestones include:159 tures and related software. ● The research community (numbered in ● Phase II: Development of prototypes of 1982 at some 150 people) is to grow by the inferencing and knowledgebase func- 50 percent in 2 to 3 years and double in tional components, to correspond to the architecture and software developed in about 5 years as a result of increases in graduate student funding, and numbers phase I. ● of faculty and research positions; Phase III: A “total system” prototype is ● - Computer equipment and software sup- to be developed in the final phase: port are to be expanded into an adequate The reactions of experts in this country to base for research in 2 to 3 years; the Fifth Generation Program goals and meth- ● An increased understanding will be ods have been mixed, ranging from enthusi- gained of knowledge representation and astic to skeptical. However, there is agreement intelligent computer interface concepts, that the goals the Japanese have established and knowledge based tasks and domains are quite ambitious for the time frame that is within 2 to 3 years; set for achieving them, and the program is ● Within 5 years, there will be substantial unlikely to succeed in all of its objectives. progress in understanding the application of knowledge based concepts using logic programming languages and parallel The Alvey Programme processors, expert systems and natural Named for John Alvey, chairman of the language understanding systems; committee that developed the plan for a Brit- ● Over the 10 year time course of the pro- ish response to the Japanese Fifth Generation gram, progressively more sophisticated Program, this academia-government-industry demonstration systems will be developed, cooperative effort will concentrate on four some early target applications being segments of information technology, one of teaching assistant programs, software which is artificial intelligence, termed by the production aids, and improved robot sys- British, “Intelligent Knowledge Based” Sys- tems; later applications might include tems.157 medical advisor systems, tactical decision aids, full 3-D vision systems, and office The British see a manpower shortfall in AI management and document production R&D similar to that in the United States, so systems. their program will initially have a strong educational component, focusing on the expan- The government has pledged a total of $310 sion of both academic and industrial training million for the entire Alvey program, with pri- facilities in the early part of the proposed 10 vate industry expected to contribute an addi- year effort. As the program progresses, em- tional $230 million.l6O The plan calls for ex- phasis will shift first to a broad effort in many penditures of about $39 million on Intelligent Knowledge Based Systems in the first 5 years, some $30 million of that going to universi- —. ‘5SL. R. Harris, “Fifth Generation Foundations, ” DatanMti”orI, July 1983, pp. 148-150, 154, 156. ‘5GIbid., p. 25. ‘6’ Ibid., p. 35. ISTThe other ~eas me Software Engineering, Man/machine “gIbid., pp. 38-40. O Interfaces, and VLSI. A Programme for Advanced Informa- “ M. Peltu, “U.K. Eyes 5th Gen., ” Datamation, July 1983, tion Technology, op. cit., p. 21. pp. 67-68, 72. Ch. 3—Selected Case Studies in Information Technology Research and Development . 107 ties. ’G’ The Science and Engineering Research prospects for artificial intelligence, and many Council (SERC), roughly the equivalent of the researchers emigrated to the United States to American NSF, will develop plans and dis- continue work in American universities.163 burse most of these funds.l62 Alvey Programme funding may reverse these trends, but there is disagreement among ex- The United Kingdom has a strong tradition perts concerning the ability of European uni- in AI, particularly in the University of Edin- versities, in general, to respond flexibly to the burgh and Imperial College, London. But in growth of a field such as AI, where interdis- the years immediately preceding the Alvey ini- ciplinary research across traditional academic tiative, efforts had been scaled back because departmental lines is so crucial. of an unfavorable government report on the —-— ‘“A Program For Advanced Information Technology, op. cit., pp. 47 and 49. “’Ibid., p. 48. “’The OTA workshop, Oct. 31, 1983. Chapter 4 Effects of Deregulation and Divestiture on Research Contents

Page Findings ...... 111 Antitrust Laws, Deregulation, and Divestiture ...... 113 Divestiture ...... 116 Management of Research at AT&T ...... 120 The-Modified Final Judgment and Bell Laboratories...... 121 Bell Labs After Divestiture ...... 122 Factors Affecting Research ...... 123 Stability of Earnings ...... 123 Allocation of Research and Development Expenditures ...... 125 Basic Research...... 128 Role of Bell Communications Research, Inc...... 129 Availability of Research Results ...... 130 Policy Implications...... 131 Chapter 4 References ...... 134

Table No. Page 23. Regional Bell Operating Companies ...... 119 24. R&D Intensities of Selected Major Telecommunication Firms ...... 125

Figures Figure No. Page 16. Pre-Divestiture Bell System ...... 117 17. Post-Divestiture Organization of AT&T ...... 118 18. Bell Operating Companies ...... 119 Chapter 4 Effects of Deregulation and Divestiture on Research

Findings

As a major source of information technol- ment. Because of the long-term nature of ogy R&D—and as an organization that has re- the work, however, it may take some cently undergone major legal, regulatory, and years for any changes to become evident. institutional changes-AT&T’s Bell Labora- The AT&T divestiture has been making tories merits special attention. In reviewing headlines since January 1982, when AT&T the potential effects of the AT&T divestiture and the Department of Justice announced the and of recent regulatory decisions on Bell settlement of the Department long-standing Labs, OTA made the following findings: antitrust suit. The divestiture marked the end ● Organizational changes within AT&T of an era. Before divestiture, AT&T had been Technologies and within Bell Labs indi- the nationwide provider of end-to-end telecom- cate that AT&T is already preparing to munications services. AT&T’s system of Bell speed the development and marketing of operating companies provided local service to new products. Other firms may also in- 85 percent of the telephones in the United crease their development activities to States; the Long Lines division carried the meet competition from AT&T. vast majority of long-distance calls; the West- • The effects on the research side are less ern Electric subsidiary manufactured most of clear. AT&T has some incentives to con- the equipment used in the system and leased tinue funding applied and basic research to end users. With assets of $150 billion and at past levels, but these stand in tension annual revenues of $69 billion, it was the big- with powerful new forces that could tempt gest communications company in the world. AT&T to direct more resources away On January 1, 1984, the size of the corpora- from research and into short-term re- tion was reduced to one-fourth as AT&T spun search and development projects. There off the Bell operating companies and gave up is little reason to think that AT&T’s com- local telephone service. petitors will perform more basic research While divestiture is indeed a dramatic event, now than they have in the past. the concern and publicity associated with it ● The areas where AT&T will be the most have tended to obscure a related regulatory likely to focus its competitive efforts are decision: the Federal Communications Com- also the areas where Bell Labs has been mission’s (FCC’s) decision in the Second Com- responsible for major scientific contribu- puter Inquiry (Computer II) detariffed the sale tions computer science, solid-state phy- of terminal equipment, deregulated enhanced sics, and photonics. Work in those areas, telecommunication services, and permitted including basic research, is likely to con- AT&T to sell these to end users through a sub- tinue into the foreseeable future. sidiary after Jan. 1, 1983.1 These changes in ● A significant portion of Bell Labs’ re- AT&T’s structure and markets have raised search base has been moved to Bell Com- some important questions related to research munications Research, Inc. (Bellcore), a and development in telecommunications. The unique new organization owned jointly by question addressed here is how divestiture and the divested Bell operating companies. deregulation will affect the functioning of Bellcore’s role in basic research is still AT&T’s research arm, Bell Laboratories, unclear. ● It is possible to monitor research activities over the next few years to determine “’Enhanced communications” are services which require add- whether the quality or direction of basic ing value to a transmission by altering the message in some research change in a deregulated environ- way, as explained below. 111

112 ● Information Technology R&D: Critical Trends and Issues viewed by some as the star of modern indus- physics and health care. The research results trial and scientific research. and technical standards are widely published in scientific and technical journals. Bell Labs In 1982, before any of the changes associ- researchers have made many fundamental ad- ated with divestiture and compliance with vances, inventing the transistor and other con- Computer II, Bell Labs had a budget of $2 bil- cepts at the base of the current generation of lion, facilities at 21 locations, and 25,000 em- computer and telecommunication technology. ployees-3,000 with doctorates and 5,000 with Among Bell Labs’ employees and alumni are masters degrees. Bell Labs provided nearly all seven Nobel laureates. R&D leading to the manufacture of Western Electric’s products, as well as systems engi- The Labs recently (1983) received its neering to support the Bell System generally. 20,000th patent; this amounts to one patent While the Labs’ principal role is in develop- per day since Bell Labs was incorporated in ing products for sale or use by AT&T, about 1925, and many other of its inventions have 10 percent of the budget has been dedicated not been patented. Traditionally, about 99 per- to scientific research. The research has had cent of AT&T’s R&D has been done internally. fallout applications in a wide variety of fields, Very little technology has been bought from from telephony and computer science to astro- outside, although AT&T does enter into cross-

Photo credit: AT&T Bell Laboratories Bell Labs developed the first 32 bit microprocessor: the dime-sized chip contains nearly 150,000 transistors and has processing power comparable to that of today’s minicomputers. Ch. 4—Effects of Deregulation and Divestiture on Research ● 113 licensing agreements. On the other hand, it reduce the scope or quality of research at Bell had been AT&T’s policy since the 1920s, and Labs alarms observers who see the Labs as a a legal requirement under the 1956 consent de- major contributor to the U.S. lead in informa- cree described below, to license its own patents ● tion technology R&D. to other firms at reasonable cost. There are The restructuring of AT&T creates pres- currently over 400 such licensing agreements sures and incentives for Bell Labs that did not outstanding in the United States and 200 exist while AT&T was a regulated, end-to-end more with foreign firms. Arno Penzias, Bell monopoly. Because of competitive pressures Lab’s Vice President, has been quoted as say- on AT&T in the deregulated markets, Bell ing that “without Bell Labs there would be no 2 Labs may choose to devote more of its re- Silicon Valley.” sources to product development and to reduce Although that may be hyperbole, it is cer- the number of long-term research projects tainly true that Bell Labs holds the basic leading to fundamental scientific discoveries. patents for the processes and products needed Such an event could be deleterious to the long- by many United States and foreign firms to run competitive position of AT&T, and more get their start in microelectronics, computers, importantly, might negatively affect the level telecommunications, or other fields. The avail- of U.S. R&D in information technology. ability of licenses and technical information This chapter discusses the problems and op- from Bell Labs greatly speeded development portunities that the new post-divestiture en- of the microelectronics industry. vironment offers Bell Labs, and the possible Bell Labs’ R&D efforts are clearly important effects that the changes in AT&T’s corporate to information technology generally. The structure may have on research at the Labs, Labs’ budget makes up perhaps 15 percent of and throughout the telecommunication and the R&D investment by information technol- computer industry. It focuses specifically on ogy firms.3 Further, if Bell Labs is producing the future stability of AT&T’s earnings, its over 370 patents per year, then it accounts for incentives to engage in research, and the pos- perhaps 5 percent of U.S. patents in informa- sible effects of deregulation on research else- tion technology fields.4 Anything that might where in the telecommunications and computer industry. Finally, it outlines some methods “’Bell Labs, Threatened Star of US Research,” Fortune, July 5, 1982, P. 47. for monitoring the health of research at Bell ‘See page 316. Investment in information technology by in- Labs and possible options for Federal Govern- dustry in 1983 is estimated to be about $10.8 billion. Bell Labs’ ment action. R&D budget of $2 billion is about 18 percent of the $10.8 billion invested in IT R&D by large IT companies in 1982. How- Before examining the effects on Bell Labs, ever, the $10.8 billion figure may be too low, as it does not include R&D expenditures of many small firms. it is necessary to briefly review the regulatory ‘This is an estimate based on approximately 5,180 Bell Labs and legal decisions leading to deregulation and patents as a fraction of 101,900 US patents in communications divestiture and to discuss the technological equipment and electronic components in 1963-1981. Data on and market forces that drove them. U.S. patents from National Science Board, ~“ence Inol”cators, 1982, U.S. Government Printing Office, 1983, p. 207.

Antitrust Laws, Deregulation, and Divestiture

American Telephone and Telegraph is no General that resulted in the Kingsbury com- stranger to antitrust litigation. In order to mitment. In the commitment, AT&T agreed: avoid a threatened Government suit under the 1) to end its policy of aggressive mergers with Sherman Antitrust Act in 1913, AT&T en- competing independent telephone companies; tered into negotiations with the U.S. Attorney 2) to allow the remaining independents to in- 114 . Information Technology R&D: Critical Trends and Issues terconnect with its long-distance system; and nical information along with patent licenses 3) to get out of the telegraph business by on payment of reasonable fees. This licensing divesting itself of the Western Union tele- provision ensured that other firms could use graph company.5 It removed AT&T from the Bell technology outside of regulated telephone telegraphy market and significantly constrained markets.6 future purchases of competing telephone com- Two major trends, each with a technologi- panies. cal and a regulatory component, developed However, the actual effect of the Kingsbury over the ensuing 25 years to make the line of commitment was to confirm AT&T as a regu- business restriction of the 1956 consent decree lated monopoly and to quell the competition increasingly unworkable. First was the devel- between Bell operating companies and inde- opment of technological alternatives in trans- pendents which had grown up in the 1895-1913 mission and switching that greatly reduced period. Under terms of the commitment, Bell the cost of providing long-distance service and companies and independents negotiated the made it economically attractive for competi- borders of their service areas and exchanged tors to challenge AT&T’s dominance of the telephones where necessary to give each other long-distance market. Second was the advance geographical monopolies. AT&T was acknowl- in computer microelectronics, which has been edged to control the entire long-distance leading to a convergence and interdependence network, and the independents used that net- of communication and computation services. work as noncompeting partners in end-to-end These technological changes, and the market service. activity that they generated, led to a number of regulatory decisions that eroded AT&T’s The next major antitrust case against AT&T, monopoly position and gradually opened the in 1949, asked for an end to AT&T’s owner- telecommunication transmission and equip- ship of Western Electric and an end to all re- ment markets to competition. strictive agreements among AT&T, the Bell Operating Companies, and Western Electric. The first chink in the long-distance monop- The suit essentially sought the separation of oly was FCC’s 1959 Above 890 decision,7 open- regulated monopoly services from equipment ing the microwave radio spectrum to private supply. users. This led eventually to FCC’s approval, in 1969 and again in 1971, of MCI’s applica- A negotiated settlement of the 1949 suit led tion for authorization to offer private line serv- to a consent decree in January 1956. The con- ice via microwave. It was also in 1971 that the sent decree imposed two important restric- FCC made its Specialized Common Carrier tions on AT&T’s future activities. First, AT&T 8 decision, in which it concluded that a general was restrained from entering other lines of business, such as the sale of solid-state com- policy in favor of entry by new carriers into specialized communications would serve the ponents or computers. It was restricted to pro- public interest. Long-distance service from viding regulated common carrier service, with “other common carriers” became more widely Western Electric as its captive equipment available to the public in 1979 after a series manufacturer. AT&T was free to develop Bell of FCC and court decisions. By the end of Labs technology, such as the transistor, for 1984, other carriers had captured 15 to 20 per- use within its own system, but was forbidden to market these products to the public. Sec- ond, AT&T was required to license all patents 5AT&T had been granting licenses and making available techn- controlled by the Bell System to any applicant ical information on its inventions before 1956. AT&T had de- at a “reasonable royalty” and to provide tech- veloped cross-licensing agreement with major manufacturers like General Electric over the previous two decades. The policy of licensing patcmts to smaller firms was in force in the 5Gerald W. Brock, The Telecommum”cati”ons Industry: The 1940s. Dynamics ofi%farket Structure (Cambridge: Harvard Univer- 727 FCC (1959). sity Press, 1981), p. 155. *29 FCC 2nd 8 70 (1971). Ch. 4—Effects of Deregulation and Divestiture on Research ● 115 cent of the long-distance market, as measured brid communications but could not provide by minutes of calls transmitted.9 Other carriers any service or product that fell into the data can now claim relatively small numbers of sub- processing categories. scribers, but they are principally the high vol- Throughout the 1960s and 1970s, AT&T ume users. AT&T estimates that other carriers had been manufacturing and selling terminals serve about one-third of the highest volume for access to mainframe computers. These residential callers (those spending over $25 per were primarily built by the Teletype Corpora- month) and one-half of high volume business tion, a subsidiary of Western Electric. Early callers (over $150 per month).* terminals were clearly communication de- In the terminal equipment market, the vices—they were only of use for sending infor- FCC’s 1968 decision in the Carterphone case10 mation to a remote computer for processing. was the first FCC action to allow consumer- As microelectronics advanced, however, more owned terminal equipment to be attached to intelligence and power could be placed in ter- the Bell system network. This decision, to- minals. It became increasingly difficult to de- gether with an equipment registration pro- termine at what point a terminal ceased to be gram authorized by FCC in the 1970s, allowed a “hybrid communications” device and be- manufacturers other than Western Electric to came a “hybrid data processing” device. enter the U.S. market, giving rise to the “in- AT&T’s applications to the FCC for permis- terconnect” market for telephones and other sion to market new terminal equipment were customer equipment. sometimes challenged as being in violation of Meanwhile, the computer industry was Computer I rules and the consent decree. ” growing rapidly and without significant gov- Further, AT&T was at a competitive dis- ernment regulation. In order to determine how advantage because it had to go through a best to deal with the policy questions that (sometimes lengthy) regulatory process before were already emerging from remote-access introducing each new product, whereas the data processing, FCC initiated in 1966 its first unregulated terminal suppliers (computer Inquiry into Regulatory and Policy Problems manufacturers) could introduce new products Presented by the Interdependence of Comput- whenever, at whatever price, they chose. It er and Communication Services and Facilities was clear that the combination of Computer (Computer I Inquiry). The decision in Com- I rules, the consent decree, and the evolution puter I, adopted in 1970, divided comput- of technology were preventing AT&T from of- er/communications services into two regulated fering state-of-the-art terminal equipment to services—pure communications and hybrid the public. communications-and two nonregulated serv- FCC initiated its second inquiry, Comput- ices—hybrid data processing and pure data er II, in 1976 and issued a decision in 1980. processing. Under the terms of the 1956 Con- That decision deregulated the sale of terminal sent Decree, AT&T could provide pure and hy- ———— equipment, both voice and data, and allowed *FAA estimate, private communication, February 1985. — —- llF e ~ , AT & T ’ rquegt *AT&T Communications briefing to OTA staff, August 17, or xm e g for a tariff to sell the Data- 1984. speed 40/4 was denied by FCC’s Common Carrier Bureau in De IO% 13 F~ Zd, 420, 437 (1968). “Terrnina.1 equipment” or cember 1976. IBM and others objected that the terminal would “customer premises equipment” termin ates the telephone wire be in direct competition with terrnin sls built by computer man- on the customer’s premises. The most common example is the ufacturers, and the Common Carrier Bureau agreed that the ordinary telephone. The terms are also used to refer to systems terminal’s storage and processing capabilities (designed to allow of telephones, like the six button “key sets” used by many small an operator to correct mistakes before sending data to the com- businesses, and to switching equipment, like the private branch puter) violated FCC rules. The full Commission overturned this exchanges (PBX) used to route calls inside large businesses. decision 9 months later. In its decision the Commission noted Modems (modulator~demodulat.ors that convert analog signals that the Computer I rules were inadequate to deal with the to digital signals) interface between the telephone wire and a changing technology and that Computer II Inquiry then be- computer and are considered terminal equipment, as are com- ginning would establish a new policy. See FCC Transmittal No. puter te rminals with built-in modems. 12449, 1977. 116 • Information Technology R&D: Critical Trends and Issues

AT&T to offer this equipment for sale to the allowed AT&T to enter computer, computer- public through a subsidiary. Computer II also related, and information services markets in allowed AT&T to offer other enhanced tele- competition with unregulated firms (although communication services through a subsidiary.12 there are still restrictions on AT&T’s actions; AT&T created that subsidiary, AT&T Infor- e.g., AT&T may not provide information serv- mation Services or ATTIS (originally called ices over its own lines for 7 years). American Bell), in June 1982. The breakup, according to Judge Greene, reduces AT&T’s ability to rely on its monopoly Divestiture at the local exchange to exact competitive ad- Meanwhile, the Department of Justice vantage in interexchange (long-distance), ter- brought an antitrust suit against AT&T in minal equipment, and computer services mar- 1974, seeking many of the same goals as in kets. AT&T’s long-distance market is still 1949. The suit alleged that AT&T monopolized regulated, but FCC regulation was not viewed the manufacturing, long-distance, and local by the court as so extensive, nor were barriers service markets; that it used its monopoly to entry seen as so high, that AT&T will be power in each market to strengthen its power able to use its currently large share in this in the other markets; and that it attempted market to provide a competitive advantage in to prevent competing equipment manufactur- unregulated segments of the industry. ers and long-distance carriers from gaining ac- Figures 16 and 17 compare the predivesti- cess to the local networks. In January 1982, ture and post-divestiture organizational struc- Department of Justice announced that it had ture of AT&T and the Bell operating compa- reached agreement with AT&T on changes to nies. Before divestiture the entire Bell system the 1956 consent decree and in August 1982, existed under a single corporate umbrella and Judge Harold H. Greene of the U.S. District the firm was organized to provide end-to-end Court for the District of Columbia approved telephone service. The Long Lines division the Modified Final Judgment. The Govern- provided interstate long-distance services; ment’s case was dismissed upon acceptance Western Electric manufactured equipment for of the terms of the Modified Final Judgment use throughout the system; the 22 wholly by all parties. owned Bell operating companies provided lo- Under the Modified Final Judgment, local cal and intrastate service; a small international Bell operating companies providing local ex- division marketed AT&T equipment abroad. change telephone services were divested by Bell Labs provided design and development AT&T, and spun off into seven regional hold- for Western Electric as well as research and ing companies. AT&T retained ownership of network system engineering for the rest of the a nationwide intercity network composed of system. The AT&T Information Systems sub- its Long Lines division and the intercity fa- sidiary was created in 1982 in response to the cilities of the Bell operating companies, and Computer II decision. continued to own Bell Laboratories and West- As figure 17 shows, AT&T after divestiture ern Electric. The Modified Final Judgment is primarily comprised of AT&T Communica- tions, AT&T Technologies, AT&T Interna- l~In it9 ~ond ComPuter Inquizy decision, the FCC distin- tional, and the subsidiary, AT&T Information guished between basic and enhanced services. Basic services Systems. AT&T Communications provides were defined to be the transmission of information, while long-distance service between local calling enhanced services involved adding value to transmission by 13 changing or acting on the message itself in some way. As an areas. AT&T Technologies includes the func- example, in voice traffic, a simple long-distance telephone call tions of Western Electric and Bell Labs. It constitutes basic service. Enhanced service would be provided now provides research and development, man- if the carrier stores and forwards calls or provides recorded messages for those who are calling. An enhanced data service might be one that provides protocol conversion so that non- laLATA—~~ A@ss and Tr~port Area-is the term now compatible computers can communicate. used to identify a local calling area. Ch. 4—Effect of Deregulation and Divestiture on Research • 117

Figure 16.—Pre.Divestiture Bell System a

Pacific/Nevada Southwestern Southern aAll entitles report to AT&T ufacturing, and marketing of equipment and for example, cannot be shared with ATTIS services both in the United States and abroad. (unless it is also shared with competitors). Western Electric no longer exists as an organizational unit, but AT&T Technologies will As shown in figure 18 and table 23, divesti- continue to use it as a trade name. Bell Labs ture places the Bell operating companies into is the section of AT&T Technologies respon- seven regional holding companies, of approx- sible for R&D. imately equal size in terms of assets and customer base. The seven jointly own and oper- AT&T Information Systems will market in- ate Bell Communications Research (Bellcore), formation services, terminal equipment and which provides technical and administrative computers to end users. Dealings between services. ATTIS and the other AT&T entities, under rules of Computer II, must be at arm’s length. Judge Greene ruled shortly after the divest- Information related to AT&T’s customer base, iture that the name “Bell” and the familiar 118 • Information Technology R&D: Critical Trends and Issues Ch. 4—Effect of Deregulation and Divestiture on Research • 119

Figure 18.-Bell Operating Companiesa

The seven regional Bell operating companies

1 Nynex New England Telephone New York Telephone — 2 Bell Atlantic C&P Telephone (4 companies) Diamond State Telephone New Jersey Bell Bell of Pennsylvania

3 Bellsouth Southern Bell I South Central Bell

4 Ameritech Illinois Bell Indiana Bell Michigan Bell Ohio Bell Wisconsin Telephone

5 Southwestern Bell Southwestern Bell

6 U.S. West Mountain Bell Northwestern Bell Pacific Northwest Bell

7 Pacific Telesis Paclflc Telephone Nevada Bell a The ~eglona~ Bell ~peratlng ~ompanle~ are h~ld,ng ~~rnpant~s for Bell operating companies that offer service In the States Inchcated above Within mOSt StateS IOcal telephone service IS also provided by Independent telephone companies SOURCE: AT&T.

Table 23.—Regional Bell Operating Companies

Total operating Value of revenue 1984 assets Net income embedded plant Access lines Regions (millions) (billions) (millions) (millions) (thousands) Ameritech ...... 8,900 16,26 1,037.1 14,409 13,970 Bell Atlantic...... 8,732 16.26 1,054.5 14,596 14,011 Bell South ...... 10,512 20.81 1,393.1 19,081 13,367 NYNEX ...... 10,006 17.39 1,029.8 15,186 12,658 Pacific Telesis...... 7,895 16.19 977.1 14,493 10,717 Southwestern Bell ...... 8,859 15.51 887.9 14,112 10,189 U.S. West ...... 7,596 15.05 910.9 13,767 10,381

SOURCE: Bell Communications Re9earch, Inc., November 1984.

38-802 0 - 85 - 9 120 . Information Technology R&D: Critical Trends and Issues logo are the property of the Bell system–that abroad. The one exception to this ruling was is, the Bell operating companies. AT&T may that the name of Bell Laboratories did not not use the name or logo in the United States, have to be changed, although it is now called although AT&T International may use it AT&T Bell Laboratories. -

Management of Research at AT&T

Bell Labs was named and incorporated in engineering. The majority of the funds used 1925, but it grew out of an in-house research for research and system engineering came to capability which AT&T had maintained since AT&T from the Bell operating companies 1907. AT&T was a groundbreaker in bringing under the “license contract. ” The contract was R&D out of the homes and private laboratories an arrangement under which the operating of individual inventors and into the industrial companies were assessed up to 2.5 percent of context. In many ways, research at AT&T was their annual revenues to pay for their use of a model for the modern industrial lab as it de- AT&T technical and administrative services. veloped in other industries. About 30 percent of these funds, together with a contribution from the Long Lines division, Most of Bell Labs’ resources have been de- were allocated to research. voted to design and development of products for sale or use in the Bell system. However, Funding of research at Bell Labs was anal- about 10 percent has traditionally been de- ogous to some of today’s attempts at joint re- voted to research. “Research” at Bell Labs en- search funding, such as Microelectronics and compasses those projects for which no specific, Computer Technology Corporation (MCC) or short-term benefit to the corporation is fore- Semiconductor Research Corporation (SRC).15 seen. Most of the research is applied or di- The operating companies were, in a sense, sep- rected systematically toward the solution of arate user companies that contributed to the particular problems, but some resources have support of a central research facility for mu- been devoted to basic research, sometimes tual benefit. The difference in this case was leading to major scientific advances. that the operating companies all existed under a single corporate umbrella, so that they had Before divestiture, Bell Labs’ work was sup- little control over how their contributions were ported by the other AT&T entities. In 1982, spent and no option of withdrawing from the and typically in the predivestiture era, about joint funding venture or establishing other ar- half of the Labs’ support (54 percent) came rangements. from Western Electric, to cover costs of specific design and development.14 In addition, A number of factors in the “climate” of Bell Western paid another 3 percent to support Labs have been cited as contributing to its work on products being developed under Gov- achievements in fundamental research. Some ernment contract. Another 11 percent came have pointed out that Bell Labs scientists had from Bell operating companies to pay for cen- access to state-of-the-art equipment, and were tralized development of computer information free to focus on their research without the re- systems. sponsibilities of teaching or serving on committees that would be required in a university The remaining 32 percent of Bell Labs budg- setting. Because of job security and the stabil- et was paid by AT&T for research and systems ity of funding, there was no need for research-

14 Fiweg from ch~les River Associates, Impacts of the ers to spend time pursuing grant support. AT& Td”vestiture on Innovative Behavior, unpublished paper prepared for OTA, 1983, p. 17. 15FOr a description of these joint research ventures, .9- ch. 6. Ch. 4—Effects of Deregulation and Divestiture on Research ● 121

There has been a tradition of staff interactions divestiture per se would hurt the quality of across disciplinary boundaries. Bell Labs service provided by the operating companies maintained an open publication policy-its re- or the research performed by of Bell Labora- searchers have published about 2,000 papers tories. He noted-that the largest potential cus- per year. With these advantages, Bell Labs tomers of Western Electric will be the divested was able to attract outstanding scientists and operating companies, hence, Western Elec- engineers to work in its research organiza- tric’s association with Bell Laboratories tion.l6 should provide an incentive to improve equipment and technology.19 The Modified Final Judgment and The Modified Final Judgment sets aside the Bell Laboratories 1956 Consent Decree and the requirement that In his opinion on the Modified Final Judg- AT&T grant nonexclusive licenses for its ment, Judge Greene commented on the pro- patents to any applicant. The elimination of posal that Western Electric and the Bell Lab- this requirement makes it easier for Bell Labs oratories be divested from AT&T. He noted to appropriate the potential benefit of new that the success of the Bell Laboratories in breakthroughs, and therefore might be consid- basic and applied research (and the beneficial ered an incentive to research. AT&T may now impact of that research on the Nation’s eco- grant or deny licenses as it chooses, and may nomic position) was due to its relationship change whatever royalty it chooses. Before with the operating companies and the Long divestiture, when revenues from local ex- Lines division. He argued that continued asso- change ratepayers were supporting Bell Labs’ ciation of the Labs with the AT&T entities research, it made sense to require AT&T to providing manufacturing and long-distance share the fruits of its monopoly financing with services would supply “the practical experi- others, according to Judge Greene. With the ence that would be useful in stimulating the divestiture of the-operating companies and the research operations. ”17 termination of the license contract fee payments, this rationale for required licensing is The possibility of negative effects on re- eliminated. Judge Greene also believed that search at Bell Labs was considered in the ne- the advance of technology and the dispersion gotiations leading to divestiture, but was not of knowledge related to telecommunications considered a matter of highest priority. Ches- l8 technology has reduced the dependence of es- sler, in summarizing the position of Govern- tablished domestic firms and foreign compet- ment negotiators, indicates that they accepted itors on information from Bell Laboratories.20 the possibility that divestiture might lead to a reduction in basic research activities: The Modified Final Judgment requires that AT&T grant licenses to the divested operat- The competitive era in station equipment, ing com-panics on all existing patents and all interexchange communications, and informa- patents issued for a period of 5 years follow- tion services under the [MFJ] will bring forth a great blossoming of progress in those areas ing approval of the Modified Final Judgment. of telephony. It was the thought of the frame- AT&T is also required to provide the operat- rs . . . that the blossoming will be so great ing companies with nonpatentable technical as to more than compensate for the loss of information that has been funded by the li- pure research at Bell Telephone Laboratories, cense contracts. The operating companies will and the reduced incentives for innovation at have the right to sublicense AT&T patents the Bell operating companies. and technic-d information to those providing Judge Greene did not believe that incentives them with goods and services. for innovation were being sacrificed or that

‘~TA, notes on interview with workshop participants. 170pinion and Order, Aug. 11, 1982, p. 62. I’Charles River Associates, “Impacts,” p. 43. “Cited in “Bell Labs on the Brink, ” Science, Sept. 23, 1983. ‘“Ibid. 122 ● Information Technology R&D: Critical Trends and Issues

Bell Labs After Divestiture after, nearly all Bell Labs personnel working on PBXs were collected in one Colorado Lab The most noticeable change resulting from facility near the Western Electric facility Computer II and divestiture is a reduction in where PBXs were manufactured. Bell Labs’ size. About 4,000 employees became part of AT&T Information Services Figure 17 also shows that research at Bell (ATTIS). FCC has interpreted the Computer Labs remains independent from the lines of II ruling that ATTIS and Bell Labs deal at business in AT&T Technologies. However, arm’s length to mean that ATTIS employees sources in Bell Labs note that research is must be kept separated from former Bell Labs undergoing review and changes as a result of colleagues, even though they are sometimes deregulation and divestiture. The loss of relocated in the same buildings. Another 3,000 search personnel to ATTIS and Bellcore caused Bell Labs employees went to the newly created some realignment of research projects. Some Bell Communications Research Inc. (Bellcore, other areas of research-for example, regu- formerly the Central Services Organization) of latory economics and social psychology-have the Bell operating companies. This leaves Bell been judged unproductive or inappropriate Labs with about 18,000 employees, returning and have been cut back. New research topics, it to approximately the size it was in 1978. such as robotics, are being undertaken. Organizational changes taking place else- A major change to Bell Labs’ funding since where in AT&T Technologies will also affect divestiture is the termination of the license Bell Labs. In 1983, AT&T Technologies was contract revenues from the local operating organized into “line of business” divisions companies, funds that were specifically dedi- defined by customer and product type. Within cated to research and system engineering. Bell Labs, development teams have been re- Under the current funding arrangements, reorganized along the same line of business cat- search is supported by AT&T Headquarters egories in order to facilitate cooperation with with funds provided by the AT&T companies manufacturing. 21 Authority for managing de- under a “composite allocator. ” AT&T entities sign and development of products within Bell will be assessed for Bell Labs research (as well Labs was given to executives running each line as administrative functions of AT&T Head- of business division, as shown by the dotted quarters) according to their size, number of lines in figure 17. employees, and revenues. The allocation formula, under the Computer II rules, must be This is a major departure from previous reviewed and approved by FCC to ensure that AT&T policy wherein Bell Labs, Western Elec- AT&T allocates a reasonable portion of re- tric, and AT&T shared this authority; unlike search costs to the unregulated portion of its practice at most firms, the old arrangement business and does not subsidize it from regu- gave Bell Labs some control over a product lated long distance revenues.23 even after it went into production. The new arrangement was chosen to make development Another major change for Bell Labs will be more responsive to the needs of marketing and an increase in the work done on military proj- manufacturing, and is a preparation to enter ects. AT&T Technologies is planning to in- competitive markets. Although the organiza- crease the number of defense contracts, and tional structure is new, it marks the continua- the design and development work will be done tion of a trend which began when the market in Bell Labs. Although defense contracts were 22 for large private branch exchanges (PBXS) once very important to Bell Labs, they had became competitive after 1968. Shortly there- been reduced to a minor part of the R&D budget during the 1970s. In 1971, Bell Labs derived ~lBrO utt~, “cold New World,” Fortune, June 27, 1983, P. 83. 30 percent of its income from defense-related ~~private Br~ch Exch~W is a generic term for the switch used on the customer premises for routing calls within a build- *gSee FCC 83-600, Dec. 22, 1983 and FCC 83-123, Mar. 31, ing or organization. 1983. Ch. 4—Effects of Deregulation and Divestiture on Research ● 123 work; by 1976 the share was down to 2 per- kinds of large complex systems that the De- cent and in 1983 about 3 percent.24 partment of Defense wants. As Solomon J. Growth of defense projects to an expected Buchsbaum, executive vice president for con- 10 percent of Bell Labs’ budget should not be sumer systems, points out, “The military side of government is a voracious eater of new tech- difficult. Based on its previous work, AT&T 25 nology, and we are good at [providing] that. ” has strong ties with the Pentagon and a good reputation for designing and building the ——. -— 24Marilyn A. Harris, “Bell Labs Looks to Military Research, ” 2s’’Bell Labs: The Threatened Star of U.S. Research, ” Bzzsi- Electrom”cs, Feb. 9, 1984. ness Week, July 5, 1982, p. 49.

Factors Affecting Research

All the changes taking place in AT&T’s mis- firm. While the predivestiture AT&T had a sion, markets, and corporate structure cannot book value of $150 billion, the new AT&T has but affect the activities of AT&T Bell Labs. assets of only about $34 billion. However, the The purpose of the Labs has always been to new AT&T is expected to have a much more provide research, systems engineering, and favorable ratio of revenues to assets. Annual product design to support the corporate activ- revenues are now expected to be on the order ities of AT&T. As those activities have evolved, of $57 billion, compared with $69 billion for the role of the Labs has also changed. Of par- the predivestiture firm. This is largely because ticular concern to many observers is the way AT&T will continue to provide long-distance in which deregulation and divestiture might service, which has traditionally been very prof- cause changes in the commitment to research, itable and is estimated to provide two-thirds particularly basic research, within Bell Labs. of the corporation’s profit base.26 Also, AT&T will continue to manufacture telecommunica- Several concerns have been voiced. Will tions equipment. Further, AT&T now has the AT&T, as a smaller corporation with a narrow- opportunity to expand into potentially prof- er revenue base, be able to support research itable computer-related markets. as it has in the past? What incentives does AT&T have to allocate funds to research, and While competitive computer markets are po- how strong are they compared to incentives tentially profitable, they are also notable for to allocate more resources to development of their volatility over the past few years: new competitive products? How will changes re- firms and new products have had meteoric suc- lated to divestiture and deregulation affect re- cesses and catastrophic failures. This kind of search at other firms in the telecommunication market may be dangerous for a firm which is and computer industries? Could a reduction unaccustomed to competition. AT&T has not in the level of research at Bell Labs have a neg- been particularly successful in markets where ative effect on U.S. research generally, and if it has been open to competition in the past. so, what can be done about it? The remainder After 1976, when customers were permitted of this chapter addresses these questions. to purchase their own private branch exchange (PBX) switching equipment from other man- Stability of Earnings ufacturers, AT&T’s market share fell sharply. Although AT&T is still the largest single man- The future funding of research at Bell Labs ufacturer, it now has 24 percent of U.S. sales, will depend, at least in part, on AT&T’s success in the market. The combination of divest- ‘Peter Hall, “AT&T and the Great Divide, ” Hnanci”al World, iture and deregulation leave AT&T a smaller Jan. 10, 1984. 124 . Information Technology R&D: Critical Trends and Issues

compared to 100 percent 8 years ago. Major tion of a new plant in Florida to make lower competitors, specifically, Northern Telecom, cost chips for use in computers and switches. Rolm, and Mitel have shares of 16,14, and 11 AT&T Technologies will continue to be a percent, respectively .27 major manufacturer of telecommunication This loss of market share is due at least part- transmission equipment, large central office ly to AT&T’s higher prices and relative slow- switches, and terminal equipment. Potential ness in bringing new products to market. Both customers include Bell operating companies, these tendencies could be major disadvantages independent telephone companies, and tele- in industries that are noted for rapid introduc- communications agencies abroad.29 tion of new products and rapid obsolescence In addition, AT&T is now free to sell prod- of old ones. In part, slowness in bringing products it developed but could not market to the ucts to market was related to the regulatory public under the 1956 consent decree. It can process-a situation that has been eased since now market computers based on the UNIX Computer II, but not eliminated. operating system, the 256 K-byte memory AT&T has traditionally designed and manu- chip, and the 32-bit processor, all developed factured its products to extremely high at Bell Labs. Its 3B computer series will of- standards; they were expected to be highly fer a range of computers of varying size and reliable with a long useful life. Such a strat- capability. egy made sense when AT&T was the owner For the first time, AT&T is acquiring some of a huge nationwide network of transmission of its new products, marketing talent, and dis- and terminal equipment that had to be depre- tribution channels through other firms. For ex- ciated over 20 to 40 years. The higher costs ample, AT&T acquired a 25 percent interest of conservative design were made up by a long in Italy’s Olivetti Co. at a cost of about $260 production run. Western Electric maintained 3o million. Olivetti is Europe’s largest word a price advantage over some other manufac- processor and computer manufacturer. The turers by producing large numbers of stand- agreement is expected not only to supply ard products over many years. AT&T with Olivetti office equipment for the Western Electric has often been at a cost U.S. market, but also to provide a European disadvantage, however, in the case of newer distribution system for AT&T products. AT&T electronic products, the very ones that are the has also entered a joint venture with a Nether- target of the competitive market. Small digital lands electronics firm, Philips, to manufacture PBXs for example, cost Western about 75 per- central office switching equipment for Europe. cent more to manufacture than those made by AT&T has also made agreements with a num- their lowest cost competitor, Mitel.28 ber of smaller US. office computer manufacturers for development of new office automa- AT&T has made a concerted effort to stream- tion equipment. line its manufacturing and to reduce costs. AT&T Technologies is reducing its work force, At the same time, and equally importantly, and several former Western Electric factories AT&T is developing its own marketing capa- have been closed down or cut back. Although bility. Before divestiture, Western Electric AT&T did not sell integrated circuits and was strictly a manufacturer and dedicated other electronic components to the public, it very few resources to marketing. One observer is the Nation’s 12th largest manufacturer. notes that a competitor, Northern Telecom, AT&T Technologies is now expanding that spends about 9 percent of manufacturing sales manufacturing capability, including construc- on marketing while Western Electric spent

‘7Northern Business Information, Inc., as cited in “ITl”s Big Gamble” Business Week, Oct. 22, 1984. ~gKathl~n K. wiegner, “Prometheus is Unbound and Seek- *8Northem Business Information, as quoted in Bro Uttal, ing His Footing, “ Forbes, Mar, 12, 1984, p. 143. “Cold New World,” Fortune, June 27, 1983, p. 83. ‘“Ibid. Ch. 4—Effects of Deregulation and Divestiture on Research ● 125 only 1.2 percent in 1982.31 Now, AT&T Tech- of manufacturing sales AT&T spent approx- nologies will be responsible for marketing all imately 9.8 percent of revenues on R&D in the products not handled by ATTIS, and a mar- 1970s, as compared with an average of about keting division and all the support functions 2.8 percent for communication firms, and 1.9 are being developed. percent for manufacturers generally .32 On average, Bell Labs has spent about 10 Allocation of Research and percent of its R&D budget on research. Among Development Expenditures the other firms in table 24, both Northern Telecom and IBM also claim to spend about As listed in table 24, AT&T and IBM have 10 percent on research. While Northern Tele- the largest R&D budgets among the U.S. com is a competitive firm, it operates under firms shown. The R&D intensity, that is, R&D the corporate umbrella of Bell Canada, and as a percent of revenues, is based on total shares the expenses of Bell-Northern Research sales, which before divestiture included revenues of the operating companies providing with the regulated firm. local telephone service. AT&T’s R&D inten- The general argument expressed by con- sity is a fairly low 3.3 percent when based on cerned observers is that AT&T may be forced, total revenues. Nordhaus cites historical evi- because of competitive pressures, to invest dence, however, to indicate that as a percent more of its R&D funds in developing salable ~lBrO ut~, “cold New World, ” Fortune, June 27, 1983, P. 83. 9Zchwle9 R,iVer Associates, Op. Cit., P. 7.

Table 24.–R&D Intensities of Selected Major Telecommunication Firms (1982)

R&D Expenses Sales R&D intensity Company (millions of dollars) (percent) (1) (2) (1)/(2) AT&T ...... $2,126 a $65,093 3.3 ”/0 COMSAT ...... 22.3 ’41O 5.4 GTE ...... 267.0 12,066 2.2 Harris...... 92.8 1,719 5.4 ROLM ...... 24.4 381 6.4 United Telecommunications ...... 11.2 2,249 0.5 Western Union ...... 11.0 1,025 1.1 Zenith Radio...... 63.8 1,239 5.1 ITT ...... 519.0 15,958 3.3 Rockwell International ...... 222.0 7,395 3.0 General Dynamics ...... 139.0 6,155 2.3 IBM ...... 2,053.0 34,364 6.0 Motorola ...... 278.0 3,786 7.3 RCA ...... 195.4 8,237 2.4 General Electric ...... 781.0 26,500 2.9 L. M. Ericsson (1980)...... 231.3 2,779.4 8.3 Northern Telecom ...... 241.4b 3,035.5b 8.0 Plessey (fiscal year 1982) ...... 241.8 1,723.9 14.0 Siemens (fiscal year 1982) ...... 3,300(DM) 34,600(DM) 9.5 Thomson-CSF (1980) ...... 3,600(FF) 22,300(FF) 16.1 The CIT-Alcatel Group (1981)...... 1,000(FF) 1O,7OO(FF) 9.3 Hitachi (fiscal year 1981) ...... 609.0 15,996.1 3.8 NEC (1981) ...... 227.8 4,820.2 4.7 Fujitsu ...... 256.1 2,769.9 9.2 ‘{”~lud~~ $1,515 million spent by western Electric Co. and other subsidiaries, not worted in ATaT’s 10K. bCanadian dollars. SOURCES: Business Week, June 20, 1983; L. M. Ericsson Telephone Co., Anrrua/ Repoti, 1980; Plessy, Report and Accounts, 1982; Siemens, Annual Reporl 198142; Thomson4SF in 1980: The Year in Review; CITA/catel Group Review; Hitachi, 1981 Annua/ Report; NEC Nippon Electric Co., Limited, Annual Report, 1981; and Fujitsu Limited, Anrrua/ Repoti, March 1981. FROM: Charles River Associates, “Impact of the AT&T Divestiture on Innovative Behavior, ” unpublished report prepared for Office of Technology Assessment, December 19S3,

126 ● Information Technology R&D: Critical Trends and Issues products, and correspondingly less in funding petitive markets and it will be a smaller cor- research projects that may lead to future poration. scientific breakthroughs. They note that the Neither the theoretical nor the empirical re- license contract fees described above, which lationships between market structure, firm provided a steady income source for Bell Labs, size, and innovative activity are straightfor- will no longer be available, and that research ward or well understood. It is not clear whether funding will depend on yearly corporate deci- innovation is most likely to occur under con- sions. Any reductions, instability, or even ditions of competition or of monopoly. In ad- uncertainty about funds could have negative dition, although many support the view that effects on the productivity of research projects larger firms have more incentives to innovate, that by their nature require long-term atten- there are many examples, especially in the in- tion and investment. formation industries, of small firms that grow The arguments related to a possible change large due to extremely successful innovations. in AT&T’s policy toward research are based Further, most existing theory deals with “in- on two major effects of the deregulation and novative behavior” or R&D as a whole, rather divestiture-AT&T will be operating in com- than with the specific relationship of market Ch. 4—Effects of Deregulation and Divestiture on Research ● 127 structure or firm size with the basic research tage of innovations by other firms. IBM was component of R&D. given as an example of such a firm.38 One theoretical argument, as proposed by The empirical evidence on the effects of firm Schumpeter and others, is that innovative size on innovation is less ambiguous than the behavior is greater in monopolistic industries evidence on market structure. R&D at small than in competitive ones because a firm with firms is sometimes more efficient than at large monopoly power: 1) can prevent imitation and ones for R&D projects undertaken by both therefore capture more profit from innovation, large and small firms. 39 However, some proj - and 2) is better able to assemble the funds and ects are simply beyond the reach of small firms bear the risks of R&D.33 On the other hand, and there may be economies of scale for other critics of this position theorize that firms in projects. It appears that R&D intensity in- competitive industries are more likely to increases with firm size until firms reach annual novate because new products or processes will sales of $250 million to $400 million (1978 help them to reduce costs or increase market prices) and then level off.40 After reviewing the share. In this view, monopolistic firms would empirical evidence on firm size and innovation, be slow innovators because they can continue Scherer concludes that an industry with a to earn profits by continuing to produce the moderate degree of concentration and a vari- current products. In addition, because the ety of firms of different sizes is most conducive monopolistic firm is under less pressure to to innovation. operate efficiently, the results of innovative 34 All things considered, the most favorable activity would be obtained at excessive cost. industrial environment for rapid technologi- Real-world markets are characterized by cal progress would appear to be a firm size varying degrees of concentration rather than distribution that includes a preponderance of extremes of pure competition or monopoly. At- companies with sales below $500 million, tempts to empirically measure the relationship pressed on one side by a horde of small, technology-oriented enterprises bubbling over between innovation and degree of industry with bright new ideas and on the other by a concentration have had mixed results.35 For 36 few larger corporations with the capacity to example, Scherer found some evidence that undertake exceptionally ambitious develop- dominant firms in highly concentrated indus- ments. 41 tries are more innovative. However, in a later study37 he found that the relationship varied After divestiture, AT&T is still many times greatly depending on the industry, and that the threshold level of size that empirical stud- there were examples where higher levels of in- ies have associated with maximum R&D. It novation were associated with more competi- will still be the dominant firm in a telecom- tive industries. In some cases, dominant firms munications industry that fits well Scherer’s were only moderately productive innovators, description of the environment most favorable but they were able to aggressively take advan- for innovation. Thus, though the details of ———. — AT&T’s R&D may change, there are no con- 33 Summarized in Morton I. Kamien and Nancy Schwartz, vincing theoretical arguments or empirical Market Structure and Innovation (New York: Cambridge evidence related to market structure or firm University Press, 1982), p. 47. size that would predict a lessening of its in- SiMorri9 E. Morkre, “Innovation and Market Structure: A Survey, ” Working Paper No. 82, Bureau of Economics, Feder- novative activity. al Trade Commission, 1982, p. vi. S~The 1itera~re is review~ in Morkre, “Innovation, p. 11. ‘F. M. Scherer, “Firm Sizes, Market Structure, Opportunity, and the Output of Patented Inventions, ’ American Econom”c 3BIbid., p. 432. Review, 55:11 19. ‘gMorkre, p. vi. “F. M. Scherer, Industrial Market Structure and Econom”c 40Charles River Associates, “Impacts,” p. 88. Performance (New York: Rand McNally, 1980), p. 431-432. ilscherer, lndust~”~ Market Structure, P. 422. 128 • Information Technology R&D: Critical Trends and Issues

Scherer and others have concluded that the at how to exploit technology, not at how to important determinant of innovation may not push it forward.’’” be market structure or firm size but rather the richness of innovative opportunities opened up When AT&T gets more experience as a competitive firm, will it continue to do basic re- by the underlying base of scientific knowledge. search, or will it begin to behave as it appears Advances in science related to semiconductors, computers, software, satellites, micro- other competitive firms do, and dedicate more resources to product-oriented research and de- wave transmission, fiber optics, and lasers provide a rich set of technological opportunities velopment? Some observers, including Nord- haus, believe that AT&T will now “tilt much upon which to base innovations in telecommu- 42 more toward a conventional equipment man- nications and information technology. ufacturer, and it will therefore have a relatively greater incentive to invest in R&D that Basic Research will enhance its equipment sales and profits” and relatively less incentive to invest in basic Economic literature on “innovation,” how- 46 ever, does not deal adequately with the effect research. of firm size or market structure on contribu- At the present time, AT&T’s management tions to the knowledge base that supports in- has voiced a commitment to continuing funda- novation. The expected effect of competition, mental research, recognizing that advances in as noted above, is investment in development science are necessary to advances in technol- of new products and services, which will re- ogy. In testimony before the Senate Com- duce cost or improve market share in the short merce Committee, AT&T President Charles L. term. Investments in research, especially basic Brown called Bell Labs the “jewel” of the Bell research, may not pay off until many years system, and pointed out that “basic research after the initial investment is made. has been the root of Bell Laboratories success One unique characteristic of Bell Labs is its and will continue to be the root of it. We 43 do not intend to skimp on it . . . . This is some- reputation for doing basic research. In gen- 47 eral, only a few firms in the information indus- thing we have as a basic tenet.” tries have spent much on in-house basic re- It is probably true that more than half a cen- search. In 1981, out of 110 firms doing R&D tury of reliance on internally developed tech- in information technology, only seven did any nology will not be quickly tossed aside. One basic research at all, according to the National Bell Labs spokesman said that the “corporate 44 Science Foundation. Speaking of Bell Labs culture” of AT&T is completely oriented to- one observer from Bell-Northern Research ward doing basic research in-house. The forces noted, “Most other organizations are looking of habit and tradition may resist some pressures to shift too many resources to develop- —. —-. 48 42Charles River Associates, Impacts, p. 85. ment. Most of the technologies that will be 48National Science Foundation, Reseamh and Development commercially important to AT&T in the fu- in Industry, 1981, NSF 83-325 (Washington, DC: U.S. Gover- nment Printing Office, 1983), p. 3. The National Science Foun- ture–computer science, photonics, and solid- dation defines basic research as “original investigations for the state physics—are the very areas where Bell advancement of scientific knowledge not having specific com- Labs has made ongoing contributions to basic mercial objectives, although such investigations may be in fields of present or potential interest to the firm. “Information technologies in this case includes firms in the 46John A. Roth, executive VP, Bell Northern Research, as following categories: office, computing, and accounting ma- cited in “Bell Labs the Threatened Star of US Research,” BusJ”- chines (SIC 357); communications equipment (SIC 366); elec- ness Week, July 5, 1982. tronic components (SIC 367). See “Table B-33–Number of R8zD 4’William Nordhaus, cited in “Bell Labs on the Brink,” Performing Companies Conducting Basic Research By Indus- Science, Sept. 23, 1983, p. 1267. try: 1981,” p. 38 in National Science Foundation, Reseamh and 47Testimony of Charles L. Brown Before the Senate Commit- Development in Industry, 1981, NSF 83-325 (Washington, DC: tee on commerce, Saence and Transportation, March 1982. U.S. Government Printing Office, 1983). ‘sInterview, March 1984. Ch. 4—Effects of Deregulation and Divestiture on Research • 129 science. It is highly unlikely that AT&T will year.50 AT&T was free to use those funds much abandon research in these areas. Further, as as a government might use tax revenues, al- was pointed out in the case studies in chapter locating some portion of those revenues to 3, the boundaries between basic and applied support activities that were for the general research in these fields are sometimes very good but provided no immediate commercial fuzzy. Bell Labs’ researchers are likely to make benefit. While some research eventually paid some contributions to the advancement of off in discoveries useful to AT&T, some never science even in pursuit of commercial ends. paid off at all. Many research results that were There are some tangible and intangible ben- not of direct benefit to AT&T were made available to others through licenses of patents or efits of performing basic research that are as through scientific and technical publication. advantageous to AT&T now as they were before Computer II or divestiture. For example, As a competitive firm, AT&T must now sup- Nelson49 points out that research often yields port its research through a different internal discoveries and inventions in unexpected funding mechanism. An important aspect of areas. The wider a firm’s scope of activities, the deregulation and divestiture rules is that the higher the proportion of these unantici- AT&T will be watched closely by FCC to be pated outcomes it will be able to use. Thus, sure that it allocates a reasonable portion of diversified firms can realize higher rates of re- research costs to the nonregulated portion of turn from research, and engage in more of it its business. Before divestiture, most of the than firms with narrow product lines. Prior to cost of research was paid for by the Bell oper- divestiture, AT&T was a vertically integrated ating companies and the Long Lines division. firm which could make use of research results Under the new “composite allocator” devel- in a large number of areas. Although the size oped by AT&T and approved by FCC, approx- of the firm is now reduced, AT&T is now in imately 50 percent of research costs will be a position to diversify in other areas, and will paid by AT&T Communications and 50 per- continue to benefit from research results. cent will be paid by AT&T Technologies and AT&T Information Systems. An additional benefit to funding basic research is that a reputation for achievements in basic science gives the Labs a certain pres- Role of Bell Communications tige, credibility, and glamour, even if its chief Research, Inc. business is not basic research. Although these Another unknown factor in the future of benefits are not quantifiable, they are useful telecommunications research is the role of Bell in attracting qualified scientists and engineers. Communication Research, Inc. (Bellcore), the Divestiture and Computer II changed the technical services organization owned by the rules under which AT&T funds research, and regional holding companies. The scope and there has been speculation that the new rules quality of Bellcore’s research effort is still un- may bring about a reduction in the amount known. One of Bellcore’s jobs will be to test spent on research over the long term. As a and evaluate products and equipment for the rate-base regulated monopoly, AT&T was able Bell operating companies. In order to do this to spread the costs of basic research over properly, Bellcore will have to stay ahead of many ratepayers. The license contract fee was the manufacturers, anticipating the state-of- essentially a “tax” on telephone calls. The the-art and doing some basic research. Accord- revenues generated provided a regular source ing to Alan G. Chynoweth, Vice President for of income that could be counted on year after ‘OAT&T points out that the license contract payments were ‘gRichard R. Nelson, “The Simple Economics of Basic Scien- not completely guaranteed income. Occasionally a Stab regu- tific Research, ” Journal of PoliticaJ Economy 67, 3 (June): pp. latory comrnission would disallow a portion of a BOC’S license 297-306. contract payment. 130 ● Information Technology R&D: Critical Trends and Issues

Applied Research, “Everything we do will be gional operating companies are creating sub- chosen because of its relevance to the long- sidiaries to enter other lines of business. term needs of the telephone companies. We’re Among the enterprises already under way are smaller than Bell Labs. We have to be more computer sales and repair, computer software selective. But in those areas we select to be sales, office equipment sales, cable television expert in, we’ll dig very deeply. ”51 Among the installation, and real estate development. In areas where research will be done are mathe- many cases the regionals are providing goods matics and computer science, materials, solid- and services in nationwide markets, in direct state science, fiber optics and photonics, and competition with one or more of the others. switches. Other ventures are being planned, subject to approval by Judge Greene’s court, under terms Nearly half of Bellcore’s technical person- of the divestiture. nel came from Bell Labs. To the extent that former Bell Labs research personnel will still The regional Bell operating companies have pursue the same sorts of problems at Bellcore, many common R&D goals because the majori- the value of their research contributions has ty of their business will continue to be the pro- not been lost to the Nation. It remains to be vision of regulated local and interstate tele- seen whether the creation of Bellcore will have phone service. However, there are a growing a positive or negative impact on basic research number of areas where their interests diverge in areas related to information technology. The or where one company wishes to withhold research agendas of Bellcore and Bell Labs will information from some or all of the others. naturally overlap in certain areas. It is possi- Bellcore is still developing an organizational ble that this duplication of effort will be inef- structure to deal with this situation. It is pos- ficient and may reduce the quality of U.S. re- sible that the growing competition between search in information technology overall. On the owners could encourage them to jointly the other hand, it maybe that the creation of fund basic research at Bellcore but to turn to this new center of initiative will have a stim- other labs for development of products needed ulating effect on research. for the competitive market. At this point it is impossible to say what Bellcore’s long-term re- The regional Bell operating companies are search agenda will be. Bellcore will be an in- the owners of Bellcore and have control over teresting experiment in jointly funded R&D. how funds are spent. Under the current ar- It remains to be seen how much of Bellcore’s rangement, they all contribute to certain resources the regional Bell operating compa- “core” projects, but each is able to limit its nies will be willing or able to spend on basic investment in “noncore” projects it does not 52 research with possible long-term payoffs. believe to be beneficial to its own business. Funding priorities for Bellcore will depend Availability of Research Results partially on actions of State regulatory com- Even if Bell Labs continues to perform re- missions. Before divestiture, a few State com- search at the current levels, it has fewer in- missions sometimes disallowed part of a Bell centives to make the results available to others. operating company’s payment for support of It has maintained a fairly open policy, en- Bell Labs on the grounds that research did not couraging its scientists to publish results, benefit the telephone ratepayers of that State. present papers, and consult informally with Support of research at Bellcore may face the other researchers. Some of the research re- same sort of problem. sults, as well as some of the patents, were of The growing competition among its owners little direct value to AT&T because it was per- may also affect Bellcore’s future. Although mitted only to provide regulated common car- they provide regulated telephone service only rier service. But some of them were of im- within their assigned geographic areas, the re- mense value to firms in related fields and even to AT&T’s competitors. 5’Lee Dembart, “Dividing Bell Labs: Breakup to Put the Best to New Test, ” Los Angeles Times, Sept. 6, 1983, pp. 1,3. ‘*Remarks of Irwin Dorros at seminar “Research at Bell” held Now, according to Bell Labs Vice President Apr. 5, 1984, Massachusetts Institute of Technology, Program Arno Penzias, AT&T will “have the opportu- on Research in Communications Policy. nity and motive to use our own technology. ” Ch. 4—Effects of Deregulation and Divestiture on Research • 131

Only a small number of basic research projects lead to results that have an obvious application. In some of those cases AT&T would probably get patent protection before publishing the results. In other cases, the published paper may report a discovery without giving details of how to duplicate it. This type of protection has been used by many labs, including Bell Labs in the past. Penzias noted that at IBM the number of papers published per dollar of research is about the same as Bell Labs, even though IBM is a competitive firm. 53 A policy of complete openness of research Photo credit AT&T Bell Laboratories results may be transferred to Bellcore. Its in- An experimental, interactive computer-based system terest is to see that research results are dis- is helping Bell Labs engineers design integrated circuits. seminated widely so that manufacturers can use them to produce the best and lowest cost products for use by the Bell operating companies. Bellcore itself, under its current char- However, he emphasized that in the area of ter, will not be able to manufacture products basic research, Bell Labs is still part of the or otherwise benefit from any discoveries or scientific and technical community where com- inventions resulting from its research. There- munication and trading of information is vital. fore, it may establish a publication and licens- In order to benefit from the results of research ing policy even more open than Bell Labs’ has elsewhere, it will have to continue to share its been in the past. research results. In order to keep good scien- ‘9Arno Penzias, remarks at a seminar “Research at Bell, ” held tists on the staff, it will have to allow them Apr. 5, 1984, Massachusetts institute of Technology, Program to publish. on Research in Communication Policy.

Policy Implications

The recent divestiture and the entry of new policies have added to the uncertainty and AT&T into competitive markets poses new change in the information industry. Several challenges for U.S. policy toward the telecom- years under the new rules will be necessary munication and information industries. The before all the effects can be assessed. 1979-83 period in which the divestiture and Similarly, the full effects of deregulation and Computer II decisions were announced and divestiture on the quality and direction of re- implemented was also a period of intensive search at Bell Labs will only become clear as congressional debate about telecommunica- this “shakedown” period goes on. Neither his- tions. Bills have been introduced to modify the tory nor economic theory seem to be of much Communications Act of 1934, to deregulate help in foreseeing the future of research at Bell parts of the industry, or to force some version 64 Labs. There appear to be only a few things of AT&T divestiture. Many of the policy is- that government can do about major changes sues raised in this legislative debate have now in research at Bell Labs. Clearly, in the post- been addressed by FCC in Computer II and divestiture era, decisions about the funding through settlement of the Department of Jus- and nature of research will be in the hands of tice suit. Speculation over the effects of the AT&T management. This is not new. Deci- “For example, S. 898, H.R. 5158, as introduced in the 97th Congress, are only two bills which proposed modifying the 1956 sions about research have always been man- consent decreq creating a subsidiary of AT&T to enter new un- agement decisions, in AT&T as throughout regulated markets, and stimulating competition in terminal U.S. industry. equipment. 132 . Information Technology R&D: Critical Trends and Issues

It is possible that, in the new climate cre- protect possible commercial advantage stem- ated by deregulation and divestiture, AT&T ming from certain types of research. management will make decisions about re- In addition to monitoring the number of pa- search that will allow the quality or quantity pers published, it might also be possible to ex- of Bell Labs’ research to decline. In that case, there may be a role for limited government ac- amine the quality of the journals in which they appear. Although this measure is subjective, tion. Some regulatory or funding policies it should reflect the quality of Bell Labs work might be developed to stimulate or facilitate as viewed by other members of the scientific research. These policy changes, discussed later community. Researchers in all fields have a in this section, might be aimed at AT&T alone, but might also be applied more generically to clear idea which of their journals is the “best.” To the extent that Bell Labs work continues raise the quality of research in industry, uni- to be published in the same sorts of journals versities, and government. as now, it may be evidence that the quality However, it would be premature to intro- of results remains unchanged. A shift to pub- duce policy changes without evidence that the lication in less prestigious journals might in- current institutional arrangements are inade- dicate a decline in quality. quate, or that the U.S. research capability is The vigor of research can also be measured in jeopardy. The first step of government ac- by its usefulness to other researchers. Thus, tion might be to monitor Bell Labs’ research a possible measure of the continuing value of over the next several years to see whether the Bell Labs research would be the number of quality of research actually changes. The monitoring effort might be expanded to in- times their work is cited in papers published by scientists in universities and other labs. In clude the whole range of industry and univer- addition, the attitude of the scientific commu- sity research in information technology. It nity toward Bell Labs could be monitored by would not be difficult to develop an analyti- its ability to attract and retain well-qualified cal framework and a set of criteria for meas- research workers. uring the vigor or quality of research. A number of possible measures are suggested below. There is the possibility that, even with a re- While none of them is decisive in isolation, to- duction of basic research at Bell Labs, research gether they might give a picture of the health in the information field generally will not suf- of research at Bell Labs and at other research fer. Scientific research may simply move to organizations. 65 other laboratories. Any decline in quality of For example, it would be possible to moni- fundamental research would certainly make it harder for Bell Labs to attract and keep a staff tor the funds that AT&T allocates to research of qualified scientists. Top graduate students over the next few years. Dollar amounts seem would choose to work at other firms or at uni- very objective and quantifiable, but alone are not a sufficient gauge of the quality or direc- versities. To the extent that researchers continue to work in the same scientific fields and tion of research effort. For example, if all basic are equally productive in their new surround- science were dropped and the research effort ings, there may be no noticeable effect on U.S. steered toward more applied projects, the total amount spent for “research” might remain the basic research. same. This criterion may be useful, but can- To get a complete picture of the effects of not be used in isolation. deregulation and divestiture on the state of in- Another measure would be the number of formation technology basic research in the United States as a whole, it would be neces- papers by Bell Labs scientists published in prestigious scientific and technical journals sary to monitor the research performed through- each year. A decline in the number of papers out industry and at university labs as well. could be an indication that the amount of re- Even then, it would be extremely difficult to attribute observed changes in the U.S. research is declining, perhaps, or that AT&T is search environment to changes occurring at significantly limiting publication in order to Bell Labs. As noted in chapter 2, the state of “Some of the measures listed have actually been used infor- information technology research is in flux and mally by Bell Labs management to monitor the strength of research in the Labs. changes will occur with or without Bell Labs. Ch. 4—Effects of Deregulation and Divestiture on Research . 133

It may be possible, however, to trace some If it is determined that changes in basic re- causal factors and to, at least, draw reason- search at AT&T have had a major effect on able inferences. the U.S. research capability, and that Govern- ment action is warranted, there is a question One difficulty with the proposed studies is of what can be done. Basically, it appears that the collection of relevant data over an ex- two general approaches might be considered. tended period of time. While the data needed Regulatory policies might be changed to mod- are not extensive, they include items that ify the rules under which AT&T operates, giv- firms do not currently report to any Federal agency (except that the FCC will continue to ing it greater incentives to perform basic research or requiring it to do so. More broadly, be concerned with AT&T’s research budget). consideration could be given to implementing Some special effort and cooperation on the funding policies that might stimulate more ba- part of industry and the Federal Government sic research throughout industry and in uni- would be needed to collect and analyze the nec- versity laboratories. essary information. In the regulatory area, for example, it might It is difficult to say who might be best be possible to allow some subsidy for basic re- suited to carry out the studies mentioned search. At the present time, FCC is working above. One possibility is the FCC, which is re- under the terms of the divestiture and Com- sponsible for oversight of many aspects of AT&T’s business. However, many of the firms puter II to make sure that AT&T does not use the revenues it earns in the regulated market and institutions engaging on information tech- to support research or development that leads nology research are not regulated by the FCC to advantages in the nonregulated market. For and it may not be appropriate for the Commis- this reason FCC must approve the “compos- sion to study them. Other possibilities might ite allocator” developed by AT&T to allocate be the National Telecommunications and In- research costs among the various AT&T en- formation Administration (NTIA), which has tities. The economic theory is that a competi- an interest in the health of U.S. information tive firm should pay its own R&D costs with- R&D, or the National Science Foundation out shifting them to regulated ratepayers. (NSF), which monitors the state of R&D and basic research in a number of fields. Yet On the other hand, when AT&T was per- another possibility might be an independent mitted to use such a cross-funding arrange- research group outside of government—per- ment, it used the funds to create a highly haps one created by a university or industry respected and productive research organiza- association. Most of the technical and scien- tion that presumably benefited the Nation as tific journals needed for bibliometric studies a whole. If experience over the next few years are already in the database of the Library of shows that it is impossible for AT&T to main- Congress. These analyses might be performed tain Bell Labs’ quality without additional by the Congressional Research Service, an in- funding, and if it is determined to be in the dependent research group, or one of the agen- national interest that such an organization be cies mentioned above. maintained, then additional funds must be provided. They could come from a direct Fed- For some of the studies mentioned above eral subsidy or from some kind of cross-fund- data may only be available several years after ing. The former is not likely to be politically the actual research has been done, and in many acceptable; the later is increasingly complex cases meaningful conclusions can be drawn only after data for 5 or 10 years have been ana- as the long-distance telecommunication market becomes more competitive. By its own lyzed. If there is a reduction in basic research estimate, AT&T now provides only 69 percent at Bell Labs, the trend might have been under 56 way for several years before the data indicate of total long-distance capacity. AT&T is a change. By that time, it might be difficult rapidly losing the market power it once had to effect any correction in the trend. “AT&T, private communication, Apr. 30, 1984. 134 ● information Technology R&D: Critical Trends and Issues to control prices throughout the industry. If Foundation (for example), direct support of it were required to raise the price of long- basic research through direct Federal sub- distance calls to provide greater support for sidies, or tax incentives for industries that basic research, it would be placed at a com- engage in basic research. Such policies might petitive disadvantage with respect to other be applied not only to Bell Labs, of course, but long-distance carriers that do not support re- to Bellcore or to other university and indus- search. Development of a mechanism by which try research organizations. Over the next few all long distance carriers contribute to fund- years, while the health of research is being ing basic research would be difficult in the in- monitored, it might be possible to structure creasingly competitive long distance market. such programs to stimulate research and to develop “trigger” mechanisms for putting An alternative regulatory approach might them into place if results indicate that the be to stimulate basic research by allowing quality of research is declining. more cooperation between ATTIS and Bell Labs. About 4,000 former Bell Labs employ- In conclusion, it is still too early to tell ees were moved to ATTIS when it was created whether the quality or direction of research at in 1982. Expertise in some research areas has Bell Labs will be adversely affected by deregu- been lost to Bell Labs through this transfer lation and divestiture, or whether any changes and through the subsequent transfer of 3,000 in its research would have major repercussions employees to Bell Communications Research. for U.S. research as a whole. There are several The FCC’s interpretation of Computer II rules possible measures for monitoring the health do not allow the exchange of market and net- of basic research at Bell Labs, but evidence work information between ATTIS and Bell of change may not be apparent for several Labs and they also prohibit the joint develop- years. While Bell Labs is a major contributor ment of certain products, especially computer to the sciences related to information technol- software. ogy, it is not the only important player. To gain a true picture of the effects of deregula- In the future, easing this requirement to the tion and divestiture it would be worthwhile to extent of allowing ATTIS and Bell Labs to co- expand such studies to monitor the state of operate on certain types of research, might basic research throughout industry and at allow greater cross-fertilization among the two university labs as well. It will be important research organizations. to begin collecting information soon in order Policies to stimulate basic research gener- to fully document the transition from the pre- ally might include such incentives as addi- to post-deregulation environment. tional grant support from National Science

Chapter 4 References

“Bell Labs on the Brink, ” Science, Sept. 23, 1983, Dembart, Lee, “Dividing Bell Labs: Breakup to pp. 1267-1269. Put the Best to New Test,” Los Angeles Times, “Bell Labs: The Threatened Star of U.S. Re- Sept. 6, 1983, pp. 1,3. search, ” Fortune, July 5, 1982, pp. 46-52. Hall, Peter, “AT&T and the Great Divide, ” Finan- Brock, Gerald W., The Telecommunications Indus- cial World, Jan. 10, 1984, pp. XX. try: The Dynann”cs of Market Structure (Cam- Harris, Marilyn A., “Bell Labs Looks to Military bridge: Harvard University Press, 1981). Research,” Electrom”cs, Feb. 9, 1984, pp. “Competing with the New AT&T,” Venture, Jan- 102-104. UWY 1984, pp. 54-57. Kamien, Morton I., and Nancy Schwartz, Market Charles River Associates, Impacts of the AT&T Structure and Innovation (New York: Cam- Divestiture on Innovative Behavior, unpub- bridge University Press, 1982). lished paper prepared for Office of Technology Morkre, Morris E., “Innovation and Market Struc- Assessment, 1983. ture: A Survey, ” Working Paper No. 82. Fed- Ch. 4—Effects of Deregulation and Divestiture on Research • 135

eral Trade Commission, Bureau of Economics, Scherer, F. M., Industrial Market Structure and 1982. Economic Performance (New York: Rand Mc- National Science Board, Science Indicators, 1982 Nally, 1980). (Washington, DC: U.S. Government Printing Uttal, Bro, “Cold New World, ” Fortune, June 27, Office, 1983). 1983, pp. 81-84. National Science Foundation, Research and Devel- “Why AT&T Will Lose More Long Distance Busi- opment in Industry, 1981, NSF 83-325 (Wash- ness, ” Business Week, Feb. 13, 1984, pp. 102- ington, DC: U.S. Government Printing Office, 104, 110. 1983), p. 3. Wiegner, Kathleen K., “Prometheus is Unbound Scherer, F. M., “Firm Sizes, Market Structure, Op- and Seeking His Footing, ” Forbes, Mar. 12, portunity, and the Output of Patented Inven- 1984, PP. 141-148. tions, ” American Economics Review, 55:1119.

38-802 0 - 85 - 10 Chapter 5 Education and Human Resources for Research and Development Contents

Page Findings ...... 139 The Concern About Manpower ...... 139 The Relationship Between Manpower Development and Economic Growth ...... 140 Identifying Particular Manpower Problems and Solutions ...... 141 The Range of Manpower Predictions ...... 145 The Supply of Manpower...... 149 The Problems in Higher Education ...... 149 Elementary and Secondary Education ...... 154 The Federal Role in Manpower Development ...... 157 The Societal Context for Determining Federal Manpower Policy ...... 161 Chapter 5 References ...... 163

Tables Table No. Page 2A. Scientists and Engineers Engaged in R&D Per Labor Force Population, By Country: 1963-82 ...... 145 25. BLS Manpower Estimates...... 146

Figures FigureNo. Page 20. Employed Scientists and Engineers by Field, 1981 ...... 143 21. Distribution of Scientists and Engineers by Primary Work Activity, 1981 ...... 144 22. Percentage of Firms Reporting Available Jobs for Scientists and Engineers . . . . . 148 23. Comparison of Growth in Engineering Undergraduate Enrollment and Number of Faculty, 1973-80 ...... 150 24. Share of All S/Es Employed in Educational Institutions by Field, 1981 ...... 151 25. Educational Gifts by Computer Vendors ...... 153 26. Selected Indicators of the Overall Quality of Mathematics and Science Education in the United States...... 154 27. Americans Science Teacher Shortage ...... 156 28. Selected Indicators of Shifts in Undergraduate Science/Engineering Education . . . 157 29. A Quarter Century of Student Aid ...... 160 Chapter 5 Education and Human Resources for Research and Development

Findings

OTA found that major Federal actions de- tion technology, Federal action, taken now, signed to affect the supply of manpower to might prove to be inappropriate in the future. perform research and development (R&D) in A number of legislative proposals have been the area of information technology would ap- made that are designed to increase the future pear to be unwarranted at this time. Forecasts of future manpower needs in this area are re- supply of highly qualified scientific and tech- plete with uncertainty. Moreover, developing nical manpower. These proposals differ con- manpower for the specific areas where poten- siderably in terms of their goals, their targets, tial shortages might tentatively be predicted their costs, and their scopes. Given the high levels of uncertainty that surround the present would be particularly hard to accomplish through broad Federal actions. The educa- manpower debate and the number of competing uses to which the Nation’s limited educational backgrounds and skills required to meet tional resources might be profitably put, the these potential shortages are at once both too most prudent course might be to adopt those broad and too narrow to be developed at the Federal level. In addition, given the length of policies that would provide for the greatest time required to develop skills and the rapid amount of flexibility and the broadest range of skills. changes taking place in the area of informa-

The Concern About Manpower

In the United States today, there is a grow- To maintain its technological edge in world ing and widespread belief that the Nation’s markets the United States must reemphasize poor economic performance is inextricably science and engineering on our agenda of na- linked to the relative decline in the size and tional priorities. When the Soviets launched the quality of its technical work force. Noting Sputnik I, a remarkable engineering accom- that Japan and West Germany, our major in- plishment, the United States rose to the challenge with new dedication to science and ternational competitors, have four times as technology. Today, our technology lead is many electrical engineers and computer scien- again being challenged, not just by the Sovi- tists, per capita, as the United States, many et Union, but by Japan, West Germany, and of the people who hold this view fear that, as others. the economies of the developed world become more technologically intensive, and thus as The negative consequences of having a R&D becomes more critical to their success, shortage of manpower in information technol- the United States will increasingly lose its ogy R&D, it is argued, may be particularly ability to compete. Typical of this perspective severe. Given the speed with which the field is the statement made by Representative Mar- is changing, even a temporary shortage might garet Heckler during hearings on Engineering impair the ability of information technology Manpower Concerns, when she said:1 industries to remain at the frontiers of re- ‘Representative Margaret Heckler, Opening Statement, Hear- ings On Manpower Concerns, before the Committee on Science ucation, before the Committee on Education and Labor, House and Technology, House of Representatives, 97th Cong., 1st of Representatives, 98th Cong., 1st sess., Jan. 26-28, 31, 1983; sess., Oct. 6-7, 1981, p. 4. For a more recent statement of this see also, Amen”ct-i Competitive Challen~ Report of Business- perspective, see also, Hearings on Mathematics and %“ence Ed- Higher Education Forum, 1983. 139 140 . Information Technology R&D: Critical Trends and Issues

search. And, because the information technol- tion’s manpower resources, there has been ogy industry represents the fastest growth very little systematic effort to clearly identify sector of the economy, failure to keep pace in and characterize the nature and the extent of this industry may have serious consequences the problems. Before making any major poli- for the Nation’s economy as a whole. cy decisions designed to affect the supply of manpower, therefore, it will be necessary to Concerned about the state of available hu- have a greater understanding of: 1) what we man resources for high-technology jobs, spokes- know and don’t know about the relationship men from business, government, and educa- between manpower development and econom- tion have called on the Federal Government ic growth; 2) what we know and don’t know to undertake a number of significant educa- about this relationship as it relates in particu- tional measures and reforms. These measures lar to R&Din information technologies; 3) the range widely in terms of goals, targets, costs, range of projections about the future supply and scope. Some of them, for instance, focus of and demand for manpower in this field; 4) on a specific curriculum area, such as math the ability of the present institutional struc- and science; others emphasize educational ture to accommodate these manpower needs; infrastructure—the training of teachers and 5) the role of the Federal Government in the need for equipment; while still others seek the development of manpower; 6) the societal con- to foster new modes of cooperation between text in which, today, decisions about educa- business, government, and educational insti- tion will be made; and 7) the range of Federal tutions. Given our Nation’s limited resources alternative strategies and options for meeting and the growing number of demands and future manpower needs in the area of informa- stresses that are being placed on our education technology R&D. The following discus- tional system at all levels, choices and decision provides a preliminary basis for such an sions will have to be made about which goals understanding. to pursue and about which measures to adopt. Notwithstanding the widespread discussion and concern about the poor state of the Na-

The Relationship Between Manpower Development and Economic Growth

The assumption of a positive relationship be- Indeed, ever since the beginning of the in- tween the size and quality of a nation’s work dustrial revolution, economists and other so- force and its economic wealth is not a new one. cial observers have argued that a skilled and Over 100 years ago, for example, the British educated work force is the most productive. Government sponsored a parliamentary com- Writing as early as 1776, Adam Smith, for ex- mittee to investigate the causes of rapid in- ample, pointed out that “the skill, dexterity, dustrial growth in the United States. Like and judgment with which it’s [the nation’s] la- many of our recent studies of economic growth bor is generally applied,” is the primary fac- in Japan, the British parliamentary commit- tor determining the size of “the fund which tee attributed much of America’s industrial originally supplies it with all the necessities success to the superior education of the Amer- of life. ”3 2 ——— ican worker. ‘Adam Smith, The Wealth of iVati”ons (New York: The Mod- em Library, 1937), p. lviv. As modem societies became more technologically advanced, an increasing amount of attention was paid to the development of the labor force. Anticipating the ef- ‘Report Fmm the Select Committee on Su”enti”fi”c Instruction, fect that technology would have on society, the German Parliamentary Papers, 15, (1867-1868) Q 6722, as cited in Wil- sociologist, Max Weber, pointed out for example, that, liam Abernathy, Kim B. Clark, and Alan Kantrow, Industzz”al in an advanced industrial society, the organization of Remissanm: l%niucing a Competitive Future for Amen”ca (New human relations could no longer be left to chance. In- York: Basic Books, 1983). stead, human beings become factors of production-their Ch. 5—Education and Human Resources for Research and Development . 141

Present government policies designed to af- One of the central historical questions con- fect the supply of manpower are also based on cerning technical progress is its extreme var- this assumption. However, as the following iability over time and place. . . . Clearly the discussion illustrates, while we can identify reasons for these differences, which are not some general linkages between education, yet well understood, are tied in numerous manpower, and economic development, our un- complex and subtle ways to the functioning of the larger social systems, their institu- derstanding of causal relationships, or of relations, values, and incentive structures. The tionships in specific situations, is extremely explanation of these differences is intimately limited. tied to such even larger questions as why so- While acknowledging that having qualified cial change occurs and why economic growth manpower is critical to the success of a na- proceeds over time and place. tion’s economy, social and economic analysts Not only is each piece of the puzzle difficult are still unable to fully account for, or to com- to solve; the whole problem is subject to the pletely explain, the nature of the relationship vagaries of external events, well beyond our between the size and the skill level of the la- anticipation and calculation. Nor are the tools bor force and economic growth and develop- of analysis particularly refined. For—although ment. As the economist, Nathan Rosenberg, demographers may tell us something about has noted:4 population trends; sociologists something ———.—.. about the institutions and processes in which relationships w be structured in accordance with there individuals are recruited, educated, and trained quirements of industrial progress. And the American economist, Thorstin Veblen, writing in the 1930s, went for work; economists something about the so far in his discussions of technology and society as to point at which, and the rate of exchange by suggest that, for technology to develop to its full poten- which, the supply and the demand for labor tial, the technical expert-the engineer-wodd have to play a key role in society’s decisionmaking process. Jay are brought into a state of equilibrium-his- Weinstein, Sociolo@I’echnology: Foundations of Post torians are sure to remind us that it is, more Academz”c fi”ence, Transaction Books, 1982, p. 32; often than not, a unique set of circumstances Thorsten Veblen, The Theoxy of the Leisure Class (New York: The Modem Library, 1934). that has had the most significant effect on a 4Nathan Rosenberg, Inside the Black Box–Technology and particular outcome. Econorm”cs (Cambridge, MA: Cambridge University Press, 1982), p. 8.

Identifying Particular Manpower Problems and Solutions

Our limited knowledge of the role of human To identify future manpower needs in a par- resources in economic growth and technologi- ticular area, policymakers have traditionally cal change is clearly evident in our efforts to relied on economic and other forecasting meth- identify and analyze specific manpower prob- odologies. While useful as policymaking tools, lems and solutions. For although manpower these methodologies are subject to a number specialists might agree that having sufficient of problems and weaknesses which stem, qualified manpower is critical to a nation’s among other things, from imperfect data, economy, they do not necessarily agree about weak forecasting models, and ill-founded as- the number of people who are required to meet sumptions. 6 To be most useful, forecasting the employment needs of a particular sector; methods need to be flexible and responsive. about the kinds of skills and experience that Acknowledgment should be made of the limi- might be required to perform particular kinds tations of these methodologies, and efforts of jobs; or about the way in which these skills should be undertaken to verify their results might best be obtained or developed.5 by conducting frequent surveys and by performing case studies designed to determine 6National Institute of Education, Education, Productivity, and the National Economy, A Research Iru”tiative, December 1981; see also Edwin Mansfield, Education, R&D, and Produc- 6R. H. Bezdek, Long-Range Forecasting of Manpower Re- tiw”ty Growth, revised, University of Pennsylvania, Jan. 31, qu.z”rements (New York: Institute of Electrical and Electronics 1982. Engineers, 1974). 142 ● Information Technology R&D: Critical Trends and Issues the changing skill requirements that are linked demand of engineers and scientists in the to the emergence of new technologies. United States, or the Bureau of Labor Statis- tics, have begun to treat computer scientists It should be noted, moreover, that manpow- as a distinct group. NSF, for example, has only er forecasts can themselves produce a pendu- recently stopped labeling everyone who works lum effect, underminingg the validity of the pro- in a computer-related field as working in the jections. This effect results from both the long area of computer theory, a heading that was period of time that it takes for people to pre- itself a subcategory of mathematics. And, pare for afield of work, and from the fact that, even today, NSF does not list a department once committed to a career path, people rarely of computer science under that heading if the change their plans in midstream to adapt to department’s name appears in a combined new circumstances. Upon hearing predictions form and if the words “computer science” ap- of an impending manpower shortage in a par- 7 pear second in that combination. Even when ticular field, for example, an inordinate num- the appropriate statistics have been collected, ber of students may seek to pursue such a ca- moreover, they have often been subject to a reer, hoping that when they have finished their 8 variety of interpretations. educations, jobs will be plentiful and competition will be in short supply. The resulting man- It is also difficult to determine not only who power glut will appear only later; but predic- or how many people are working on or with tions of it may induce a number of students these technologies but also who or how many to avoid the field, leading to another shortage people are performing specific R&D tasks in in the future. this area. NSF estimates that in 1981,3.1 million scientists and engineers were employed It should also be remembered that man- in the United States. Of these, 47 percent were power predictions can be interpreted differ- employed as engineers (including engineers do ently by different kinds of people. Economists ing management jobs), and 13 percent were might describe a shortage, for example, when 9 working as computer specialists (see fig. 20). they see a rapid increase in wages due to a gap NSF reports, moreover, that 34 percent of all between the supply and demand for labor. scientists and engineers are involved in R&D Businessmen might consider that there is a 10 activities” (see fig. 21). As table 24 illustrates, shortage of manpower when they are dissat- compared to other countries, this is a high pro- isfied with the quality of preparedness of the portion of R&D scientists and engineers rela- pool of people from whom they have to select tive to the total labor force. Figures are not employees. New graduates may interpret a available, however, for the percentage of scien- manpower shortage to mean that they face lit- tists and engineers who specifically perform tle competition in seeking employment. R&D tasks in the area of information tech- The problem of predicting manpower needs nology. in the area of information technology R&D is Use of aggregated data based on broad skill even more complicated, because the field is categories or outmoded technologies reduces new and in a rapid state of flux. There is, for the value of manpower demand forecasts. A example, very little historical basis for iden- category such as computer programmer, for tifying who the people are who might typically example, is much too broad to use for forecast- perform R&D tasks in the area of information ing manpower demand in R&D. Further sub- technology; what skills they should possess in order to perform these tasks most effectively; or what their optimum career patterns might 7Kent K. Curtis, “Computer Manpower-Is There a Crisis?” be. (Washington, DC: National Science Foundation, January 1983). ‘Ibid. It is only very recently that either the Na- ‘Su”ence Inch”catms 1982: An Analysis of the State of U.S. Sa”ence, Engine, and Tdnology, National Science Board, tional Science Foundation (NSF), the key 1983, p. 63. agency mandated to monitor the supply and IOIbid., p. 66. Ch. 5—Education and Human Resources for Research and Development . 143

Figure 20.-Employed Scientists and Engineers by entailed in R&D and those entailed in produc- Field, 1981 tion are becoming increasingly blurred in this Mathematical area. This problem is clearly evident in the case of software engineering. Calculations of Physical future manpower needs that focus specifically scientists 8% on R&D activities are, therefore, particularly difficult to make in the area of information technologies. Questions also arise with respect to how the skilled technical workers, who provide support Engineers 47% to the R&D process but who do not perform the most highly skilled tasks, might best be factored into manpower projections. These workers might include, for example, all of those who maintain, troubleshoot, repair, and sometimes fabricate sophisticated equipment, those who build and help develop prototypes

Environmental of new products, draftsmen and nondegree designers, computer system operators, and tech- Psychologlsts 30/o nical writers. Since the skills that these work- 40/0 ers require may be more easily obtained or alncludes earth sclentlsts, oceanographers, and atmospheric scientists. may be more easily substituted—either by NOTE: The total number of scientwts and engineers m 1981 was 3.1 mdlion. other workers or by technology-than the SOURCE: Science Indicators, 1982, skills required for the more highly technical jobs, the manpower projections for this sec- division into types such as entry level, applica- tor of the R&D process, and the policy impli- tions, and systems programmers based on cations that might be drawn from them, may existing job specifications, while useful, will also be quite distinct. Policies designed to af- probably not suffice for long, since the mix of fect the availability of skilled technical work- computer-related skills required for R&D is es- 11 ers, for example, might call for some general pecially sensitive to technological innovation. educational changes at the elementary and/or Some of the most important skill categories, secondary levels whereas those that are de- for example, lie at the frontiers of information signed to influence the supply of manpower technology, and these do not show up in the in a highly technical area such as artificial in- broad categories based on aggregated data. telligence, or software engineering, might call The problem of identifying these skills, there- for targeted incentives at the university or fore, is not just one of substituting one set of graduate studies level. static descriptors for another; it is a problem of gathering information about new skills over Defining R&D manpower and estimating time and in response to changing conditions. manpower needs for this area becomes even more troublesome the further one looks into The task of identifying research and devel- the future. Over the long run the future de- opment workers in the field of information mand for manpower, for example, is likely to technology is complicated, moreover, by the depend on the extent to which and the speed fact that the traditional distinctions that have at which the new technologies are deployed always been made between the tasks that are throughout society. However, the rate and degree of their deployment will depend, in turn, ——. —— llAb~ hfowshowi~, “on Predicting R&D Skill [email protected] on the kinds of social variables that are most for Information Technology,” paper prepared for the Office of often left out of forecasting models, and that Technology Assessment, February 1984. are the most difficult to predict.

144 ● Information Technology R&D: Critical Trends and Issues

Figure 21 .–Distribution of Scientists and Engineers by Primary Work Activity, 1981

and computing, consulting, other, and

SOURCE: Science Indicators, Ch. 5—Education and Human Resources for Research and Development ● 145

Table 24.–Scientists and Engineersa Engaged in R&D Per Labor Force Population, By Country: 1963.82

Country— —. 1965 1968 1972 1975 1979 1982 S/Es a engaged in R&D per 10,000 labor force population France ...... 21.0 26.4 28.1 29.3 31.6 NA West Germany...... 22.7 26.2 36.0 41.0 47.7 NA Japan ...... 24.6 31.2 38.1 47.9 50.4 NA United Kingdom ...... 19.6 20.8 30.4 31.2 33.2 b NA United States...... 64.1 66.9 57.9 55.5 57.9 63.8 U.S.S.R. (lowest) ...... 44.8 53.5 66,5 78.2 84.4 89.8 U.S.S.R. (highest) ...... 48.2 58.8 73.2 87.5 95.5 102.4 S/Esa engaged in R&D (in thousands) France ...... 42.8 54.7 61.2 65.3 72.9 NA West Germany...... 61.0 68.0 96.0 103.9 122.0 NA Japan ...... 117.6 157.6 198.1 255.2 281.9 NA United Kingdom ...... 49.9 52,8 76.7 80.5 87.7 b NA United States...... 494.5 550.4 518.3 532.7 620.2 716.9 U.S.S.R. (lowest) ...... 521.8 650.8 862.5 1,061.8 1,216.4 1,340.4 U.S.S.R. (highest) ...... 561.4 715.2 950.1 1,187.6 1,377.4 1,340.4 Total Iabor force (in thousands) France ...... 20,381 20,744 21,817 22,310 23,059 NA West Germany ...... 26,887 25,968 26,655 25,323 25,573 NA Japan ...... 47,870 50,610 51,940 53,230 55,960 NA United Kingdom ...... 25,498 25,378 25,195 25,798 26,464 NA United States...... 77,178 82,272 89,483 95,955 107,050 112,383 U.S.S.R ...... 116,494 121,716 129,722 135,767 144,201 149,215 alncludes all scientists and engineers engaged in R&Don a full-time equivalent basis (except for Japan whose data include persons Primarily employed in R&D ex. ciudlng soc!al scientists, and the United Kingdom whose data include only the Government and industry sectors) blg78 NA—Not available. SOURCE National Science Foundallon,Sc~ertce /ndicators 1982

Future manpower requirements in software changes or their future impacts is extremely engineering, for example, might be significant- difficult, given the newnessofthe field and the ly reduced as new software tools are developed fact that they are dependent on a number of and introduced, if new institutional practices social variables-e.g., the willingness of indi- are adopted to improve the efficiency of those viduals and institutions to both adopt and working in the area, and if more and more ap- adapt to technological changes-variables placations software are developed on personal that are themselves notoriously unpredictable. computers by end users.12 Predicting these “Ibid.

The Range of Manpower Predictions

Given the problems involved in identifying shortages of Ph.D.s to teach at the university future manpower requirements, it is not sur- level in the fields of engineering and computer prising that there has been considerable conscience. To a somewhat lesser degree, man- troversy over and discrepancy between many power experts concur that the future growth of the projections that have been made about in demand for computer scientists will be ex- the need for high-technology manpower. Among traordinary. They disagree, however, about those making forecasts, the consensus has whether or not the educational system, as it been the greatest with respect to projected exists today, can effectively respond to meet 146 . Information Technology R&D: Critical Trends and Issues that demand. Agreement is lowest with regard moreover, tended to exaggerate the growth of to whether or not there will be a future short- technical occupations and underestimate the age of engineers. decline of certain traditional jobs. One useful measure of the accuracy of these projections The following summaryprovides some sense is the recent finding that 60 percent of the of the range of projections. Because these 1980 forecasts fell within a 10 percent range forecasts are based on different assumptions, of the actual employment level for that year.15 methodologies, baselines, and timeframes, it is impossible to compare and contrast them BLS projects the rate of growth and the fu- analytically as a whole. Evaluations about ture demand for manpower in given occupa- their reliability and accuracy are quite depen- tional categories. The Bureau’s most recent dent, therefore, on judgments about the va- projections for those categories most relevant lidity of their methods and assumptions. to information technology are listed in table 25.16 For each category, there are three pro- Generally speaking, however, it can be said jections—high, medium, and low—each corre- that manpower forecasts are more reliable the sponding to one of the three economic scenar- greater the number of and the more refined the ios used by BLS in developing their forecasts. underlying analysis. 13 The least sophisticated forecasting methodology, for example, might Because it does not make predictions about be one based solely on survey data, or simply the future supply of manpower, the Bureau of extrapolating trends on the basis of the present. Labor Statistics does not predict labor short- A much more comprehensive approach, on the ages or labor surpluses per se. However, re- other hand, might be one that takes into ac- view of the recent articles in the BLS pub- count such things as changes in the relative lication, Occupational Outlook Handbook, prices of capital and labor, and/or that posits suggests that there will be a multi-tiered mar- a set of alternative assumptions about the future. The most ambitious forecasts are those ‘51bid. ‘K!onversation with Tom Nardone, Manpower Economist, Bu- that try to factor into their analysis the im- reau of Labor Statistics, Department of Commerce, May 15, pact of technological change.14 1984. Perhaps the most widely referred to, and among the more sophisticated projections, are Table 25.—BLS Manpower Estimates those that have been put forward by the Bu- reau of Labor Statistics (BLS). These projec- Base year: 1982 1995 tions cover a period of 10 to 15 years. They Electrical and electronic engineers: are not only based on a set of alternative eco- 320,000 ...... LOW . 531,000 nomic scenarios, positing different rates of Moderate 528,000 High 540,000 growth; to some extent, they also seek to take Computer specialists: technological change into account. Moreover, Programmers: BLS has consistently sought to improve its 226,000 ...... LOW 465,000 Moderate 471,000 methodology by systematically evaluating the High 480,000 accuracy of its own projections. Such evalua- Systems analysts: tions show that BLS has had more success in 254,000 ...... LOW 469,000 Moderate 471,000 determining how technology might effect fu- High 480,000 ture job growth than it has in identifying at Technicians: what point such changes in employment pat- 55,000 ...... Low 106,000 Moderate 108,000 terns might take place. Past projections have, High 108,000 Computer operators: 211,000 ...... , . . Low 366,000 l~Henry M. hvin and Russell W. Rumberger, me Edu~ti”ozI- Moderates 371,000 al Implications of High Technology, The National Institute of High 378,000 Education Report #83-A4, 1983. SOURCE: Tom Nardone, Manpower Economist, Bureau of Labor Statistics, 141bid. Department of Commerce, personal communication, May 15, 19S4. Ch. 5—Education and Human Resources for Research and Development . 147 ket for these job categories, with a shortage ysis because they are major centers of defense of people with some specific skills and a sur- activity. The ROPES study found that the plus of those with others. need for all skill groups involved in “the use, operation and repair of computer equipment The National Science Foundation also makes will grow at an alarming rate throughout the projections of the supply and demand for sci- 1980s and that these skills will be particularly entific, engineering, and technical personnel. affected by increased expenditures for de- These projections are developed using a multi- fense. ” Although the Air Force study group step model, similar to the one used by BLS. did not project the supply of manpower for Like the BLS model, for example, the NSF these areas, they concluded, based on the pro- model tries to anticipate how technological jected rate of growth in demand, that the field change might affect future manpower needs. of computer science, and perhaps electrical and In one way, however, the NSF model goes fur- mechanical engineering, may be areas of po- ther than that of BLS: its alternative scenarios tential national shortage. posit different levels of defense spending as well as different levels of economic growth.17 A 1982 study by Betty M. Vetter of the Sci- entific Manpower Commission summarizes in- In its recent report, Science Indicators 1982, formation on the present and future supply NSF pointed out that computer specialists and utilization of scientists and engineers in accounted for almost 45 percent of the total the United States. Vetter concludes that, ex- growth of scientific employment over the pe- cept in the field of computer science, the sup- riod 1976-81. Matching this growth against ply of scientists appears sufficient to fill near- the future supply of computer science person- term demands. For engineers, she notes that nel, it predicted a future shortage in this area. the number of new graduates at the baccalau- Indicators for the future supply and demand reate level has been rising since 1975, but that of engineers were more mixed, however. For “a high level of demand has not only fueled while these indicators revealed a shortage in that increase, but has utilized so many engi- 1981, they also suggested that by mid to late neering graduates at the baccalaureate level 1982, the situation already appeared to be that graduate enrollments of U.S. students shifting back towards a balance between the have not climbed commensurately and short- supply of and the demand for engineers in ages of Ph.D. engineers have become serious, general. at least at academic institutions. ” She con- A recent survey conducted for NSF raises cludes, however, that there is no general agree- some questions about the degree to which ment about the adequacy of the future supply manpower shortages may in fact materialize of engineers, even when considering particu- in the future, even in the area of computer lar specialties. science. Nearly one half of the 351 firms sur- Industry projections of future manpower veyed, for example, reported fewer openings needs are quite inconsistent with one another. for scientists, engineers, and technicians dur- Derived, as they are, by stakeholders, their ing the 1982-83 recruiting year than in the conclusions must be regarded with some de- 1981-82 period. These results are broken down 18 gree of caution. by area in the following figure (fig. 22). Basing its conclusions on a survey of 815 Focusing on manpower needs for defense, manufacturing facilities, the American Elec- the U.S. Air Force, in The Regional Planning tronic Association (AEA) predicts that through Project, and Evaluations Systems (ROPES) 1987, the need for technical professionals will forecast manpower needs for 30 States and 70 grow by 69 percent; while the need for techni- major cities. The States were selected for anal- cal paraprofessionals will increase by 60 per- ‘7Levin and Rumberger, op. cit., pp. 12-13. cent. Respondents to their survey estimate, ‘“Ibid. moreover, that in 5 years they will need to

148 ● Information Technology R&D: Critical Trends and Issues

Figure 22. —Percentage of Firms Reporting Available Jobs for Scientists and Engineers Engineers Scientists (Percent of firms) (Percent of firms) o 20 40 60 100 0 20 40 60 100

Petroleum Computer science

Electronic Geology/geophysics

Electrical Chemistry

Systems analysis Computer

Physics Civil

Mathematics Chemical

Mechanical Biochemistry

Industrial Biology

Metallurgical

Manufacturing

SOURCE: National Science Foundation. employ more than 100,000 each of new tech- age of 113,406 computer scientists and elec- nical professionals and new technical para- trical engineers. Discounting employment re- professionals. l9 The occupational groups for lated to defense, the shortage would be 81,780. which they foresee a tremendous amount of More skeptical about the likelihood of an growth during this period—defined as over impending shortage of electrical engineers is 100 percent total increase—are software engi- David Lewis, Council Chairman for the Career neers, electronic engineering technologists, 20 Activities Council of the Institute of Electrical and computer analysts/programmers. and Electronic Engineers, Inc. (IEEE). Not- To project the future supply of these key oc- ing that the engineering profession is made up cupational groups, AEA uses data that as- of people who are trained in a range of dis- sume that U.S. colleges will continue to in- ciplines, projections of shortages, he says, fail crease the number of Bachelor of Science/ to take into account the extent to which elec- Computer Science degrees at the same rate as trical engineers can be substituted for by engi- they have over the past 5 years. Comparing neers trained in other specialties. He has ex- projected supply and demand, the AEA report pressed concern, moreover, that an uncritical concludes that by 1987 there will be a short- acceptance of such predictions might lead to a surplus of electrical engineers, a situation not dissimilar to the one that existed for aero- nautical engineers in the early 1970s.21 *UAEA, p. 10. Ch. 5—Education and Human Resources for Research and Development ● 149

The Supply of Manpower

Although manpower projections tell us Today, as we move towards what has been something about the number of people who characterized as a high-technology society, will be needed and who may appear to fill ex- businesses have themselves become involved, isting high-technology positions, they say very both formally and informally, in performing little about the quality of skills and experience educational tasks. This has been particularly that the people who are available might bring true in the area of information technology, to these jobs. To evaluate the quality of our where the larger corporations like IBM, Xerox, existing and future supply of manpower for and Digital Equipment Corp. have set up their R&D in information technologies, we have to own educational centers. Moreover, informal look at the major source of this manpower— training takes place and is diffused within the at the Nation’s educational institutions. business community as people, trained in large companies, move on to form new companies Formal educational institutions are, of of their own. course, neither the only nor necessarily the most significant institutional setting for man- Recognizing that a number of different power training and development. At one point kinds of institutions are presently involved in in history, for example, it was the family that the development and training of future man- dominated in preparing its members for eco- power, this chapter will nonetheless focus on nomic roles. Later, as society became more those that are a part of the formal educational technologically advanced, a somewhat more system. For it is chiefly within the context of formal system of apprenticeship emerged. For- these institutions that the Federal Govern- mal schooling became especially important ment plays out its role in manpower devel- during the age of industrialization. opment.

The Problems in Higher Education

While the American university system has is becoming a rapidly growing and wide-open always been renowned for the number of schol- high-technology market. ars and the amount and quality of research The difficulties are well illustrated in the that it has generated, today many people are case of engineering and computer science edu- beginning to question whether universities can cation, where a large proportion of faculty continue to effectively perform all of their positions are unfilled and where the number traditional roles. And, although almost all of Ph.D.s graduating each year has dropped areas of university education have suffered substantially. The problem in this area is not from the problems of increased educational one of attracting highly qualified undergrad- responsibilities and increased educational 22 uate students. Over the past decade under- costs, the problems that universities face ap- graduate enrollments in these areas have pear to be particularly acute in the areas that grown at a tremendous rate—by 80 percent in generate manpower to perform R&Din infor- the case of engineering29 and by 20 percent in mation technologies. University departments in these areas are having an especially diffi- ZZJeme MCDel-rnOtt, “’1’echnical Education: The Quiet Cri- cult time because, given their limited funding, sis,” High ‘1’echnology, November/December 1982, p. 87. they are finding it almost impossible to com- ZgJerrier A. Haddad, “Key Issues in U.S. Engineering Edu- pete for manpower and other resources in what cation, ” NAE Bridge, summer 1983, p. 11. 150 . Information Technology R&D: Critical Trends and Issues the area of computer science.24 And, given the ment. According to the American Council on growing popularity of electrical engineering Education, 1,583 teaching positions were va- and computer science and the limitations that, cant in the Nation’s 244 accredited depart- in almost all engineering schools and departments of engineering during the 1980-81 aca- ments of computer science, are now being demic year.26 placed on the number of admissions, the high The gap would probably be much greater, qualifications of new entrants are without 25 moreover, were it not for the sizable number precedent. of foreign engineers who teach in American Rather, as figure 23 illustrates, the problem colleges and universities. A recent survey of at universities-has been one of recruiting suf- engineering schools found, for example, that ficient faculty members to support this enroll- 25 percent of all junior faculty members in ——.— .— engineering received their bachelor’s degree ““AS Students Flock to Computer Science Courses, Colleges outside of the United States.27 ~r~ble TO Find Professors,” T..e Chronicle of Higher Edu- cation, Feb. 9, 1981. The shortage of faculty members in the field ‘5John Horgan, “Technology ’84 Education,” IEEE Spec- trum, January 1984. [Data on Admission Illustrations.] of engineering and computer science has been attributed to the fact that industry, by offering higher salaries and other, nonmonetary in- Figure 23.—Comparison of Growth in Engineering centives, has been able to draw a number of Undergraduate Enrollment and Number academics and students away from universi- of Faculty, 1973-80 ties.28 The extent of the problem is illustrated 100 by figure 24, which shows that, in contrast to other areas of science, only a relatively small 90 proportion of the Nation’s engineers and computer scientists are employed in academia.29

80 Discrepancies between the salaries earned by scientists working in industry and academia have, in fact, been quite extensive. It has not been atypical, for example, for an inexperienced electrical engineer with a bachelor’s degree to earn more than an assistant professor of engineering with a Ph.D.30

*g”As Students Flock to Computer Sciences Courses,” op. cit. 20 ““AS Students Flock to bmputer Science Courses,” op. cit. *a’’ Supply of Engineering Faculty,” Electronz”c Market Trends, January 1982, p. 14: 10 It should be noted in this regard that the continued supply of foreign faculty will depend to some extent on the fate of The Immigration Reform and Control Act, 0 a bill that was recently passed by the Senate and that 1973 1974 1975 1976 1977 1978 1979 1980 would require fonigners to return home after graduation Year (fall term) for at least 2 years. An amendment maybe attached to NOTE: Faculty numbers drawn from ASEE’s Research and Graduate Study the bill, however, allowing certain students studying in directory were adjusted on the basis of enrollment data to include the high-technology fields to stay. same number of schools as in EMC’S enrollment survey. ‘eScience lti”cators, op. cit., p. 123. SOURCE: Engineering Education, November 1982. ‘McDermott, op. cit., p. 47. Ch. 5—Education and Human Resources for Research and Development • 151

Figure 24.—Share of All S/Es Employed in quired to perform R&D at the leading edge of Educational Institutions by Field, 1981 the field.33 (Percent) o 10 20 30 40 50 Other institutional factors that have been I I I I I I I 1 I cited as reasons for the exodus from academia All S/E . to industry include uncertainty of tenure, fields heavy teaching loads, inadequate funds and Mathematical scientists I institutional support for research, and increases in educational fees compounded by di- Life scientists minishing financial aid for students. Social scientists It is difficult to assess the extent to which these problems will persist in the future. The Physical scientists number of students studying for Ph.D.s in Environmental engineering significantly decreased over the scientists course of the decade 1972-82. Educators of en- Computer gineering suggest that, as a result, there will specialists not be enough new faculty members to replace 34 Engineers I even those who die or retire. If this kind of I I trend continues, there will probably be a fac-

SOURCE: Science Indicators, 1982. ulty shortage in the future. However, there are some indicators that point to a reversal of this university campuses to industry is the poor trend. In 1982, for example, the number of en- condition of most university research facilities. gineers earning doctorates increased for the In an effort to reduce costs, for example, many first time in 8 years. 35 In 1983, the number in- colleges and universities have failed to pur- creased again—from 2,888 to 3,023.3’ chase, maintain, and upgrade their buildings On the other hand, the academic manpower and equipment. As a result of such decisions, problem may be more difficult to overcome in university instrumentation inventories are the field of computer science where the per- now nearly twice as old as those of leading 31 centage of faculty leaving for industry is two commercial labs. The cost of adequately im- 37 proving these facilities in the area of engineer- times that of any other field of engineering. ing education alone has been estimated to be, Even in this area, however, there is some anecdotal evidence reported by NSF to suggest at a minimum, between $1¼ billion to $2 bil- 32 that the number of graduate students study- lion. This cost, moreover, is rapidly escalat- 38 ing with inflation. ing in this area is now increasing. It is possible, however, that faculty short- Problems of university instrumentation are ages could become even more critical in the particularly serious in those fields where the future in some specific areas, limiting the cost of equipment is especially high and where amount of research and the amount of teach- it plays an essential, if not an integral, part ing that can be done in these fields. This is par- of the educational and research processes ticularly true, for example, in an area such as themselves. This is the case, for example, in 39 the area of artificial intelligence. Only very few artificial intelligence (AI). Because the field universities can afford the cost or have the 8WTA Case Study on Artificial Intelligence. space and facilities available to house and sup- MNAE Bridgq op. cit., p. 121. sb~”aa Indz”catirs, op. cit., p. 123. port the kinds of sophisticated equipment re- ‘Manpower Comments, vol. 21, No. 3, April 1984, p. 24. sTC@i9, Op. cit., p. 8. 81John Brademas, “Gradua~ Education: Signs of Trouble,” ‘Conversation with Kent Curtis, National Science Foun- Sa”ence, vol. 223, Mar. 2, 1984, p. 881. dation. ‘zHaddad, op. cit. S90TA ewe study on Artifi”a”al InW”gence.

38-802 0 - 85 _ 11 152 . Information Technology R&D: Critical Trends and Issues

of AI is so specialized and because the size of account the changing style of doing engineer- the research community is so small to begin ing and computer science. To be effective, for with, the number of qualified faculty members example, the engineers and computer scien- cannot be multiplied rapidly enough to meet tists of the future may have to be adept in the rising number of students who are now be- communications as well as in technical skills.43 ginning to enter the field. The problem is likely As one young computer scientist described it, to worsen if, as might reasonably be expected, The old stereotype of an engineer was a the greater commercialization of AI applica- guy with a slide rule hooked to his belt and tions together with enhanced military inter- who was not into sports, not into team play. est in the field lead to increased competition The jobs that we have to do now are not for those trained in artificial intelligence in the 1-man year projects but are 5- to 10- to 20- future. The effect of a faculty shortage may man-year projects. Integrated circuits, even be particularly acute in an area such as this linear ICs, are so complex that they are no where the required skills and knowledge are longer designed by one engineer, but by best taught in an apprenticeship-type sit- teams. . . this concept may be new to us, but uation. it’s not to the Japanese. To alleviate the present faculty shortage, The mounting problems that engineering more than half of all engineering colleges have schools are facing would appear to be having a had to eliminate courses, and two-thirds are negative effect on the quality of education increasingly substituting graduate assistants that they provide. In the last 3 years, for ex- for full professors for some coursework, a sit- ample, 30 percent fewer engineering schools uation that could adversely affect the teaching were given full 6-year accreditation. Moreover, of advanced technical courses. A number of there has been an increase of over 70 percent in the number of institutions that have been universities have also increased the teaching 40 asked to show cause why their accreditation load. It has been estimated, for example, that should not revoked. 44 university offi- over the past 10 years the teaching burden of be Many cials agree with this assessment. In a survey the average professor of engineering has increased by 40 percent.41 While such actions of engineering colleges conducted by the Amer- of may ameliorate the problem in the short term, ican Council Education, 49 percent of the in the long term they may—to the extent that respondents reported that the quality of education provided by their institutions had ei- they discourage people from pursuing teach- 45 ing careers-actually exacerbate it. ther greatly or moderately declined. Concerns about the decline in the quality of engineer- Compounding their problems of loss of facing education are also echoed in the industrial ulty and deterioration of equipment, schools community, where a number of companies of engineering and departments of computer have independently taken steps to improve the science will also have to find ways to modern- pool of engineering manpower.46 ize their curricula to meet the educational needs of a high-technology society undergo- The current situation is not, however, set in ing rapid technological change. It has been stone. To the contrary, the American educa- estimated, for example, that given the speed tional system is today in the midst of consid- at which technological change is occurring, erable change. In adopting policies to affect today’s engineering education, acquired over the future supply of manpower, therefore, pol- the course of 4 or 5 years, will be obsolete in icymakers should be reminded of the “pendu- about the same amount of time.42 Up-to-date lum effect.” In making decisions about the fu- coursework, moreover, will have to take into 4sIbid,, pp. 127-128. 44McDermott, op. cit., p. 90. @McDemot,t,, op. Cit., P. ‘- 4’High Technology Manpower in the West: Strategies for Ac- 41~nfowor~d, May 30, 1983, P. 32. tion, op. cit., p. 12. 4“’The Changing Face of Engineering,” Electronics, May 31, *For a discussion of many of these efforts see Chapter 6: New 1983, p. 127. Roles for Universities in Information Technology R&D. Ch. 5—Education and Human Resources for Research and Development ● 153 ture, they need to take into account not only Figure 25.—Educational Gifts by Computer Vendors the projections of future manpower needs but IBM $50 million in cash and equipmen to 20 advance research in CADICAM. also how present reactions to those projections • Co-donation of $50 million in equipment to MIT may, in fact, undermine their validity. to research data transfer. ● $15 million pledge to Brown University to establish institute for research in information Concerned about the present state of high- and scholarship. technology manpower training in the United ● $2.4 million in graduation fellowships to science and engineering students. States, a number of businesses and corpora- Digital • $ co-donation of $50 million in equipment to MIT. tions, for example, have already taken it upon Equipment - themselves to improve the situation. In par- Corp. • $1.6 million to Boston University to fund new compute science programs. ticular, they have adopted a number of meas- ● Total of $45 million in fiscal 1982 to higher ures that are designed to keep faculty in aca- education. demia and to encourage students to pursue Apple ● $21 million in equipment to California schools computer, grades K-12 for “Kids Can’t Wait” Program. academic careers. Bell Labs has recently be- Inc. Company wants a Similar nationwide pro- gun a program that, over time, will provide $2 gram, but wants Federal tax deductibility first via so-called “Apple Bill" million to doctoral candidates studying in ● $500,000 to Brown University in form of 5O fields related to telecommunications. The Gen- Lisa Computer. eral Electric Co. has donated $2 million to Hewiett- ● Approximately $22 million, mostly in equipment, Packard to universities in fiscal 1983. engineering schools to be used primarily by co. teachers and Ph.D. candidates. IBM has es- Wang ● Total of $3.7 million in equipment and $458,000 tablished 150 graduate and postdoctorate fel- Laboratories, in cash to 23 universities and secondary lowships in engineering, computer science, and inc. schools in 1982. NCR • $140,000 in equipment to Michigan State information systems. And Hewlett-Packard Corp. University. recently initiated a student loan forgiveness ● $170,000 in equipment to Cornell University. • Several gifts of companies NCR systems to program that allows students to write off a other universities. part of their loans in accordance with a prear- Honeywell, ● S220,000 to Arizona State University. ranged schedule for each year that they teach. Inc. ● $30,000 to United Negro College Fund. ● Total of $3 million in education contributions in 1982. Companies are also providing funds for SOURCE: Compurerworld, July 25, 1983. equipment and instrumentation. Digital Equip- ment Corp. and IBM together have donated The States have also taken steps to improve $50 million for equipment to MIT. Sizable con- and foster the development of manpower for tributions have also been made to a number a knowledge-intensive society. Not only have of other universities by IBM, Hewlett-Pack- they joined in cooperative ventures between ard, Apple Computer Inc., Wang Laborator- industry and the universities, they have also ies, Inc., NCR Corp., and Honeywell, Inc. (see taken steps to enhance the research environ- fig. 25). ment and to improve faculty salaries. In addition, businesses and corporations are, While most State and corporate efforts to more and more, joining with universities in improve manpower training have been focused cooperative ventures to overcome perceived on the most prestigious universities, there shortages of scientific and technical manpow- have been a number that have sought to im- er. The goal of training manpower, for exam- prove the education and training provided to ple, was the major reason for establishing the less advantaged colleges and universities. Ex- Microelectronics and Information Sciences xon, for example, recently donated $1.8 mil- Center at the University of Minnesota, and the lion to six predominantly black engineering most important inducement in gaining indus- colleges, to be spent during the course of the try support. These cooperative ventures, as next 3 years. IBM has lent several of its em- well as a number of others, are discussed in ployees to teach in schools that serve “disad- some detail below, in chapter 6. vantaged students. ”

154 • Information Technology R&D: Critical Trends and Issues

Elementary and Secondary Education

To determine whether or not there will be Figure 26.— Selected Indicators of the Overall enough highly qualified people available to Quality of Mathematics and Science Education perform R&D in information technology, one in the United States Scholastic Aptitude Test score averages from must look not only at institutions of higher 1963 to 1981 education, but also at elementary and second- 500 i 1 ary schools. Generating the pool of students from which university and graduate students 480 are drawn, these institutions have an effect on both the number and the quality of students 460

who desire to pursue fields of study that might 440 lead to work in this area. 420 A number of reports have been released recently that raise serious questions about the ability of elementary and secondary schools 0 in the United States to adequately prepare the 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 Nation’s youth to work in a knowledge-inten- Years sive society. While differing in the focus of their concerns, all of these reports find that Proportion of 1980 high school leavers who have taken our schools are inadequately preparing Amer- various courses ican students for their futures. As evidence, they point to declining test scores, the need Algebra I for colleges and business to provide remedial Algebra II education and training programs, the level of Geometry functional illiteracy among the general popula- Trigonometry tion and the Nation’s poor showing in in- Calculus ternational comparisons of student achieve- 47 ment. Physics Of particular concern in all of these studies Chemistry is the overall poor quality of math and science o 10 20 30 40 50 60 70 80 education. A number of indicators give cause for such concern (see fig. 26). Enrollments in Percent math and science courses, for example, are Average scores on science tests among school children around generally low. Only one-third of all U.S. high the world (1973) school graduates have completed 3 years of mathematics, and less than 8 percent have England completed a calculus course.48 Moreover, in grades kindergarten through six, students Scotland spent only approximately 20 minutes per day on science lessons. And by 10th grade, only Germany about half of all students study any science Japan

B. Education in Reports on USA and Recommendations for Change, Issue Brief # IB83106 Congressional Service, Library of Congress, NOV. 17, 1983. o 10 20 30 40 50 60 “High School Science Problems Gain Spot- light,” & News, May 24, 1982, p. 39. Percent of maximum score Ch. 5—Education and Human Resources for Research and Development . 155 at all.49 In contrast, students in many other al Science Teachers Association, since 1971 industrialized countries spend about three there has been a dramatic decline in the num- times more class time on math and science ber of people training to teach in these areas— than the most scientifically oriented Ameri- 79 percent in the case of math and 64 percent can student.50 in the case of science (see fig. 27). Student attitudes about science and math The lack of new entrants into the teaching also reflect problems in the education of these profession is not surprising insofar as there courses at the elementary and secondary are few incentives to draw well-qualified indi- school levels. Science, for example, is the sub- viduals into the field.56 With the financial prob- ject that is preferred the least by most Amer- lems facing schools today, a career in teaching ican students.51 Moreover, although most chil- is no longer considered to be secure. And it is dren regard scientists themselves as being considered by many to be less likely to pro- intelligent and dedicated, they consider the vide the rewards of public status or personal work that scientists do as being “boring, dull, esteem. Since teaching salaries are by no and monotonous. ”52 Mathematics is not much means competitive with those in private indus- more popular. While it is the favorite subject try, many of the most qualified teachers— of 45 percent of students in the third grade, especially those with training in math and it is rated tops by only 18 percent of those in science—are giving up teaching to sell their the 12th.53 skills on the open market. Moreover, well- qualified women–a traditional source of high- Overall, student performance in these areas quality educators—are particularly likely to reflects the lack of interest and low enrollment seek out the wider range of employment op- in these subjects. In the three studies con- 57 portunities now available to them. ducted by the National Assessment of Educa- tional Progress, which were designed to assess The actual extent of the present shortage of the achievements of precollege students in a qualified math and science teachers is a mat- number of areas over the years 1969, 1973, and ter of some debate, however. A survey of ed- 1977, all age groups showed significant aver- ucational placement offices conducted by the age declines on mathematical applications National Science Teachers Association found, which involve the use of mathematical knowl- for example, that no less than one-half of the edge, skills, and understanding to solve prob- Nation’s math and science teachers were hired lems.54 In the area of science, the results were on an emergency basis.58 The Association for similar. There was a downward trend for all School, College and University Staffing has age groups, from the first to the second assess- reported, moreover, that there is a consider- ment, although this decline appeared to be di- able nationwide shortage of teachers in the minishing for 9 and 13 year olds in the third area of mathematics, physics, and computer assessment.55 programming on the other hand, a recent study conducted by the General Accounting As in the case of higher education, many of Office concluded that the gaps in the available the problems in precollege math and science education have been attributed to the short- — . .——. ‘The New Sa”entist, July 14, 1983. age of high-quality teachers to teach in these In explaining the shortage of teachers, much of the at- areas. According to the Director of the Nation- tention has focused on salary differentials between teaching and industry. While salary levels are no doubt important, one recent study found that internal morale “Herbert J. Walberg, “Scientific Literacy and Economic pro factors, such as the potential for personal growth, are ductivity in Intematimal Perspective, ” Dzmdalus, p. 12. even more so. Edward B. Fiske “Teacher Fulfillment Put ‘OHeylin, op. cit. The New York Times, 4, 1983, p. Cl. ] Above Pay, ” Oct. * Walberg, op. cit. Amer- 5$ “J. Myron Atkin, “Who Will Teach in High School?” Heylin, op. cit.. ica Schools: Pubh”c and Pn”vate, Daedalus, summer 1981. “Walberg, op. cit. 68 The New Scjentjst, Op. cit. “Sa”ence Indicators, op. cit., p. 74. W’eacher SupplyDemand 1984, Association for School, Col- 551bid. lege, and University Staffing. 156 ● Information Technology R&D: Critical Trends and Issues

Figure 27.—America% Science Teacher Shortage A 1982 survey reveals just how many new science and mathematics achieving the appropriate qualifications. Another important factor is teachers in the U.S. do not hold the appropriate qualifications to do the far more attractive salaries offered to graduates in the sciences, their jobs. This is partly a result of a fall in the numbers of those whether qualified teachers or not. - 840/0 Teachers 460/o 43 ”/0 (all disciplines) 3 Those entering government or Men industry, by Women discipline: Humanities

Social sciences Pacific Mountain 63%— 40% Biological sciences South Atlantic . Agricultural sciences East South Accounting Central Health professions

Chemistry 1971 base 1-0 Mathematics 0-75 Computer sciences

0-50 Earth sciences

0-25 Engineering

0 1 1 I 1 I 1 I 1971 1973 1975 1977 1979 1971 1973 1975 1977 1979 12 14 16 18 20 22 24 26 28 Thousand dollars - Qualified teachers available for employment _ Accepting teaching positions Teachers taken on in 1982 to teach science and math not qualified tween 1971 and 1980 (above left). Starting salarlis for new graduates in those subjects (top left). Decline in number of new teachers be- entering careers in government or Industry in 1982 (above r!ght). SOURCE: New Scientist, July 14, 1963. information are so great that whether or not among those students who are most likely to there are shortages of math and science teachers, pursue careers requiring a strong background and whether or not the quality of technical in math and Science. 61University math and sci- teaching has declined in recent years, cannot ence departments and schools of engineering be determined.60 report, moreover, that they are continuing to recruit high-quality students, many of whom Since a strong foundation in math and sci- are being drawn from the humanities depart- ence is, in most cases, essential to a future ca- ments with the idea of having better future reer in information technology R&D, these job prospects62 (see, e.g., fig. 28). On the other findings raise a number of questions about the hand, given the diminishing supply of high- future supply of manpower in this area. Their quality educators in these fields and the answer, however, is not particularly clear or predicted decline in the school age popula- straightforward. For although the data show that the level of math and science understanding and competency for the student popula- el~”ena In~”ca~rs, Op. cit., P. 77” tion as a whole has declined, it has not declined ufiti J. A-, stud~t Quali”tyin the ti”ences and Ew”- neering Opku”on of Sam”or Academi”c Offi”a”als, Higher Educa- 60New D&@”ons for Federal Programs to Aid Mathematics tion Panel Report No. 58, American Council on Education, Feb- and Su”ence Teachi”ng, GAOIPEMD-84-5, Mar. 6, 1984. ruary 1984. Ch. 5—Education and Human Resources for Research and Development ● 157

Figure 28.—Selected Indicators of Shifts in Undergraduate Science/Engineering Education

Chart 1. Perceptions of academic officials about change over Chart 2. Opinions of academic officials about shift in distribution last five years In quality of undergraduate science/engineering of most able undergraduate students between science/engineering (S/E) students, 1982 (S/E) and other fields over the last five years, 1982

Percent Percent o 10 20 30 40 50-- 60-- 7(-I o 10 20 30 40 50 60 70 80 Type of 1 I I I I I Type of Institution institution No change NO Signifcant change I

All institutions

Shift away S/E trends All institutions significant improvernent

Doctorate Doctorate granting granting

Nondoctorate granting granting

Publicly Publlcly control led J controlled

Privately Privately controlled control led I

100 largest S/E 100 largest S/E baccalaureate baccalaureate . I

SOURCE: National Science Foundation and American Council on Education, tion,63 a more serious problem may emerge in ondary education is not static and changes are the future. taking place that might affect the long-term Just as the situation in the case of higher supply of scientific and technical manpower. education, the situation in elementary and sec- The public’s response to the numerous reports — on the condition of American education has WI’he number of Americans graduating from high school is been immediate and widespread, involving projected to drop from a peak of 3.2 million of 1977 to a low of 2.3 million in 1992, following a roller coaster pattern through parents, teachers, businessmen, and govern- the end of the century. High Schcnd Graduates: Projtwti”ons for ment officials at all levels. the Fifty States (1982-2000), Western Interstate Commission for Higher Education, Boulder, CO.

The Federal Role in Manpower Development

As the largest employer of scientists and en- Given the Constitutional limitations on the gineers, the Federal Government has a direct Federal Government’s role in education, the interest in whether or not the educational sys- responsibility for developing manpower has tem can provide an adequate supply of tech- always been shared by a number of different nically trained people. Its role in the develop- social institutions ranging from the family to ment of manpower, however, has traditionally the business community. As American society been indirect. For it is the States, and not the has become more technologically advanced, Federal Government, that have had the pri- however, the Federal Government has been in- mary responsibility for educational policies. creasingly called onto play a more significant 158 ● Information Technology R&D: Critical Trends and Issues role. Pressure on it to be more active in this came to believe that special technical knowl- area is particularly strong today, as the Na- edge was the key to prosperity in the modern tion seeks to maintain its place in a highly age, secondary educational institutions were technical and competitive world environment. restructured to prepare American youth for an increasingly differentiated set of economic While conscious of the economic benefits as- roles. Not only were vocational courses added sociated with having a skilled labor force, to the educational curriculum, but the schools Americans did not originally adopt a formal themselves were remodeled to conform to the system for transmitting vocational and tech- prevailing business standards of efficiency. nical skills, in the earliest years of their his- The business community played a major role tory, when agriculture was the dominant mode in bringing about these changes. Concerned 64 Instead, most formal educa- of production. about strikes, labor turnover, and increasing tional institutions were designed to serve gen- worker absenteeism, they hoped that school- eral social and political functions, while gen- ing would socialize a growing number of im- eral vocational skills were left to be passed on migrant youths for the workplace.68 69 more or less informally, by family members or through apprenticeship systems.65 The growing enthusiasm for scientific and technical knowledge also had an impact on the In particular, there was little effort given to, nature of higher education. Accustomed to or even much concern for, the development of training gentlemen as preachers, lawyers, and scientifically trained manpower. When scien- doctors, the universities began to expand their tists were needed in the United States, they roles and to train people in the more vocational were brought over from Europe. Thus, it was 70 applications of education. Efforts to move in only in 1902 that the Federal Government first this direction met with considerable resistance became involved in the development of scien- from traditional academics, however, who tific manpower. Concerned at this time about were disdainful of the study of experimental the security dangers entailed in relying almost science and even more so of the teaching of exclusively on foreigners for scientific exper- 71 72 the “useful arts.” tise, Congress established the Army Corps of Engineers at West Point.66 The transformation of the university system was greatly facilitated by the passage of Fed- It was only with the rapid industrialization eral legislation establishing the land grant col- of society, at the end of the century, that edu- leges. Provided for under the Merrill Act of cation came to be really valued in economic 73 67 1862, these colleges, open to children of all and technical terms. And then, as Americans backgrounds, were established to provide education in practical fields such as agriculture, engineering, home economics, and business

‘For a discussion of American education in the preindustrial ‘David Tyack and Elizabeth Hansot, “Conflict and Consen- period, see Bernard Bailyn, Education in the Fornu”ng of Amer- sus in American Education, ” ti”a /Wwls: Public and Pri- ican Sou”ety (New York: W. W. North, 1980); Lawrence Cremin, vate, Daedalus, summer 1981. TradI”tiorIs in Amen”can Education (New York: Basic Books, ‘sIbid. Harper, 1976); and Rush Welter, Popular Education and ~OEmest L. Boyer ~d Fr~ Heckinger, Higher Education in Democratic Thought in Amerz”ca (New York: Columbia Univer- the Nati”on’s Serw”ce (Washington, DC: Carnegie Foundation sity Press, 1962). for the Advancement of Teaching, 1981); and Edward Shils, ‘A. Hunter Dupree, Sci”ence in the F’dmd Governm ent: His- “The Order of Learning in the United States From 1865-1920,” toxy of Pd”a”es and Actk”tiee to 1940 (Cambridge, MA: Belknap Minerva, vol. 21, No. 2, summer 1978. Press of Harvard University Press, 1957). nDa~d Noble, Am~”ca by Detu”gn: &i”eXlce, T~OIOKY, ‘d sdlbid. Initi~ly trained in the s~s neCeSSSry b cm out the Rise of Corporate Capitalism (New York: Alfred A. Knopf, explorations of the West and to conduct surveys of the east- 1977), p. 24. ern coastline, these West Point graduates played a key role in T2Boyer ad HW~@r, op. cit..; ~d ShilS, oP. cit. many of the scientific ventures carried out by the Federal Gov- 7~his law provided land to the States, the proceeds of which ernment up until the end of the 19th century. were to be used to teach in the fields of agriculture and me o~David K. Cohen and Barbara Newfeld, “The Failure of High chanical arts. Subsequent legislation provided Federal finan- Schools and the Progress of Education,” Amen”ca’s Schools: cial support for research and the operation of the landgrant Public and Pn”vate, Daedalus, spring 1981. colleges. Ch. 5—Education and Human Resources for Research and Development ● 159

administration. Through their agriculture ex- organizational mission, was explained by Van- periment stations and their service bureaus, nevar Bush in Science—the Endless Frontier, their activities were designed to serve the his report to the President on a program for state .74 postwar scientific research. About the need for 79 The impact of the Merrill Act on develop- scientific manpower, he said, ment of scientific and technical manpower is . . . Today, it is truer than ever that basic re- clearly evident in the case of the engineering search is the pacemaker of technological profession. Before its passage, State legisla- progress. In the 19th century, Yankee me- tures had been reluctant to invest in techni- chanical ingenuity, building largely on the ba- cal education. Responding to the offer of Fed- sic discoveries of European scientists, could eral grants, however, they quickly sought to greatly advance the technical arts. Now the situation is different. establish the new types of schools, while private colleges, caught up in the movement, also A nation which depends on others for its established departments of engineering.75 new basic scientific knowledge will be slow Schools of engineering expanded rapidly, in its industrial progress and weak in its competitive position in world trade, regardless of thereafter, numbering 110 by 1886. The num- its mechanical skill. ber of engineering students similarly increased, from 1,000 in 1890 to 10,000 in 1900.7’ As Provoked by the successful launching of the more and more engineers were educated in for- Soviet spacecraft Sputnik, defense considera- mal institutions, there was a greater empha- tions also motivated the passage of the Na- sis in engineering on science. Moreover, with tional Defense Education Act of 1958 (NDEA), the establishment and growth of these insti- which was aimed at improving instruction in tutions, a profession was developed and with mathematics, science, and foreign languages. it a means of preserving, transmitting, and Under this law, funds were provided on a increasing an evolving body of engineering matching basis to public schools, and as long- knowledge. 77 term loans to private institutions, for needed equipment in these instructional fields, for cur- The real impetus for manpower develop- riculum development, for guidance counseling, ment, and for a strong Federal role in it, came, for vocational education in defense-related however, after World War II, when advanced fields, and for teacher training in foreign lan- technology had proven to be critical not only guage instruction. The passage of NDEA for the Nation’s economic growth, but also for resulted in substantial increases in Federal aid its defense. It was in recognition of this fact, to education (see fig. 29). Since Federal dollars for example, that the National Science Foun- had to be matched by State and local funds dation was created in 1950 with the task of under provisions of the act, the overall invest- improving the Nation’s potential in scientific 78 ment in NDEA programs was large. Between research and in science education. 1958 and 1961, $163.2 million in Federal funds The philosophical basis for establishing were dispersed. Approximately 75 percent of NSF, and the rationale for including the de- these funds were directed to the development velopment of scientific manpower within its of science curricula. “Clark Kerr, The Uses of the Um”versity (Cambridge, MA: Although the U.S. educational system is or- Harvard University Press, 1972). ganized on a local basis, the Federal Govern- 76 Noble, op. cit., pp. 38+9. “Edwin T. Layton, Jr., The Revolt of the En@”neers: %cial ment has, over the years and in response to R.esponsibih”ty and the Amen”can Engineering Pmfesm”on (Cleve changing technological developments, come to land, OH: The Press of Case Western Reserve University, 1971). influence the development of scientific and 771bid. ‘oThe Natjon~ ~ience Foundation and Pre-COflege Science technical manpower in a number of ways. Par- Education: ~950-1975, report prepared for the Subcommittee ticularly in the area of science and math- on Science and Technology, U.S. House of Representatives, 94th ematics, the Federal Government has, for Cong., 2d sess., by the Congressional Research Setice, Library of Congress, January 1976. 7e1bid., p. 19.

160 ● Information Technology R&D: Critical Trends and Issues

Figure 29.—A Quarter Century of Student Aid

25 years ago, the National Defense Education Act was signed $6,718 Student loans NDEA, 1958 1983 National Defense Student Loans, for The program, renamed National Direct which about $30 million was appropri- Student Loans and transferred to the ated in the first year, were provided at an Higher Education Act, provided loans at interest rate of 3 percent. The law told an interest rate of 7 percent. About $193 colleges to give “special consideration” million was appropriated for fiscal 1983. to outstanding students who planned to The debts are forgiven for borrowers who become teachers in elementary and become teachers in schools serving dis- secondary schools and to those with a advantaged students and those who teach background of high achievement in the handicapped. science, mathematics, engineering, or a

International education NDEA, 1958 1983 The law authorized support The programs have been for fellowships, research, and modified and incorporated in university centers for foreign- the Higher Education Act, with language and area studies, increased emphasis on support- ..,. . . . ,k putting emphasis on the training the education of nonspe- ing of specialists in languages cialists as well as specialists in “needed by the federal govern- foreign language and area ment or by business, industry, studies. About $26 million was or education in the United spent on those programs in ~, States.” Some $5 million was fiscal 1983. ,* appropriated in the first year. .,=[3 n,

fellowships for students plann- vamped and transferred to the I ing to become college teachers Higher Education Act in 1972. and grants to encourage the The Education Department’s establishment and expansion principal graduate-fellowship of graduate schools. About program provided support for $5.3 million was spent in the women and for blacks and first year. other minorities. About $10 roil- lion was provided in fiscal t 1983. , f

1959 ’60 ’61 ’62 ’63 ’64 ’65 ’66 ’67 ’68 ’69 ’70 ’71 ’72 ’73 ’74 ’75 ’76 ’77 ’78 ’79 ’80 ’81 ’82 ’83 Years SOURCE: Compiled by the American Council on Education, based on data from the U.S. Department of Education. Totals do not include Veterans Educa[lion Belnefits or Social Security Benefits for college students. Ch. 5—Education and Human Resources for Research and Development • 161 example, provided support for curriculum de- also influence the direction of manpower development and the production of new course velopment. 80 This was particularly true, for ex- materials. In addition, it has also lent consid- ample, during the years following Sputnik erable institutional support to colleges and when there was a widespread belief that the universities, in the form of direct grants, normal pool of technically trained graduates research grants and contracts, equipment could not meet the manpower needs of the ci- grants, and the establishment of institutes and vilian space and the military ICBM programs. specialized instrumentation centers. The Fed- Where a large number of people have been in- eral Government has, moreover, assisted involved, the Government has tended to provide dividuals through training programs and fel- only generalized incentives. Where the num- lowships as well as by providing scholarships ber is much smaller and the area of expertise and other forms of financial aid. quite specific, it has sought more to narrowly target its support. By targeting scholarships or fellowships for specific fields of study, the Government can ‘“Ibid.

The Societal Context for Determining Federal Manpower Policy

The role of the Federal Government in de- move the educational system more in the di- velopment of scientific and technical manpow- rection of manpower ‘development can be er is not only indirect, it is also complicated. achieved without sacrificing other equally, if It is complicated by the fact that the Nation’s not perhaps more important, educational and educational system-the primary institutional societal goals. means by which the Federal Government can Education has always played a particularly affect manpower— serves a number of societal important role in American society. For while goals in addition to, and at times competing educational institutions are publicly supported with, those that directly relate to manpower in many societies, in no other country have development. they been established with such deliberate pur- Today, given the growing concerns about pose and public expectation, or been conceived the United States’ declining position in the in- of as being such an integral part of the politi- ternational economy of high-technology goods cal, social, and economic order. Contrasting and services, the American educational sys- the attitude of Americans towards education tem is being called on to develop the Nation’s with that of Europeans, for example, Alexis scientific and technical manpower as a means de Tocqueville, the well-known French com- of gaining for the United States a greater com- mentator on American society, noted in 1831:81 petitive edge. In examining this issue, this Everyone I have met up to now, to what- chapter has posed a number of questions ever rank of society they belong, has seemed about: 1) the causal relationships between edu- incapable of imagining that one could doubt cation and economic performance, and 2) the the value of education. They never fail to extent to which there will actually be a criti- smile when told that this view is not univer- cal shortage of manpower in the future to per- sally accepted in Europe. They agree in think- form R&D in such high-technology areas as ing that the diffusion of knowledge, useful for information technology. Before policy altern- all peoples, is absolutely necessary for a peo- atives can be adopted to address the manpower problem, it is necessary to inquire further and a] Alexis de Tocqueville, Journey to Amen”ca, translated by to ask first whether or not Federal efforts to George Lawrence, J.P. Mayer (cd.) Anchor Books, 1971. .

162 ● Information Technology R&D: Critical Trends and Issues

ple like their own, where there is not property ing in the United States and is increasing the qualification for voting or for standing for potential economic and social penalty for those election. That seemed to be an idea taking who do not respond to those demands. More- root in every head. over, the study found that the information The goals that Americans have sought to revolution is creating new stresses on many achieve through education have changed over societal institutions, particularly those such time and in different historical circumstances.82 as public schools and libraries that tradition- In the earliest years of American history, for ally have borne the major responsibility for example, education was considered essential providing education and other information for the survival of the new democratic nation. services. Later, with the need to acculturate immi- New demands will also be placed on the ed- grants into society and to unite a divided Na- ucational system by virtue of the changes that tion in the aftermath of the Civil War, it was are now taking place in the American econ- considered the means for building a nation of omy. Programmable automation, for example, citizens. At the turn of the century, education as the recent OTA study, Computerized Man- was expected to train and socialize American ufacturing Automation: Employment, Educa- youths to participate in a modern, industrial- tion, and the Workplace, points out, is one of ized society. In the 1960s, Americans saw in the economic forces that is currently reshap- education a way of providing equal access to ing the roles for and the values being assigned social and economic opportunity. to education, training, retraining, and related Today, much of the national discussion services such as career guidance and job coun- about education has focused on the goal of seling. This study concludes, moreover, that meeting the manpower needs of a high-tech- the inadequate capacities of the present in- nology economy. Unlike earlier periods of structional system may constrain the estab- American history, however, when the educa- lishment of strategies to develop adequate tional system was itself undergoing tremen- skills for programmable automation. dous expansion and enjoying a period of con- Educational institutions will also be called siderable prosperity, today the educational onto perform a number of cultural tasks. Fore- system is having to take on a multitude of new casts of demographic trends suggest, for ex- tasks at a time when it faces the prospect of ample, that Hispanics, Asians, and other cul- shrinking economic and human resources. Thus tural groups with specialized educational needs today, given limited social and economic re- will soon comprise a major portion of the school sources for education, the choice to take Fed- population. Moreover, the increase in the num- eral actions to increase the supply of man- ber of school-age children living with a single power for R&D in the area of information parent or two working parents will force schools technology might be made at the expense of to provide more supervision and socialization. addressing other educational problems or of In addition, a growing number of adults are turn- meeting other educational and societal goals. ing to the educational system to fill in the non- Some of the most important of the new vocational, liberal arts gaps in their educational tasks that educational institutions will have backgrounds. 83 Thus, instead of cutting back on to perform are those that relate to the emer- their educational activities, schools and univer- gence of an “information society. ” In the re- sities will, most likely, have to cater to a grow- cent OTA study, Information Technology and ing number and variety of educational needs. it was Its Impact on American Education, The role of the educational system as a found, for example, that the growing use of means of providing equitable access to eco- information technology throughout society is nomic and social opportunities could become creating new demands for education and train- ——— —— ‘a’’More Adults Return to College h Study the Liberal Arts,” esw~ti~, op. cit.; cohen and Newfield, op. cit.; ‘1’’ya~, oP. cit. Chronicle of Higher Education, Apr. 25, 1984, p. 1. Ch. 5—Education and Human Resources for Research and Development . 163 all the more important in an age of high tech- tive to those earned by technical workers.ab nology. The Bureau of Labor Statistics, for ex- Under such circumstances, the competition for ample, has predicted that the number of low- educational credentials may become more in- skilled jobs will constitute an increasingly tense in the future, and the issue of equitable larger proportion of the total work force than access may become at least as important as, the highly technical jobs.84 Other social observ- if not more important than, the issue of man- ers have suggested, moreover, that the wages power development. earned by low-skilled workers will decline rela- —— .— 8’Bob Kuttner, “The Declining Middle, ” The Atlantic BiBLS, Levin and Ruxnberger, op. cit. Monthly, July 1983.

Chapter 5 References

“Academic Gap Between High and Low Achievers “More Foreign Students Enroll, But Only Some Narrowing, ” Chem”cal & En~”neering News, Stay to Work, ” Physics Today, October 1983. Feb. 29, 1983. New Directions for Federal programs to Aid Math- America’s Competitive Challenge, Report of the ematics and Science Teaching (GA0/REMD—84- Business-Higher Education Forum, 1983. 5, Mar. 6, 1984). “Comrnentary: The Science Education Stampede,” “NSF Survey Shows Academic Employment Science & Government Report, Oct. 1,1983, pp. Gains for Scientists and Engineers, ” AFIPS 7-8. Washington Report, September 1983, p. 4. “Computer People: Yes, They Really Are Differ- “Recent Reports on Education: A Digest, ” ent, ” Business Week, Feb. 20, 1984, pp. 66-77. Change, November/December 1983. Education, Productivity, and the National Econ- Science and Engineering Education for the 1980’s omy: A Research Initiative, National Institute and Beyond, National Science Foundation and of Education, December 1981. the Department of Education, October 1980. “Education for a High Technology Future: The Science Indicators 1982: An Analysis of the State Debate Over the Best Curriculum, ” School Re- of U.S. Science, Engineering and Technology, search Forum, Educational Research Service, National Science Board, 1983. May 1983. “Supply of Engineering Faculty,” Electronic Mar- “Employment Up in Manufacturing for Scientists, ket Trends, January 1982. Engineers, Technicians, ” National Science “The Changing Face of Engineering, ” Special Re- Foundation, Mosai”c, March/April 1983. port, Electrom”cs, May 31, 1983, pp. 125-148. “Engineers Need More Learnin g,” Infoworld, May The National Science Foundation and Pre-College 30, 1983, p. 32. Science Education: 1950-1975, report prepared Hem”ngs on Mathematics and Science Education, for the Subcommi ttee on Science and Technol- before the Committee on Education and Labor, ogy, U.S. House of Representatives, 94th Cong., House of Representatives, 98th Cong., 1st sess., 2d sess., by the Congressional Research Serv- Jan. 26-28, 31, 1983. ice, Library of Congress, January 1976. “Is Shortage of Engineers a Matter of Defini- “The Science Education Stampede, ” Science & tion?” Science, vol. 223, p. 800. Government Report, Oct. 1, 1983, p. 6. “Labor Market Slackens for Science and Engineer- “Q and A on Job Market for Science, Engineer- ing Graduates, ” National Science Foundation, ing, ” Science & Government Report, vol. XII, Mosm”c, January/February 1983. No. 14, Sept. 1, 1983. Manpower Concerns, Hearings, before the Com- Abernathy, William, Clark, Kim B., and Kantrow, mittee on Science and Technology, House of Alan, Indust~”al Renm”ssance: Producing a Com- Representatives, 1st sess., Oct. 6-7, 1981. petitive Future for America (New York: Basic “Manpower Surveys Continue to Disagree, ” Elec- Books, 1983). tronics, July 28, 1983. Adler, Mortimer, “The Reform of Public Schools, ” 164 • Information Technology R&D: Critical Trends and Issues

The Center Magazine, September 1982, pp. Dollar, Bruce, “What Is Really Going On in 12-33. Schools,” Social Policy, fall 1983, pp. 7-19. Aiken, James N., Teacher Supply/Demand 1983: Dupree, A. Hunter, Science in the Federal Govern- A Report Based Upon an Opinion Survey of ment: A History of Policies and Activities to Teacher Placement Officers, Association for 1940 (Cambridge, MA: The Balknap Press of School, College and University Staffing, 1983. Harvard University Press, 1957). Atkin, J. Myron, “Who Will Teach in High Evangelauf, Jean, “Salaries of New Assistant Pro- School?” America Schools: Public and Private, fessors Vary More Than $8,000 Across Disci- Daedalus, summer 1981. plines, ” The Journal of Higher Education, Feb. Bailyn, Bernard, Education in the Forming of 29, 1984, pp. 15-17. American Society (New York: W. W. North, Evangelauf, Jean, “Top Students Move to Science 1980). Studies, Leave Humanities,” Chronical of High- Bezdek, R. H., Long Range Forecasting of Man- er Education, Feb. 11, 1984, p. 1. power Requirements (New York: Institute of Finn, Chester F., Jr., “American Education Re- Electrical and Electronics Engineers, 1974). vives, ” Wall Street Journal, July 7, 1982. Blakesless, Sandra, “Teacher Fulfillment Put Fox, Jeffrey L., “The Uneven Crisis in Science Ed- Above Pay,” New York Times, Oct. 4, 1983, p. ucation,” Science, vol. 221, p. 838. C1. Ginzberg, E., and Voita, G. J., “The Service Sec- Botkin, James, Dimancescu, Dan, and Stata, Ray, tor of the U.S. Economy, ” Scientific American, “High Technology, Higher Education, and VOl. 299, March 1981, pp. 48-55. High Anxiety,” Technology Review, October Haddad, Jerrier A., “Key Issues in U.S. Engineer- 1982, pp. 49-52. ing Education, ” NAE Bridge, summer 1983. Boyer, Ernerst L., and Hechinger, Fred, Higher Harris, Marilyn A., “Manpower Surveys Continue Education in the Nation’s Service, the Carnegie to Disagree, ” Electronics, July 28, 1983, pp. Foundation for the Advancement of Teaching, 108-110. Washington, DC, 1981. Herman, Ros, “Science Slops in the Land of Op- Brademas, John, “Graduate Education: Signs of portunity,” The New Scientist, July 14, 1983, Trouble, ” Science, vol. 223, No. 4639, Mar. 2, pp. 110-113. 1984, p. 881. Heylin, Michael, “High School Science Problems Cohen, David, and Neufeld, Barbara, “The Fail- Gain Spotlight,” Chemical and Engineering ure of High Schools and the Progress of Edu- News, May 24, 1982, pp. 39-41. cation, America's Schools: Public and Private, Hubbard, Pat Hill, “Technical Employment Pro- Daedalus, summer 1981. jections: 1983-1987, ” paper prepared for the Cooper, Edith Fairman, U.S. Science and Engi- Symposium on Labor-Market Conditions for neering Education and Manpower: Background: Engineers, National Academy of Sciences, Supply and Demand; and Comparison with Ja- Washington, DC, Feb. 2, 1984. pan, the Soviet Union, and West Germany, Con- Irwin, Paul M., “Education in the 98th Congress: gressional Research Service, Library of Con- Overview, ” Issue Brief # IB83055, Congres- gress, May 24, 1982. sional Research Service, Library of Congress, Cooper, Edith Fairman, United States Supply and Oct. 28, 1983. Demand of Technology, Part I, Current Situa- Jordan, K. Forbis, “Precollege Math and Science tion and Future Outlook, Congressional Re- Education: Issues and Proposals, ” Issue Brief search Service, Library of Congress, Nov. 6, # IB82092, Congressional Research Service, Li- 1981. brary of Congress, Aug. 1, 1983. Crernin, Lawrence, Traditional in American Edu- Jordan, K. Forbis, “Teachers for Precollege cation (New York: Basic Books, Harper, 1976). Mathematics and Science Programs, ” Congres- Curtis, Kent K., “Computer Manpower-Is There sional Research Service, Library of Congress, a Crisis?” National Science Foundation, Jan. 4, July 28, 1982. 1983. Kerr, Clark, “The Uses of the University” (Cam- Dennison, E. F., SIower Economic Growth (Wash- bridge, MA: Harvard University Press, 1972). ington, DC: Brookings Institution, 1979). Kutscher, Ronald E., “Future Labor Market Con- Ch. 5—Education and Human Resources for Research and Development ● 165

ditions for Engineers, ” paper presented to the Porter, Beverly Fearn, and Czujko, Roman, “Sci- Symposium of the National Academy of Sci- entific Employment in a Tightening Economy, ” ence, Feb. 2, 1984. Physics Today, February 1983. Kuttner, Bob, “The Declining Middle,” The Atlan- Prewitt, Kenneth, “Scientific Illiteracy and Dem- tic Monthly, July 1983, pp. 60-72. ocratic Theory, ” Daedalus, spring 1983, pp. Layman, John W., “Overview of the Problem, ” 49-64. Physics Today, September 1983. Richie, Richard W., Hecker, Daniel E., and Bur- Layton, Edwin T., The Revolt of the Engineers: gan, John, “High Technology Today and To- Social Responsibility and the American Engi- morrow: A Small Slice of the Employment Pie, ” neering Profession (Cleveland, OH: The Press Monthly Labor Review 106, November 1983, of Case Western Reserve University, 1971). pp. 50-58, Levin, Henry M., and Rumberger, Russell W., Rosen, S., “Human Capital: A Survey of Empirical Forecasting the Impact of New Technologies on Research, ” Research in Labor Economics, the Future of the Job Market, Project Report (Greenwich, CT: JAI Press, vol. 1, pp. 2-39. #84–A4, National Institute of Education, Feb- Rosenberg, Nathan, Inside the Black Box–Tech- ruary 1984. nology and Economics (Cambridge, MA: Cam- Levin, Henry M., and Rumberger, Russell W., The bridge University Press, 1982). Educational Implications of High Technology, Schultz, T. W., Investment in Human Capital The National Institute of Education Report # (New York: Free Press, 1971). 83-A4, 1983. Setlzer, Richard, J., “Impact of Technology on McDermott, Jeanne, “Technical Education: The Employment Probed,” Chemical & Engineering Quiet Crisis,” High Technology, November/De- News, Aug. 1, 1983, pp. 23-24. cember 1983, pp. 87-92. Shils, Edward, “The Order of Learning in the McDonald, Kim, “Engineering Deans Ask Con- United States from 1865 to 1920, ” Minerva, vol. gress to Give Their Field Equal Standing With 21, No. 2, summer 1978. Science in Mission of NSF, ” The Chronical of Silverstri, George T, Lukosiewicz, John M., and Higher Education, Feb. 29, 1984, p. 13. Einstein, Marcus E., “Occupational Employ- Mansfield, Edwin, Education, R&D, and Produc- ment Projections Through 1995, ” Monthly La- tivity Growth, revised, University of Pennsyl- bor Review 106, November 1983, pp. 37-49. vania, Jan. 31, 1982. Smith, Adam, The Wealth of Nations (New York: Miller, Harry G., Dean, School of Technical Ca- The Modern Library, 1937). reers, Southern Illinois University at Carbon- Stedman, James B., Education in America: Re- dale, Carbondale, IL, “Higher Education for ports on Its Condition and Recommendations Employment: Considerations of Degree Deter- for Change, Issue Brief #83106, Congressional mination for Emerging Technical Careers. ” Research Service, Library of Congress, Aug. 5, Missimer, William C., Jr., “Business and Indus- 1983. try’s Role in Improving the Scientific and Tech- Tyack, David, and Hansot, Elizabeth, “Conflict nological Literacy of America’s Youth, ” T.H.E. and Consensus in American Education, ” Amer- Journal, February 1984, pp. 89-94. ica Schools: Public and Private, Daedalus, Moshowitz, Abbe, “On Predicting R&D Skill Re- summer 1981. quirements for Information Technology,” un- Upthegrove, William R., “Engineering Manpower published paper prepared for the Office of Tech- and Education: A Report of the Business-High- nology Assessment, February 1984. er Education Forum, ” paper presented at the Noble, David, America by Design: Science, Tech- Symposium on Labor-Market Conditions for nology, and the Rise of Corporate Capitalism Engineers, National Academy of Science, Feb. (New York: Alfred A. Knopf, 1977). 2, 1984. Olson, Larry, Future Technical Management Re- Vanski, Jeane E., “Projected Labor Market Bal- quirements and Availability in Information ance in Engineering and Computer Speciality Technology R&D: United States and Compari- Occupations: 1982 -87,” prepared for the Sym- son With Japan, France, and the United King- posium on Labor-Market Conditions for Engi- dom, unpublished paper prepared for the Office neers held at the National Academy of Sciences, of Technology Assessment, 1983. Washington, DC, Feb. 2, 1984. 166 ● Information Technology R&D: Critical Trends and Issues

Veblen, Thorsten, The Theory of the Leisure Class tions of Post Academic Science, Transaction (New York: The Modern Library, 1934). Books, 1982. Walgerg, Herbert J., “Scientific Literacy and Eco- Welter, Rush, Popular Education and Democratic nomic Productivity in International Perspec- Thought in America (New York: Columbia Uni- tive,” Daedalus, spring 1983, pp. 1-28. versity Press, 1962). Weinstein, Jay, Sociology/Technology: Founda- Chapter 6 New Roles for Universities in Information Technology R&D Contents

Page Introduction ...... 169 A Conceptual Framework ...... 169 Forces Driving New Relationships ...... 171 Impacts of New University Arrangements ...... 173 Expected Benefits ...... 174 Potential Costs ...... 175 New Roles for Universities: Selected Case Studies ...... 176 Massachusetts Institute of Technology Microsystems Industrial Group ...... 177 Microelectronics and Information Sciences Center University of Minnesota ...... 178 Rensselaer Polytechnic Institute Center for Industrial Innovation ...... 180 Stanford University Center for Integrated Systems ...... 182 Microelectronics Center of North Carolina...... 184 Semiconductor Research Corporation ...... 189 Microelectronics and Computer Technology Corporation ...... 193 Chapter preferences ...... 195

Tables Table No. Page 26. Stanford University, Center for Integrated Systems Research Topics ...... 185 27. A Distribution of SRC Funding by Region, March 1983 ...... 192

Figures Figure No. Page 30. A Conceptual Framework: New Roles for Universities in Information Technology R&D ...... 170 31. Stanford University, Center for Integrated Systems Patent Policy ...... 183 32. Communications System Linking MCNC Participating Sites...... 187 33. MCNC Participating Institutions ...... 187 34. Working Relationships, Microelectronics Center of North Carolina ...... 188 Chapter 6 New Roles for Universities in Information Technology R&D

Introduction

Throughout history, new institutional ar- of advancing technologies in new products and rangements have been created to satisfy processes; the efficient use of high-technology changing needs. As one of society’s major in- manpower and ensuring a renewable supply stitutions, the university, too, has evolved. of manpower resources; economic survival and Particularly in the case of information tech- future industrial growth; and maintenance of nology R&D, new roles for universities are national security and defense. developing. Other major players–industry, Since these efforts are essentially in their State and local governments and the Federal formative stages, it is difficult to draw con- Government–are involved along with the uni- clusions now about their long-term impacts. versity in the formation of new institutional In establishing a framework for analysis and relationships. policy options, OTA developed a series of sev- The new institutional arrangements been case studies (described later in this chapter) tween university and industry, as well as those that were selected as examples of the range among university, industry and government, of new institutional relationships. Taken to- are driven by many factors. These include the gether, they illustrate several of the key issues need for new knowledge and the application in today’s debate.

A Conceptual Framework

OTA created a conceptual framework to id advances in research and development at analyze the changing institutional R&D rela- both the university and in the industry. Sim- tionships being played by the Nation’s aca- ilarly, the development of new industries and demic institutions–one that emphasizes the subsequent economic growth are directly tied pivotal role being played by them. Figure 30 to the products coming from the university— outlines this framework and focuses on the highly trained technical graduates and new connections among university, industry, and knowledge, new processes, and new applica- government in terms of education, research, tions—as well as to the advances and offshoots and economic development. At the same time coming directly from industry. there are forces converging on these institu- In examining the institutional players in tions that, while creating new opportunities terms of their relationship to education, re- and strengthening connections among univer- search, and economic development, we see that sity, industry and government, also are cre- both universities and industry are directly in- ating strains and producing tensions. volved in the creation of new knowledge While education, research and development, through research, that both universities and and economic development are separate func- State and local governments are directly con- tions, they, like the institutions that foster cerned with the educational process and the them, are becoming increasingly interrelated. provision of a renewable supply of trained For example, concepts in advanced computer graduates, and that both State and local gov- architecture taught in a university program ernments and industry are directly concerned or course are directly dependent upon the rap- with economic health and growth. 169 170 . Information Technology R&D: Critical Trends and Issues

Figure 30.–A Conceptual Framework: New Roles for Universities in Information Technology R&D

*Changing nature of R&D: increasing complexity; high cost of research. *Dwindling resources: faculty and students drawn to industry; obsolete facilities. UNIVERSITY

New knowledge Trained manpower innovation and expertise

INDUSTRY ECONOMIC GROWTH STATE AND LOCAL GOVERNMENT

● International competition joint lndustry- 1 ‘ Economic development; need for new national R&D efforts. industries; strengthening the tax base. Products,servlces, *Need for cooperation to reduce high * Expanding Iimited resources for jobs, revenue risks, high cost of R&D. economic development, education, and public services, * Forces driving new Institutional relationships

SOURCE Off Ice of Technology Assessment

There are also important indirect relation- with a renewable supply of highly trained ships. The educational program and knowl- manpower. edge resources of the university can have an impact on the economic well-being of the re- There is also a national dimension to this gion. Industry is involved in education indi- triad. These relationships and interactions are rectly through the training and resources it affected by and in turn affect national issues has to offer to the other players. Industry is and the Federal role. The strength and effec- also a consumer of talent and new ideas, the tiveness of the educational system, the quality products of the university. States and local- of research, and the level of economic growth ities are finding that they need to be concerned and industrial innovation and productivity de- with research not only because the creation of termine, in part, the Nation’s national secu- new knowledge can lead to the development rity and economic well-being. The role of the of new processes and products by industry, Federal Government has also been both direct but because the quality and scope of univer- and indirect. For example, the passage of the sity research efforts can strengthen the educa- Merrill Act in 1862 fundamentally affected the tional program and in turn provide the region nature of research and education in the univer- Ch. 6—New Roles for Universities in Information Technology R&D ● 171 sities, as did the direct collaboration among government for various discretionary educa- government, industry and academic research- tion programs, providing a positive climate for ers during World War 11.1 Recently, the Fed- industrial joint ventures, and encouraging eral Government’s role has become more in- joint sponsorship of R&D through tax incentives.2 —-—direct, increasing— authority to State and local ‘See ch. 6, “The Provision of Education in the United States,” in Information Technology and Its Impact on American Edu- ‘See ch. 2, “The Environment for Research and Development cation, OTA, 1982. in Information Technology in the United States. ”

Forces Driving New Relationships

One can argue that the institutional frame- traction of top-quality faculty and the recruit- work has been in place a long time, even though ment of advanced-level students, many of the interconnections may not have been sharp- whom are being drawn to industry, are criti- ly defined. Why, then, are the relationships cal problems.5 Further, as information technol- among the university, industry, and State and ogy R&D advances, multidisciplinary efforts local government increasing in strength and are required to achieve new breakthroughs.G activity today? Although there are many fac- Not unexpectedly, universities have had to tors that could be considered, several forces seek new ways to operate educational and re- appear to be critical. search programs. One of the principal factors has been the Given the resources for R&D within the ma- change in the direct and indirect roles played jor information technology corporations,’ it is by the four participants shown in figure 30. logical to ask why industries would initiate or The changing Federal role in education, re- be responsive to new institutional research research, and economic growth has shifted cer- lationships. The change in the scope of the in- tain areas of responsibility to State and local formation technology industry from a national governments, academia, and the private sec- to a global arena has been a critical factor. tor. While Federal R&D funding has stayed Competition within the industry has expanded roughly constant in real dollars over the past from the U.S. to a new situation where the decade, recent increases have been targeted to competition derives from nationally coordi- specific areas, such as defense. The consolida- nated industry-government efforts world- tion of federally funded discretionary pro- wide.8 grams in education has increased local and be more than $1 billion. National Research Council, Revitaliz- State decisionmaking and control. ing Laboratory Instrumentation (Washington, DC, National Academy Press, 1982). Universities have been constrained by the ‘Louis Branscomb, former Chairman of the National Science resources available for support of research, fac- Board, points to needs for advanced degree training in com- ulty, students, and facilities.3 The rapid obso- puter science, electrical engineering, polymer science and . materials engineering-a problem which requires both fellow- lescence of laboratory and research tools, cou- ship support and the strengthening of university instructional pled with the highly complex and sophisti- and research facilities, At the same time, he points out that cated nature of the equipment now needed for incentives must be provided to make university research careers as attractive as offers from industry-principally through the advanced information technology research, re- provision of research and equipment support. See: L. Branscomb, sults in capital costs beyond the reach of most “The Computer’s Debt to Science,” Perspectives in Computing, academic institutions.4 The retention and at- vol. 3, No. 3, October 1983, p. 18. 61 bid, pp. 13-15. See also ch. 3, Case Studies on Advanced — . . . . —. . —- Computer Architecture, Fiber Optics, Software Engineering, ‘W. R. Lynn and F. A. Long, “University-Industrial Collab- and Artificial Intelligence. oration in Research, ” Technology in Society, vol. 4, 1982, p. 199. ‘See table 23, R&D Intensities of Selected Major U.S. Tele- ‘For example, in 1970 a need for $200 million in new instru- communications Firms, 1982, ch. 4, Divestiture. mentation in the Nation’s university research laboratories was Wee ch. 7. Information Technology R&Din the United King- identified: a decade later the accumulated need is estimated to dom, France, and Japan, for examples of such efforts. 172 ● Information Technology R&D: Critical Trends and Issues

Based on the perception that industrial well as local communities have become active growth, productivity and competitiveness are competitors in seeking to attract high-technol- dependent on new knowledge and innovation, ogy firms. Just as U.S. industries have had and a renewable supply of highly trained man- to acknowledge the international change in the power, industries have turned increasingly to competitive forces for their products, so too the universities. The increased cost of R&D have State and local governments had to rec- makes cooperative efforts highly desirable ognize that the competition for high-technol- among the industries themselves, and among ogy industry is interregional. The intensive cosponsored efforts with universities. Such co- State bidding for location of the Microelec- operation goes beyond cooperation among tronics and Computer Corporation (MCC) is large companies. It includes cooperation be- such an example, with some 60 mayors and tween large and small companies, and among 27 governors involved. Notes Arizona Gover- business, academia, and government.9 In sum, nor Bruce Babbitt, given the rapid advances in technology, the The great MCC bidding war marks a spe- escalating costs of R&D, and the global inten- 1 cial chapter in American industrial history. sity of competition, intranationa. cooperation State and local governments across the coun- is seen as a means of maintaining international 10 try have discovered scientific research and competitiveness, and universities are seen as technological innovation as the prime force cornerstones of the cooperative effort. for economic growth and job creation. And local officials have also uncovered a broad Over the past two decades, several regions base of public interest that can be translated of the United States have developed strong into support for aggressive action programs. local economies based on high-growth, tech- With the exception, perhaps, of the post- nology-based firms that are engaged in sys- Sputnik era, such grassroots enthusiasm for tematic development and commercialization science and technology has not been seen of new products, processes, and services. since the Gilded Age of the 19th century, These firms, and the industries they represent, when communities vied to finance the trans- have provided a major source of new jobs in continental railroads. * 11 12 the manufacturing sector. Thus States, as To attract such industry, incentives such as ‘See for example, testimony by Erich Bloch, Vice President, tax breaks, donations of real estate, venture IBM, and Chairman, Semiconductor Research Corp. “In order capital for industry and funding for educa- to cope with increasing competition in the world market, the semiconductor industry must increase its efforts in research tional programs have been provided. ’3 Not all and development. At the same time the research tasks are be- State and local high-technology initiatives coming more complex and more capital-intensive; lead time is have focused on education, nor does every increasing and the shortages of sufficiently trained manpower make the staffing of needed projects difficult. For all these rea- State or locality have equal resources on which sons, some research efforts are beyond the affordability of in- to draw. However, a strong educational base dividual companies. ” Hearings Before the Subcommittee on In- is seen as a way of becoming more competi- vestigations and Oversight and the Subcommittee on Science, Research and Technology, House of Representatives, 98th tive. In a survey of 691 high-technology firms, Cong., 1st sess., June 29-30, 1983, Japanese Technological Ad- completed for the congressional Joint Eco- vances and Possible United States Responses Using Research nomic Committee, “the importance of skilled Joint Ventures, p. 46. IOW. B. Norris, “How to Expand R&D Cooperation. ” 13usi- labor points up the necessity of linking State ness Week, Apr. 11, 1983, p. 21. Keynote Address “Coopera- and local development efforts with a region’s tion for Improving Productivity, ” San Diego, July 20, 1983, universities in order to attract high-technology IEEE Task Force on Productivity and Innovation. “U.S. Congress, Office of Technology Assessment, Technol- ogy, Innovation, and Re~”onal Economic Development, Census of State Government Im’tiatives for High Technolo~ Industrial Development-Backgmund Paper OTA-BP-STI, May 1983; En- *B. Babbitt, “me Sta@s and the Reindustriahzation Of Amer- couraging High-Technology Development—Background Paper, ica,” Issues in Science and Technolo~, fall 1984, p. 85. OTA-BP-ST1-25, February 1984. “For examples of such efforts, see the report by the National 12Tot~ employment in the information technology industry Governor’s Association, Zkwhnology and Growth: State lm”tia- experienced considerable growth in the decade between 1972 tives in Technolo~”cal Innovation (Washington, DC: National and 1982, despite economic recessions. See ch. 9. Governors’ Association, 1983), pp. 23-45.

Impacts of New University Arrangements

While we can find examples of State and lo- other arrangements,15 the university today is cal high-technology initiatives and numerous in a special position. Universities are being examples of long-standing university-industry courted by all of the principal actors and many interactions including industry’s support for are initiating programs of their own. Most im- research through gifts of funds or equipment; cooperative research grants and contracts; the Foundation, Research use of university consultants, exchange of per- Relationships (Washington, DC: National Science Foundation, sonnel between universities and industries and 1982). 174 . Information Technology R&D: Critical Trends and Issues

portantly, they are the linking element in mul- Expected Benefits ti-institutional R&D relationships. The increased interactions among the uni- The new institutional relationships take versity-industry-government triad depicted in many forms; some efforts represent new and figure 30 underlie potentially successful ap- largely experimental ways of working togeth- proaches to several critical problem areas: er; many efforts that are being developed are ● not really new, but are evolving from previ- Research and new knowledge: The cou- ous efforts and relationships. However, all of pling between university and industry the efforts that OTA examined involve a set may match needs of both. Industry gets of agreements whose principal characteristics access to the research and knowledge that are multidisciplinary arrangements and com- resides in the university. Similarly, uni- mitments to research with long-term objec- versity researchers can benefit from the tives. While several of the ventures have been pool of industry expertise. The academic initiated by one or few individuals, the nego- community obtains R&D laboratory facil- tiated agreements themselves are made at the ities and research tools, as well as fund- institutional level. It is the level of commit- ing to undertake research. With increased ment and the extent of the involvement that interaction, the university has an oppor- differs from previous university-industry tunity for better understanding of the in- efforts. dustry’s practical concerns, and, conversely, industry may get a closer look at It is too early to know with any certainty the university’s research findings, speed- the benefits and costs of the new university- ing technology transfer. centered activities. However, in breaking new ● Education and manpower: With the in- ground, the university arrangements, cospon- creased opportunities for research, and a sored efforts, and high-technology State and strengthening of the academic research local initiatives have generated high expec- program, the educational program can l6 tations amid questions of appropriateness. also be affected positively. Incentives The number of meetings, conferences, hear- that would attract and retain top-level ings, and publications on this subject has been faculty and advanced graduate students significant. 17 The debates over these relation- are derived from higher levels of support ships have involved university leaders and of new facilities and research. Moreover, academicians, governors, congressmen, and as a result of interacting with industry corporate executives. personnel, students can make more informed decisions about their future em- -.— — .— — ployment. The combination of top-level “This concern has been most focused on biotechnology, where personnel, adequate facilities, and a vig- se~reral university-industry agreements have involved large orous research agenda can strengthen the sums of funding, over multiyear periods, and where a major research unit of the university is involved with a single com- educational program, as new courses are pany with varying agreements for industry participation on developed and learning opportunities in- campus, and on some agreement to delay publication or pro- crease. The university products can then vide exclusive licensing to processes and products developed during the duration of the research agreement. feed back into both industry and the com- 174’ Academe and Industry Debate Partnership, ” Science, vol. munity. 219, No. 4481, January 1983, pp. 150-151; T. W. Langfitt, S. ● Economic growth: New institutional ef- Hackney, A. P. Fishman, et al., Partners in the Research Enter- prise, University-Corporate Relations in Science and Technol- forts are aimed at a strengthened research ogy (Philadelphia: University of Pemsylvania Press, 1983). U.S. base and a renewable source of highly House of Representatives, University/Industry Cooperation in trained manpower, which are needed by Biotechnology. Joint hearings of the Subcommittee on Inves- tigations and Oversight and the Subcommittee on Science, Re- industry for its economic growth. This, in search, and Technology, June 16-17, 1982. turn, can strengthen the regional econom- Ch. 6–New Roles for Universities in Information Technology R&D ● 175

ic base through new jobs, new spin-off in- Education and manpower: It is possible dustries, and the continued development that these new, highly visible, exciting of entrepreneurial efforts to fill new ventures will cause competition between niches. 18 research and education, drawing faculty away from teaching, and recruiting stu- Potential Costs dents from other areas. There is the danger that these new efforts will skew the The increased interactions among academia- balance among programs and capture un- industry-government also raise questions of equal attention and support from univer- long-term impact: sity administration. Economic growth and development: ● Research and new knowledge: Closer interaction may lead to subtle changes in While there are many other joint indus- the setting of research goals both in terms try-university activities involving small of the selection of topics to be studied and and large businesses, a variety of academ- in the shortening of time horizons for re- ic institutions, and individual faculty sults. Breakthroughs in fundamental re- members, the industry sponsors and ma- search often require decades of study. jor participants in both the multidiscipli- Thus some problems may be overlooked nary university centers and the industry if the institutional time frames for re- cooperative ventures have mainly been search are 3, 5, or even 10 years. How will the large information technology corpora- topics not of direct interest to industry tions. Fewer numbers of smaller compa- be covered in such institutional ar- nies have joined projects as ‘‘associate’ rangements? (in contrast to fully participating) members. Full membership includes access to In addition, industry’s traditional em- research as well as personnel exchanges, phasis on secrecy is in direct conflict with and active participation in planning re- academic practices. The new arrange- search, selecting proposals for funding ments have involved extensive negotia- and evaluating ongoing programs. Thus, tion regarding patent and licensing agree- cooperative joint-industry ventures among l9 ments. Thus far, most universities have the information technology giants may resisted industry pressure to limit access put the smallest, entrepreneurial compa- to research results, maintaining their tra- nies at a disadvantage. ditional role “to protect and to foster an environment conducive to free inquiry, The costs of increased interaction come from the advancement of knowledge and the two directions. First, in coming together, each free exchange of ideas. “2° The conflict be- of the institutions may lose some measure of tween openness and control is of continu- autonomy and relinquish aspects of their tradi- ing concern,21 and will require solutions tional roles. Conflict is inevitable, for exam- on a case by case basis. ple, between the university’s need for openness and industry’s need for protection of proprie- “For example, in a recent study of the Route 128 High Tech- nology Industrial Corridor, Massachusetts’ advantage h at- tary interests. Another conflict may develop tracting and supporting industries has resulted from the rich if the traditional distinctions that have sepa- university environment. “Entrepreneurs in the electronics fields rated the use of public funds from private income mainly from the staffs of universities and their research 22 labs, and from other high tech firms (already established).” N. S. terest and gain are blurred. Dorfman, “Route 128: The Development of a Regional High Second, there is the “cost” of nonparticipa- Technology Economy, “ Research Policy, 12:6, December 1983, p. 309; see also table 1, p. 301, ibid. tion. Most of the debate has focused solely on 1’Fowler, “University-Industry Research Relationships: The Research Agreement, ” Journal of College and University Law, 9:4, 1982-83. 22’’ WeighinK the Social Costs of Innovation, ” Science, vol. ‘“A. B. Giarnetti, “The University, Industry and Cooperative 223, Mar~h 1384, p. 1368. A suit involving the University of Research, ” Science, vol. 218, December 1982, pp. 1278-1280. California and its research, raises the question of the legality “D. Nelkin, “Intellectual Property: The Control of Scientific of spending public funds for research that allegedly benefits Information, ” Science 216:14, May 1982, pp. 704-708. large agribusiness more than small farmers and laborers. 176 • Information Technology R&D: Critical Trends and Issues

the generic institution without recognition of remain critical. Therefore, a diversity of efforts the differences within each of the institutional and approaches needs to be explored and sup- communities. Thus, it has not looked very ported. much at how the establishment of university At the State and regional level, the ability industry relationships may affect a smaller or to compete for new industry, for research cen- less prestigious research institution, nor how ters of excellence, and for expert manpower joint-industry ventures may affect smaller may also become equity issues. As noted ear- businesses, nor how local and State initiatives lier, while the number of high-technology cen- place some other regions or other institutions ters has increased in recent years, the competi- within the region in an inequitable position. tion among State and regional localities is There is concern that the already existing becoming fierce. The ability of States to assure differences between the Nation’s top-tier and significant support for new facilities, addition- second-tier universities may grow even great- al faculty and graduate students, the availabil- er as the competition for industrial resources ity of venture capital, and the cooperative ef- and partnerships accelerates. Thus, there is forts of business and academic leaders are concern for the needs of the range of the Na- critical factors in attracting new institutional tion’s universities. Even though programs research ventures. Moreover, as a result of suc- may be less ambitious in scope and scale, the cessful bids, regions expect to attract other needs for sophisticated equipment, advanced high-technology companies while the univer- research facilities, highly trained and knowl- sities hope to attract senior faculty and the edgeable faculty, and advanced-level students top graduate students.

New Roles for Universities: Selected Case Studies

University-industry-government information ● Institutional arrangements involve long- technology R&D efforts demonstrate a varie- term, multiyear commitments with agree- ty of approaches which have been only recent- ments that include facilities, equipment, ly implemented. Because these efforts are es- and human resources. sentially in formative stages, an assessment ● These arrangements bring together multi- of their effectiveness and impact on R&D is disciplines, multi-institutions, and multi- premature. Yet it is clear that the efforts ex- funding resources to support wide-ranging amined in this chapter provide important ex- research, educational, and development amples of new directions and major commit- efforts. ments. The case studies provide examples of ● These arrangements involve leadership how problems or barriers raised by these new and support of individuals at the highest relationships are being addressed, as well as levels of the university, corporations, and those issues which are not yet resolved. The government. case studies also provide an understanding of ● While Federal funds continue to support the motives and factors stimulating change, a significant portion of the research at the how the institutional players are responding, university centers, the Federal Govern- and the role played by incentives and direc- ment had a limited role in development tions from the Federal Government. Thus, of the institutional arrangements, influ- they provide a framework for analysis and the encing them by providing limited funds development of policy options. for startup activities, by creating tax While each of the case studies is unique, sev- credit incentives, and by its supportive eral themes emerge: policy towards joint ventures. Ch. 6–New Roles for Universities in Information Technology R&D ● 177

Massachusetts Institute of Technology contribute $250,000 annually for 3 years, and Microsystems Industrial Group associate member companies contribute $50,000 annually for 5 years. MIT’s experience with joint industry ar- Thus far, 18 companies have joined the rangements is extensive and reaches virtually group. 25 Full member companies can send one every program area in the Institute. More technical staff member to MIT to work in the than 300 joint programs are currently spon- Microsystems research program annually for sored at MIT by industrial firms. Federal 3 years. Each visiting professional submits a sources nonetheless provide the bulk of fund- plan of proposed research topics, and these are ing for research. In 1983, only 10 percent of matched with an appropriate faculty member MIT’s sponsored research was funded by in- 23 or research group. More than a dozen indus- dustry. It is estimated that support from in- try people have participated in the program. dustry will not grow beyond 15 or 20 percent. The opportunity to work in a new area of in- However, MIT faculty point out that indus- terest, “get caught up in the MIT atmos- try involvement is important because it pro- phere” and interact daily with students and vides exposure to and understanding of indus- faculty members is viewed very positively .2’ trial concerns, motivations, and needs. This interaction is seen as critical for the future of Research projects under way have been de- most students who graduate from the univer- veloped by faculty and reflect their traditional sity to work in the information technology roles as principal investigators. The director field. Equally important, it provides necessary and faculty meet with the industry member expansion of academic research concerns that advisory group, who provide information and would otherwise be guided primarily by the advice. Even more directly, the technical peo- interests of particular Federal funding agen- ple from industry have contributed to the re- cies.24 search efforts, and have broadened the view While there have been long and well estab- of faculty and students. The director of the lished industry-university research ties at Microsystems Industrial Group explains: MIT, the formation of the Microsystems In- “These are smart people with different back- dustrial Group breaks new ground. Increased grounds than my University colleagues. It is industrial involvement in the MIT Microsys- very important for those of us who do research tems program was stimulated by the pro- to have the industrial viewpoint in front of us. ‘ ’27 gram’s need for advanced state-of-the-art equipment and laboratory facilities. A propo- Understanding grows both generally sal to reach out to industry for help in devel- through interaction with the Council of mem- oping these facilities was made by members ber companies, who offer advice and guidance, of the faculty, who argued that the amount and in the process of working out specific vis- needed (originally estimated at $10 million) iting relationships. This understanding helps could not be supported by any Federal pro- faculty, students, and the academic program. gram or by the university itself. The advanced But it is also clear to faculty and administra- research Very Large Scale Integration (VLSI) laboratory and facilities are supported by in- z5Full member Compwies included AT&T Bell labs, Di@~ dustry sponsors; contributions are estimated Equipment, General Electric, General Motors, GTE, Harris, to be $10 million (half of the actual cost of ren- IBM, Raytheon, and United Technologies. Associate member companies are Analog Devices, GCA, Genrad, NCR, Polaroid, ovation and operation). The rest of the cost is Sanders Associates and Teradyne. being recovered through overhead charges on *’Personal communication, March 1984. According to Paul current contract research. In agreements nego- Penfield, Director, MIG, these experiences provide industry associates unique opportunities for professional growth and de- -—tiated —-— ———with industry, full member companies velopment, and appear to be a way for a company to retain “Kenneth A. Smith, “Industry-University Research Pro- highly valued employees. grams,”’ Physics Today, vol. 37, No. 2, February 1984, p. 24. “Personal communication. Paul Penfield, Director, Microsys- “Ibid., p. 24. tems Industry Group. 178 ● Information Technology R&D: Critical Trends and Issues tors who have responsibility for university- benefit the public and the research par- industry research that, even though time and ticipants and institutions; and competitive pressures are high, the academic 5. the issue of conflict of commitment— integrity of both the research and educational assuring that faculty are primarily com- program must not be compromised. Thus, in mitted to the university: its research and 30 some cases, decisions are made not to under- its educational programs. take certain projects, for example, when the proprietary stakes are too high, or the time- Microelectronics and Information frames are inappropriate or if scientific ex- Sciences Center (MEIS) 28 change is jeopardized. From industry’s point University of Minnesota of view, the Microsystems Industrial Group, as well as other similar efforts such as the Cen- The Microelectronics and Information Sci- ter for Integrated Systems at Stanford, are ences Center (ME IS) is a joint endeavor be- working because the research effort is tween the University of Minnesota’s Institute focused.29 of Technology and Minnesota industry. It was New institutional relationships can benefit created to establish a center of excellence in both university and industry if the agreements these sciences as well as to meet local industry’s technical manpower needs. Such joint ef- meet the needs of the partners. In analyzing 31 the aspects of such negotiations at MIT, the forts are not new to Minnesota. The impetus Associate Provost and Vice President for Re- for the Microelectronics and Information Sci- search identifies the fundamental issues to be ences Center came from Minnesota industries— addressed: Control Data, Honeywell, 3-M, and Sperry Corp. —who committed $6 million to launch 1. the relevance of a proposed line of inquiry the effort. The Minnesota State legislature to the essential missions of the university allocated an additional $1.2 million. Current and the industry—maintaining a balance operation is at $2.5 million a year matched by between the pursuit of research as an inte- $4.0 million in external grants and contracts. gral part of the educational process and industry’s need for useful knowledge to Faculty members, university officials, cor- be applied in the development of prod- porate executives and center administrators ucts, processes and services; have worked together to define the directions 2. the organization of a program that meets for research and educational programs, the the different time constraints of industry center’s operation, and the university-industry and the university—accommodating the interface mechanisms. This negotiation took time to work out, and programs were phased multiyear efforts of graduate students 32 with the shorter time pressures of the in gradually over a several-year period. The marketplace; 3. the issue of proprietary rights versus openness—achieving openness and free ‘°K. A. Smith, “Industry-University Research Programs, ” op. cit., p. 25. exchange of research results while protec- 31 For example, in the early 1970s such a joint effort initiated ting the industrial partners’ proprietary the development of the Minnesota Educational Computing Con- rights; sortium to provide instructional time-sharing capability to the 4. State’s colleges and universities, as well as the elementary and the issue of patents and copyrights— secondary schools. See: a case study of “Minnesota Schools and determining licensing agreements that the Minnesota Educational Computing Consortium, Informa- advance scientific and technological dis- tional Technology and Its Impact on American Education, (Washington, DC: U.S. Congress, Office of Technology Assess- coveries in ways that are most likely to ment, OTA-CIT-187, 1982) pp. 214- 221. 32The Center’s slow start has been criticized by some. How- ever, the benefits of taking the time to work out an arrange- ‘“Personal communication, George Dummer, MIT, March ment that suited the needs of both the university and the spon- 1984. soring industries outweighed the costs of delay. Personal *gPersonal communication, Bill Nelson, GTE, April 1984. communication, Dr. Martha Russell, March 1984, Ch. 6—New Roles for Universities in Information Technology R&D ● 179 results of the negotiations are embodied in the Education ME IS Center goals: 1. to sponsor and conduct research at the Strengthening the educational program has frontiers of microelectronic and informa- focused on increasing the number of faculty tion sciences; members, starting new courses, and attracting 2. to strengthen the course offerings of the top-quality graduate students. Eighteen new University of Minnesota in these sciences; graduate and undergraduate courses have and been added in computer science, electrical 3. to provide active interplay between uni- engineering, materials science and chemical versity researchers who seek discovery engineering; seven new faculty members have and industrial firms that apply those re- already been recruited through a 3-year cost- search results to the development and sharing program with the university, and marketing of innovative products and plans call for hiring an additional four mem- services. bers in Computer Science as well as a director; 54 graduate students and four post-doc- Research toral assistants were supported by MEIS funds in the five departments receiving ME IS Through the sponsorship of interdisciplin- research sponsorship; 16 fellowships will be ary research projects, the center links faculty, available for 1984-1985.33 students, and industry. Proposals for research are submitted by faculty members. ME IS- Technology Transfer sponsored research is reviewed annually by technical experts at the university and sup- The exchange of knowledge and technology porting companies. In 1983, projects in 3-D between MEIS member companies and the Integrated-Circuits, Processor Array Concepts university community has been a major goal for Engineering, Design Automation and Soft- of the Center. Through direct scientist-to- ware Engineering, and Ultrasmall Electronic scientist interaction, it is anticipated that the Research received ME IS seed funding, total- time between discovery and application will ing $625,000; additional research funds of be shortened. Faculty, students, and industry $4,312,000, principally from Federal grants, technical staff have worked jointly on projects, was obtained. In 1984, MEIS has awarded in some cases using industry’s state-of-the-art both seed and matching funding to three inte- facilities for design, special fabrication or grated team efforts in Intelligent Systems Re- testing. A major assumption is that graduate search, II I-V Semiconductor Materials and students serve a key role in the transfer of High Speed Devices, and High-Performance technology between industry and university. Integrated Circuits. Another planned effort After the first year of graduate study in the will include a project on Artificially Structured doctoral program, students work in the re- Materials. search laboratories of the industry sponsors, learning what drives industrial use of innova- Renovation and development of laboratory tion in science and technology, and bringing facilities has been directly tied to the research their recently acquired knowledge and skills efforts. The University is planning a new Com- to the task. Research projects developing from puter Science and Engineering building which these experiences expand the involvement of will house both offices and laboratories. ME IS faculty, students, and the industrial scientists. co-owns, with Argonne National Laboratory, a Synchrotrons X-Ray Beamline Facility, lo- Like other joint efforts, the center has fos- cated in Stoughton, WI. In addition, MEIS tered the exchange of ideas through confer- and the University share the newly remodeled — microelectronics laboratory and the VLSI “Microelectronics and Information Sciences Center, 1983 An- engineering design laboratory. nual Report, February 1984. 180 ● Information Technology R&D: Critical Trends and Issues

ences and seminars. ME IS technical reports The effort to improve RPI’s educational pro- and newsletters have also been widely dissem- gram was begun in 1975-76 and resulted in inated. Center participants from both academ- several new interrelated directions: expansion ia and industry point out that this open ex- of the graduate program (from approximately change has been facilitated by concentrating 500 to a goal of 2,400 graduate students by on long-term research areas conducted over a the year 2000); an institutional commitment 5 to 7 year period. The Center has thus far to research through the expansion of faculty avoided the issues regarding exclusive re- and facilities as well as of the number of stu- search and proprietary information. dents, and a revision of the undergraduate curriculum to overcome the lack of hands-on engi- Continued ability to recruit high-quality neering experiences. graduate students, as well as recruiting and retaining excellent faculty, is critical to the Center for Interactive Graphics long-term stability and growth of the program. Stable funding and full implementation of pro- The first step was the creation of an interac- grams are anticipated by 1985. In addition, tive computer graphics laboratory designed as the Center expects to attract additional State a service facility for undergraduates. This was and private support. based on the belief that an important emerging tool for engineering was the interactive Rensselaer Polytechnic Institute Center computer graphics terminal. The facility and for Industrial Innovation the applications have grown beyond the original classroom to a Center for Interactive The RPI Center for Industrial Innovation Graphics. The growth was due not only to the is the result of a focused University initiative increased use in almost all engineering that has involved key participants from aca- courses, but also to the decision to combine demia, industry, and New York State govern- research with practice as the means for keep- ment—including the governor. This initiative ing up to date with the advancing technology. was based on the experiences of RPI’s three The Center for Interactive Graphics was cre- established Centers for Interactive Graphics, ated in 1978, with initial funding from the Na- for Manufacturing Productivity, and for Mi- tional Science Foundation. From its inception, croelectronics. With a $30 million interest-free it was intended to involve industry, and to loan from the State and an additional $30 mil- share research results with industry. A meas- lion commitment by RPI, construction of a fa- ure of its success is that the Center has grown cility to house these Centers is under way. from 20 supporting companies with $20,000 These Centers involve more than 100 arrange- annual fees to 35 companies with $40,000 an- ments and agreements with industry, includ- nual fees. An early concern that it would not ing support for ongoing research through indus- be possible to keep up with the continually ad- try affiliates, specific research and vancing hardware and software has been re- problem-solving agreements, continuing edu- duced: companies have been willing to donate cation and training, adjunct industry-faculty their latest equipment. Just recently, for ex- arrangements, faculty-industry consulting, in- ample, the Center received a $3 million equip- dustry fee payments, and gifts or loans of 3S ment grant from IBM. equipment and software. However, it was not the quest for industrial partnerships, but Center for Manufacturing Productivity rather the desire to improve the undergraduate engineering education program, that The Center for Manufacturing Productivity served as the initial catalyst for these activi- and Technology Transfer was the result of a ties.34 deliberate decision to train students in areas — .— ——-— ...— “G. M. I.ow, “The Organization of Industrial Relationships —— —..-—— in LJni\’ersities, ” Partners in the Research Enterprise, ‘I’. W’. ‘ Personal communication, Dr. Christopher I,eMaistre, Direc- I.angfitt et al. (wis. ), (Philadelphia: Uni\’cv-sity. of Pennsylvania tor Center for Industrial Innovation and ~lssistant Dean, School pl.(=$s, 1 983), P. 71 of I’engineering, March 1984. Ch. 6—New Roles for Universities in Information Technology R&D ● 181

ance from a board of advisers comprised of founding member company representatives.

Center for Integrated Electronics The Center for Integrated Electronics also has industrial sponsors, who support research efforts which are fully open with no restrictions on disclosure.37 The Center is equipped with $8 million in hardware, much of it donated, from companies such as IBM, Calma, and Computervision. RPI also provides “in- cubator space” for small fledgling companies on campus and provides administrative support to help companies while they find venture capital, develop management capability, and begin to grow.

Graduate student and faculty member using a In addition, RPI has created an industrial CAD/CAM workstation, in the Center for park, located 10 miles south of the institute, Interactive Computer Graphics, RPI on a 1,200-acre parcel of land owned by RPI. With strong leadership from then RPI Presi- that would be needed in the decade of the dent George Low the Institute committed $3 1980s and at the same time to meet industry’s million for initial preparation of the site in research needs so that industrial funding 1982. National Semiconductor was the first would be available. Support from Federal company to move in, with several others fol- sources was not available. Thus, the Dean of lowing. RPI is currently constructing its own the School of Engineering made initial con- building in the park to provide startup space tacts with executives from General Electric. for companies that are not yet large enough He and two GE executives traveled to Europe to be on their own. to see how industry and universities were working together. RPI’s Center was modeled Center for Industrial Innovation after one at the University of Aachen in West All of these activities led to the Center for Germany. Industrial Innovation. The RPI President and With a commitment from GE, the Center the Chief Executive Officers of GE and Kodak, started in May 1979. Presently there are eight along with other corporate executives, met founding companies and five affiliate compa- with the Governor to push for a State Tech- nies who support the operation of the Center nology Initiative to be funded through the and its research.36 In addition, the Center en- State legislature. The arguments, as in other gages in contract work and involves under- industrial States, were that the smokestack graduates and graduates in “real life, real industries were dying, new technology indus- time” industrial problem solving, all under the tries were locating elsewhere, and that to over- direction of faculty and a project manager. The come this, new catalysts were needed. RPI intent is to create experiences that are directly argued that it had the necessary infrastruc- relevant for students’ entry into the industrial .— world. The Center reports to and receives guid- —- - . -—. 3TFounding members include Harris COW., Computervision, “Member companies include General Electric, General Mo- Digital Equipment Corp., Eastman Kodak, General Electric, tors, Boeing, Norton, IBM, Alcoa, Digital Equipment Corp., Raytheon, Polaroid, GTE, IBM, Phoeti Data Systems, Eaton and United Technologies. Affiliate companies include Kodak, Corp., and AIR Products. Affiliate members include Sperry, Cincinnati Milacron, Fairchild Republic, Fairchild Schlumber- Xerox, Hewlett-Packard, Perkin-Elmer, Fairchild Semiconduct- ger, Altech, and Timex. or, BTU Corp., Matheson, I’IT, and the PEW Memorial Trust. 182 . Information Technology R&D: Critical Trends and Issues ture in place and that what was needed was Planning Activities a $30 million interest-free loan. As early as 1977, Stanford engineering fac- The ground has been broken for the Center ulty discussed the idea of a center for inte- at RPI and the university expects completion grated systems research, a multidisciplinary of the facility in September 1986. RPI expects endeavor involving the interaction of people to put an additional $30 million into the facil- knowledgeable about integrated circuits with ities. In return for the State loan, RPI will pro- another group knowledgeable about computer vide an outreach program to 2-year colleges and information systems. “From the outset and to industry to upgrade the level of tech- it was clear that a collaboration, more intense nological expertise. than had ever before occurred, between IC types and systems engineers needed to evolve.”41 While there has been strong support for Furthermore, such a center could vertically in- these activities and agreement that they have tegrate the research process, with a state-of- helped the Institute, their industrial orienta- the-art facility for design, fabrication and test- tion does cause concern to some faculty. The ing of VLSI chips. This fast-turnaround facil- Center argument is, however, that there is a ity would allow a systems designer, in collab- healthy balance between uncommitted re- oration with an IC designer, to create search support and both focused engineering experimental devices in a shorter time than and applied projects. There is evidence that ever before. the program brings together a blend of research and application for students, and that The faculty took their idea to the Dean of the quality of the instructional program has the School of Engineering, and subsequently been significantly improved.38 a formal proposal was submitted to the university. By January 1980, the Center for In- Stanford University Center tegrated Systems was under way with approv- for Integrated Systems al from the Board of Trustees. Executives of Hewlett-Packard, TRW, Xerox, and Intel Like the MIT, RPI, and University of Min- formed a development committee to raise funds. nesota efforts, the institutional relationship By March 1981, 10 corporations agreed to con- developed between Stanford University and tribute $750,000 each, spread over a 3-year industry breaks new ground. In May 1983, period. By 1983-84, an additional 10 sponsors more than the construction of a $15 million fa- brought the total to 20, with each also agree- cility to house the Center for Integrated Sys- ing to provide $100,000 annually for educa- tems (CIS) was being celebrated. According tion, research, and administration of the facil- to participants from the faculty, university ity.42 administration, and industry, this project and others like it are part of a new willingness by Formulating New Policies industry and the academic community to be- Not unexpectedly, the most controversial come “allies in basic research.”39 According to aspect of the plan was not the facility, but the William Hewlett, “CIS is a clear and distinct intention to involve industrial companies as answer to three major problems that face the sponsors of the Center and offer them “facili- United States–the failure of our national pro- tated access” to the research program. Of con- grams of basic research to keep pace with the needs of our universities and industries, the ——- —- —.—. . need to strengthen our system of education, “John G. Linvill, Director Industrial Programs, Center for Integrated Systems, Stanford University. Personal communi- and the challenge to U.S. trade and technol- cation, April 1984. ogy posed by foreign countries.”40 ‘zCorporate sponsors are General Electric, Hewlett-Packard, TRW, Northrop, Xerox, Texas Instruments, Fairchild, Honey- 38Low, op. cit. well, IBM, Tektronix, Digital Equipment Corp., Intel, 11’”1’, “F. H. Gardner, “Special Report: The Center for Integrated GTE, Motorola, United Technologies, Monsanto, Gould/Amer- Systems, ” Hewlett-Packard Journal, November 1983, p. 23. ican Microsystems, Inc., North American Philips/Signetics ‘“I bid. Corp., and Rockwell International. Ch, 6—New Roles for Universities in Information Technology R&D . 183 cern to faculty members and corporate spon- under way at CIS. Moreover, both sponsors sors was how patent ownership, licensing, and and faculty point out that it is in their inter- intellectual property rights would be deter- est that the Center produce new basic knowl- mined. The companies’ initial posture was in- edge as well as students trained as broadly as sistent on exclusive proprietary rights to in- possible. ventions which involved their own people. The Another area of concern focused on the ques- university, on the other hand, argued that re- tion of research direction: would the nature search in the Center was funded principally and direction of basic research be distorted by with Federal support and that the Univer- industrial sponsorship? The answer appears sity’s legal obligation was to make sure that to be that this is unlikely, given the tradition any resulting patents would be brought into of independent research teams led by principal the stream of commerce as quickly as possi- investigators. At the same time, the invest- ble, with any company capable of commercial- ment of industrial sponsors is not insignifi- ization having the right to bid and obtain a cant, and there is the sense that faculty will license. be receptive to good problems posed by indus- The successful resolution of the issue cen- try. There is also the sense that research ques- tered on categorizing the patent in terms of tions can be shaped to examine fundamental the inventor (see fig. 31, CIS Patent Policy). issues likely to be of interest to all.43 There In addition, if a corporate visiting scholar de- is additional concern that there will be pres- velops a patentable product jointly with a sures to keep research secret. Such pressures Stanford faculty member or student, the com- are likely to be strongly resisted; the number pany may request a 90-day delay of publica- of seminars, publications, and open meetings tion to get the patent filed; it also gets free have demonstrated the University’s and the but nonexclusive rights to exploit the product. Center’s intent to maintain openness. CIS sponsor companies have also agreed to a Finally, there is the question concerning the cross-licensing plan, sharing any inventions advantage corporate sponsors have over non- developed at CIS. While intellectual property participating companies. The questions here rights have generated much discussion and have taken time to work out, it appears un- —. —..— “John Linvill, Director, CIS, notes, “We gain much more likely that highly commercial applications will through access with each other. The watchword is not isola- result, given the basic nature of the research tion, but interaction. ” Personal communication, March 1983.

Figure 31. —Stanford University, Center for Integrated Systems Patent Policy CIS Patent Policy Disposition of Rights

Visiting Scholar

. . p-~ -.---z= 1. Stanford takes title and Inventor receives nonexclusive, fullv paid Iicense Including right to sublicense 2. Upon request of inventors, Stanford will authorize a petltlon by Inventor’s employees to U.S. Government for greater r!ghts than 1

SOURCE: HewlettPackard Journal, November 19S3.

38-802 0 - 85 - 13 184 ● Information Technology R&D: Critical Trends and Issues relate not just to this project but to other proj- cipally by the Federal Government, total $12 ects as well. Proponents of the Center argue million a year. (see table 26: CIS Research that the fundamental nature of the university Topics). Seventy-one faculty members, repre- has not been changed, that everything the uni- senting seven departments, are affiliated with versity does is open to dissemination-a way the center, and 30 Ph.D. and 100 MS degree of life that undergirds every relationship—no candidate students a year are being trained. matter “whom we get money from. ”44 While It is the people who are the most important sponsoring companies have access to graduate output of the Center for Integrated Systems: students, and may seem to have an advantage “To the extent that we educate people with in recruiting them, the networks between the right background, have them do interest- Stanford faculty and their contacts in hun- ing research of significance to the Nation’s dreds of companies in the Silcon Valley remain problems but at a fundamental level, and do strong. In addition, if a CIS research team this in close collaboration with the industries wishes to include a nonparticipating company which need such people, we will significantly on a research project, they can do so. Reaching modify the nation’s productivity and competi- this agreement, noted several participants, tiveness. “46 was a hard fight to win. Nonetheless, the controversy persists, and Microelectronics Center of questions are likely to remain.45 The negoti- North Carolina (MCNC) ated agreements concerning patents, and the Universities play a critical role in the Micro- disposition of licenses are seen as experiments electronics Center of North Carolina (MCNC). which may or may not work out and which As a multi-institutional R&D effort, MCNC may need to be revised as research and work combines the resources of Duke University, progresses. The experimental nature of the North Carolina A&T State University, North center is not limited to intellectual property Carolina State University, University of North arrangements, note participants, but includes Carolina at Chapel Hill, and University of as well the social and organizational arrange- North Carolina at Charlotte, as well as the Re- ments built around cooperation—both among search Triangle Institute. These institutions, the various departments and among the cor- along with the MCNC core organization, the porations. new MCNC research facility, and the communications system linking all the facilities pro- Center Operation vide a concentration of resources for educa- While the Center itself is still evolving in tion, research, technology transfer, and terms of the agreements for intellectual prop- industrial development. erty rights, the working relationships between Established in July 1980, MCNC is orga- corporate sponsors and faculty, and the facil- nized as a private, nonprofit corporation to ities under construction, the research and in- assist North Carolina’s development of mod- volvement of the faculty and students are well ern electronics and related high-technology in- under way. CIS research projects, funded prin- dustry. MCNC has been funded primarily with —. —..—-. . —.- 441 bid. State grants, as a result of strong leadership 45A recent article in the New York Times highlights the con- from the governor and support from the leg- tinuing questions and controversy. While the Vice Provost notes islature. Thus far, $43 million has been allo- that CIS is an “innovative setup for Stanford, ” a professor of history argues, “It’s potentially very dangerous for a univer- cated by the North Carolina General Assem- sity to give privileged space and privileged access to informa- bly for constructing, equipping, and operating tion to particular companies. There is a danger that research- MCNC. An additional $34 million for new capers will create relationships that are likely to influence what they study and what they do not study. It is a threat to the ital facilities at the participating institutions autonomy of the university. ‘“ See: R. Reinhold, “Stanford and has also been provided. Industry Forge New Research Link, ” New York Times, Feb. 10, 1984. “John Linvill, personal communication, April 1984. Ch. 6–New Roles for Universities in Information Technology R&D ● 185 —

Table 26.—Stanford University, Center for Integrated Systems Research Topics

Although its building won’t be completed until early 1985, ● Multilevel-metal interconnection technology the Center for Integrated Systems is already coordinating a ● Biomedical silicon sensors $12-million-a-year basic research program at Stanford, funded ● Fast turnaround laboratory for VLSI largely by the Federal government. Here’s a sampler of re- Solid State cent and current topics investigated by CIS affiliated faculty ● Ion implantation and laser annealing in semiconductors in various existing labs. and related materials Computers ● Laser and electron beam processing of semiconductors ● Research on streamlined-instruction -set microprocessor ● Characterization of high-speed semiconductor device (featuring partnership of computer and IC engineers) materials using advanced analytical techniques ● Research in VLSI systems ● Ion implantation and laser processing of 3-5 compound ● Knowledge-based VLSI project conductors ● Image understanding ● Defects at electrode-oxide and electrode-silicon interfaces ● IntelIigent task automation in submicron device structures ● Network graphics ● Microstructure fabrication using electron beams of con- ● Partitionable computer systems ventional and very low energies . Analysis and verification of high-order language programs ● Advanced concepts in VLSI metalIization ● Study of very high-speed integrated circuits (VHSIC-Phase ● Advanced packaging concepts for VLSI 3) ● Studies of surfaces and interfaces of 3-5 compounds and ● Real-time communications systems: design, analysis, and Si:silicides implementation ● Silicon photocells in thermophotovoltaic energy con- Ž Structured design methodology for VLSI systems version ● Logical methods for program analysis . Investigation of metalIic impurities introduced into SiO2 ● Ultra-concurrent computer systems and Si by various candidate VLSI metallization systems ● Silicon compilation ● Modeling of emitters ● Data base theory ● Structural and bonding studies of practical semiconduc- ● Computer languages for VLSI fabrication tor layers Information Systems Space Telecommunications and Radioscience ● Multiple user channels and information theory ● Establishment of a Center for Aeronautics and Space In- ● Computational complexity, efficiency, and accountability formation Sciences at Stanford University (Funded by the in large-scale teleprocessing systems U.S. National Aeronautics and Space Administration. The ● Multiplexed holographic reconstruction methods for 3D focus will be on applying VLSI techniques to develop new structures hardware and firmware for space instrumentation and com- ● Information theory and data compression mand and control.) ● Signal processing and compression ● Communication satelIite planning center ● Statistical data processing, system modeling, and re- Material Science and Engineering liability ● Atomic-level physics modeling of the thermal oxidation ● Algorithms for Iocating and identifying multiple sources process by a distributed sensor network ● Fabrication and properties of muItilayer structures ● Fast algorithms for improved speech coding and recog- ● Computer simulation of surface and fiIm processes nition ● Photoelectronic properties of 2-4 heterojunctions ● Dual-energy digital subtraction radiography for noninvasive ● Photoelectronic properties of zinc phosphide crystals, arteriography films, and heterojunctions Integrated Circuits ● Photovoltaic heterodiodes based on indium phosphide ● Computer modeling of complete IC fabrication process ● Preparation and properties of CdTe evaporated fiIms com- ● Integrated electromechanical and optical sensor arrays for pared with single-crystal CdTe Optacon II (a reading aid for the blind) Ginzton Laboratory ● BME Center for Integrated Electronics in Medicine (to pro- ● Superconducting thin fiIms, composites, and junctions duce implantable telemetry systems for biomedical re- ● Acoustical scanning of optical images search) ● Research on nondestructive evaluation ● Computer-aided design of IC fabrication processes for VLSI ● Evaluation of machining damage in brittle materials devices ● Optical and acoustic wave research ● Submicron device physics and technology ● High-frequency transducers ● Development of multichannel electrodes for an auditory ● Research on acoustic microscopy with superior resolution prosthesis ● Study of properties of material by channeling radiation ● Study of transdermal electronics for an auditory prosthesis ● Surface acoustic wave MOSFET signal processor

SOURCE Hewlett Packard Journal, November 1983

186 • Information Technology R&D: Critical Trends and Issues

While the extent of State support is signif- ing engineers and scientists from industry, icant, such a role is not new for North Caro- and visiting faculty and graduate students lina. The precedents for government-industry- working on special research projects. The university cooperation span two decades. The l00,()()()-squar~foot,” $30 million facility has ca- development of the Research Triangle Park, pability for performing advanced manufactur- the Research Triangle Institute, the establishing processes, including high-density inte- ment of the North Carolina Board of Science grated circuit fabrication, system design, and and Technology, and now MCNC, are seen as design tool research. models of government-industry-university co- The MCNC $6.5 million communications operation to develop new technology-based in- 47 system, scheduled for completion by 1985, will dustries. put in place a ISO-mile microwave network linking the educational and research activities New Facilities for Education, at MCNC, the universities, and the Research R&D, and Technology Transfer Triangle Institute (see fig. 32, MCNC Commu- The MCNC facility, under construction since nications System). In the first phase hookup May 1982, will house core MCNC staff, visit- of the system, computer science students at —. Duke University in Durham and UNC at “U.S. Congress, of Technology Assessment, Technol- ogy, Innovation, and Development, Census Chapel Hill take classes (which originate from of State Government for High Technology Indus- Durham) together without leaving their own trial Development–Background Paper, OTA-BP-STI-21, May 1983, p. 56, campuses. Similarly, courses on Computer

MCNC dual source electron beam/r.f. metal evaporator for next generation integrated circuit manufacturing research Ch. 6—New Roles for Universities in /formation Technology R&D . 187

Figure 32.—Communications System Linking MCNC Participating Sites MCNC Communications System

● Unique high-performance microwave system. ● 152 miles from end to end. \ ● 2 two-way, full-motion video channels. . 7 program-origination locations. . 192 high-speed data paths.

SOURCE Mlcroelectronlcs Center of North Carollna, Research Triangle Park, NC

Graphics and VLSI Design originate from Figure 33.— MCNC Participating Institutions Chapel Hill. Microelectronics Center of North Carolina MCNC and the participating institutions Like the MCNC research facility, the development of the telecommunications system required funding beyond the reach of each of the individual institutions. “Before the center was created, each of our participating universities hoped to develop its own major microelectronics program. But the financial realities and the difficulties of attracting talent from the limited talent pool, soon made it apparent that the only way to develop a first class program was to join forces and work together. ”48

MCNC Working Relationships MCNC is more than a consortium of universities sharing resources and interacting with The MCNC community industry. The participating institutions are SOURCE: Microelectronics Center of North Carolina, Research Triangle Park, linked together by MCNC (see fig. 33, The NC. MCNC Community) and MCNC as an organizational entity and actor bridges the interests search and technology projects tie together and functions of the industry and university the commercial technology activities under communities (see fig. 34, Working Relation- way in industry and the basic research being ships). As members of industry work along conducted at the universities. with university researchers, the applied re- Since 1980, more than 30 new faculty mem- —..— . . bers in microelectronics-related disciplines “D. S. Beilrnan, President of MCNC, “New Initiatives in Mod- em Electronics, ” address before the Materials Research Society have been recruited. In contrast to the recent Annual Meeting, Boston, Nov. 14, 1983. trend of faculty leaving universities for em-

188 ● Information Technology R&D: Critical Trends and Issues

Figure 34.—Working Relationships, Microelectronics Center of North Carolina

SOURCE: Microelectronics Center of North Carolina, Research Triangle Park, NC ployment in industry, 20 of the 30 new faculty staff of the Center number more than 70, with have come directly from industry .49 Full-time 12 having joint institutional appointments. Moreover, the combined microelectronics-re- that was able to compete with be- lated manpower resources at the participating cause it offered industry-level salaries, access to an advanced institutions consist of over 150 faculty mem- state-of-the-art research and manufacturing facility, a strong university R&D environment, and a thriving high-technology bers and 450 graduate students, a significant industrial center. pool of talent. Ch. 6–New Roles for Universities in Information Technology R&D • 189

MCNC spokesmen argue that MCNC’s abil- The establishment of MCNC appears to be ity to bridge university/industry concerns is an example of effective university-industry- enhanced by its structure as a nonprofit “neu- government collaboration. Some point out tral” facility. The Center’s permanent staff, that it is an example of how State leadership specialists with joint university and Center ap- and initiative can be the driving force in pull- pointments, resident scientists and engineers ing together traditionally independent public representing industrial affiliates, visiting and private academic institutions and forging scientists from other national centers, and new relationships to attract significant indus- graduate students involved in special projects try participation in education, research, tech- physically work together in the MCNC micro- nology transfer, and industrial development. electronics manufacturing research complex. What has made MCNC work? In reflecting on the experiences thus far, the president of In contrast to MCC (see below), a joint ven- MCNC lists three basic requirements: ture owned by its corporate companies, MCNC leaders envision this effort as providing tech- 1. the need for a long-term strategic ap- nology functions that are not now provided by proach to substantial funding. This in- either MCC, other planned industry joint de- cludes building upon existing programs velopment programs, or other joint industry- and investments, as well as obtaining at university collaborative efforts in the micro- least a 3-year commitment from all mem- electronics field. “All of the industrial joint de- bers of the collaborative effort as a development programs or companies are spon- pendable commitment to common goals; sored by the first tier of large electronic 2. the need to structure in-depth interaction companies. The perceived need for these large- among the limited talent available. Full company joint efforts to remain competitive participation by personnel from industry, greatly amplifies the need for similar support universities, and government is necessary to the larger number of second-tier and evolv- for understanding each other’s perspec- ing smaller companies in the electronics indus- tive and for crystallizing and mutually ac- try. “5° MCNC expects to involve a broader cepting responsibility for important com- segment of companies, in part because of lower mon goals; and fee levels, and because of substantial, continu- 3. the need to plan for and accelerate the ing State support. The industrial affiliate pro- transfer of research into technology and gram is just getting under way. By 1985, the to promote R&D in progressively more Center expects to have 20 industry affiliates.” scientific endeavors while making use of all basic related investments.” Affiliates can come and use the MCNC facility to develop products, can participate in MCNC spokesmen are confident of MCNC’S research and educational programs (tuition future. Continued support for two-thirds of its and fees are provided for three staff members operation are expected to come from the State. at a time for each affiliate), are represented on Industrial and Federal support are anticipated an advisory council, participate in semiannual to cover the remaining third. reviews, and in the process have increased access to faculty and students. Nonexclusive, Semiconductor Research nontransferable licenses for intellectual prop- Corporation (SRC) erty rights are available to affiliates on a preferred royalty basis. It is expected that the Under the aegis of the Semiconductor Indus- majority of cooperative research will be openly tries Association, the Semiconductor Research disseminated. Corp. (SRC) was formed in 1981, to establish

—.—..—— — ———— ““D. S. Beilman, personal communication, April 1984. “D. S. Beilmen, “New Initiatives in Modern Electronics, ” “I+~ach industrial affiliate pays an annual $250,000 fee. op. cit. —

190 • Iformation Technology R&D: Critical Trends and Issues

a cooperative organization that would spon- semiconductor industry had to increase R&D.56 sor university research needed by the indus- Moreover, more complex and more capital- try. SRC was incorporated in February 1982. intensive research tasks, coupled with increas- As noted by Robert Noyce, of the Intel Corp., ing lead time and a perceived shortage of suf- “The semiconductor industry is fiercely com- ficiently trained manpower, created additional petitive and that competition has resulted in difficulties to undertake such research, partic- the vitality and success of the industry. ”53 The ularly for a single company .57 And drawing on cooperative institutional arrangement that the examples of other nations, such as Japan, makes SRC possible represents a change for joint coordinated efforts are seen as ways of the semiconductor industry. assuring long term competitiveness through cooperation. In the case of that industry, the forces stimulating cooperation centered around three These factors resulted in four major objec- issues: research, manpower resources, and in- tives for SRC: ternational competition. There was growing 1. increasing semiconductor research in the concern that the industry’s basic research ef- United States; forts, the foundation of its future well-being, 2. sharing research efforts among industry were increasingly directed towards the solu- sponsors; tion of near-term problems; that industry re- 3. strengthening and upgrading research in search efforts were often duplicative and re- the universities; and dundant, as each corporation tried to stay on 4. attracting more students to this field of top of the competition; and that fierce com- study and improving the quality of edu- petition and the squeeze for profits, combined cation. with increasing costs of R&D, created disin- centives and high risk for long-term industry Since its formation in 1982, SRC has grown research efforts.54 from 10 to 40 companies, of varying size, companies that manufacture or purchase semicon- At the same time, there was growing recog- ductor devices for manufacturing other prod- nition that the Nation’s research universities ucts, or companies that manufacture equip- were underutilized resources that industry ment or materials for use by the semiconduc- could turn to for long-term basic research and tor industry .58 Membership fees are tied to a creation of new knowledge. There was also con- company’s IC sales or purchases worldwide, cern that the pool of experienced and trained with annual fees ranging from $60,000 up. All manpower was being “overgrazed” by the in- — . -- - . ) dustry itself, and that both faculty and ad- ‘’’ E:. f?loch, prepared statement, Hearings, Japanese Techno- vanced graduate students were leaving univer- lo~”cal Advances and Possible United States Responses Using sities, irresistibly drawn to industry .55 Joint Research Ventures, Subcommittee on Investigations and Oversight and the Subcommittee on Science, Research, and In view of the growing competition for semi- Technology, U.S. House of Representatives, June 29-30, 1983. “Ibid., p. 46. conductor products, and the increased and co- 58SRC membership, as of September 1984, includes: Advancwl ordinated R&D efforts undertaken by its for- Micro Devices, Inc., AT&T Technologies, Inc., Burroughs Corp., eign competitors, it was argued that the U.S. Control Data Corp., Digital Equipment Corp., E.I. du Pent de Nemours & Co., Eastman Kodak, Eaton Corp., E-Systems, Inc., GCA Corp., General Electric Co., General Instrument Corp., .——— General Motors Corp., Goodyear Aerospace Corp., GTE Lab- “R. Noyce, “Competition and Cooperation–A Prescription oratories, Inc., Harris Corp., Hewlett-Packard Co., Honeywell, for the Eighties, ” Research Management, March 1982, pp. Inc., IBM Corp., Intel Corp., LSI Logic Corp., Monolithic Mem- 13-17. ories, Inc., Monsanto, Co., Motorola, Inc., National Semicon- “R. M. Burger, and L. W. Summey, “An Update on the Semi- ductor Corp., Perkin-Ehner Corp., RCA, Rockwell International, conductor Research Corporation”, QIE Conference Proceedings, Silicon Systems, Inc., Sperry Corp., Texas Instruments, Inc., 1983, pp. 51-59 Union Carbide Corp., Varian Associates, Inc., Westinghouse “Robert Noyce uses the analogy of the “Tragedy of the Com- Electric Corp., Xerox Corp, and Zilog, Inc.. In addition, the fol- mons’ where the commonly held resources, in this case, re- lowing companies are in the Semiconductor Equipment and search and manpower are exploited by industry self-interest, Materials Institute, Inc.: Micro Mask, Inc.; Pacific Western Sys- op. cit., p. 15. tems, Inc.; Probe-Rite, Inc.; Pure Aire Corp. Ch. 6—New Roles for Universities in Information Technology R&D ● 191

member companies have equal privileges: ac- for major research “centers of excellence” and cess to all sponsored research through semi- major research projects. In its initial solicita- nars, annual meetings, and newsletters and tion for proposals from the universities, SRC reports, access to research data bases and received 166 proposals from 63 universities. license rights, as well as an expanded recruit- In the first. year of operation, eight universities ing base. received research contracts. In 1983, more than 30 universities, involving approximately The SRC 1984 budget is over $15 million, 100 researchers and 125 graduate students, re- up from $6 million in 1982, and $10 million in ceived $10 million for research through 47 con- 1983. Currently, approximately $12 million is 60 tracts with SRC. By May 1984, 34 universi- available for university research projects, an ties involving 125 faculty and research staff amount which “substantially increases total and 202 graduate students were supported by available funding for basic research in semi- 59 $12.275 million in SRC research funding. conductor technology. ” Spokesmen point out that SRC has promoted research in engineer- It has been SRC policy to distribute the con- ing, mathematics, and the physical sciences tracts for centers and individual research proj- underlying semiconductor technology. Major ects on a broad geographic basis among lead- areas of focus established by the industrial ing research centers as well as to universities board members and the technical advisory whose expertise in these areas is not as well board are: established (see table 27, “Regional Distribu- tion of SRC funding”). Thus, SRC efforts may • Microstructure Sciences: have an impact that goes beyond the specific —Materials, Phenomena, and Device research projects: in helping to expand a uni- Physics, versity’s research capabilities, it may help it —Microsciences, attract high caliber faculty and graduate stu- —Device Fabrication Technologies. dents. Moreover, in the long run the SRC sup- ● Systems and Design: port may contribute to additional university- –Design Automation, industry cooperation, and new high-technol- –System Component Interactions. ogy industrial development. ● Production and Engineering: –Reliability, Quality Assurance, and SRC research “centers of excellence” include Testing, Cornell University, University of California at —Packaging, Berkeley and Carnegie Mellon University. Ma- —Manufacturing. jor research programs are supported at the North Carolina consortium (MCNC), Massa- Impacts chusetts Institute of Technology, Clemson, Stanford, Rensselaer, and the University of In planning the research activities to be California at Santa Barbara. Over the next few undertaken, several different levels of funding years, there are plans to support 8 to 10 more and effort were envisioned by board members. of these centers and programs, and to to con- SRC has awarded individual university re- duct research into design of microstructure, search projects, as well as several contracts properties of silicon material, computer-aided —.—. ——-— design and automation of design, lithography, “Erich Bloch, Former Chairman of the Board, SRC, estimates that the semiconductor industry allocates 3 to 5 percent of its beam processing, fault tolerance, micropack- R&D budget to basic research–approximately $35 million to aging and cooling, three-dimensional silicon $50 million annually. He notes that the R&D tax credit was structures, and manufacturing systems re- an important factor in the decision to proceed with the formation of SRC. Moreover, if there were a differentially larger tax search. incentive for industry-sponsored university research, there could be an even broader expansion of industry funding of university research in the future. Testimony before the Subcommit- tee on Taxation and Debt Management, Senate Committee on ‘(’N. Snyderman, ‘‘I ndustry Observer,Electrom”c .%’ews, ,Jan- Finance, Feb. 24, 1984. uary 1984. 192 ● Information Technology R&D: Critical Trends and Issues

Table 27.—A Distribution of SRC Funding by Region, involved both university researchers and in- January 1985 dustry participants. Beginning with abroad Region/institution Funding array of potential research needs, the groups New England $1,377,588 were able to reach consensus on research topic MIT ...... $ 976,110 priorities and areas for future focus. Yale ...... 211,258 Brown ...... 104,000 There is other evidence that SRC’s approach Vermont ...... 86,220 has fostered closer links between industry and Middle Atlantic: $4,441,210 Cornell ...... $1,776,651 academia. In addition to the technical advi- CMU ...... 1,414,580 sory board, composed of member company RPI ...... 800,000 representatives, SRC has established indus- Penn State ...... 126,489 Rochester ...... 119,000 trial mentors for each contract. With recoin- Johns Hopkins ...... 114,589 mendation from the technical advisory board, Columbia ...... 89,901 an industry engineer or scientist in an SRC North Central $1,734,900 Illinois ...... 750,607 member company becomes the direct contact Michigan ...... 337,155 point for each of the SRC contracts. The in- Minnesota ...... 206,755 dustrial mentor can help identify important Notre Dame ...... 126,000 Iowa ...... 118,013 problem areas, and may from time to time be Purdue ...... 102,870 able to provide direct technical assistance to Wisconsin ...... 93,500 the university research community. Through South Atlantic: $1,814,927 MCNC ...... $ 900,000 topical research meetings, additional industry- Clemson ...... 215,424 university contacts are strengthened. Georgia Tech...... 190,553 Auburn ...... 152,342 Program reviews of SRC Centers of Excel- North Carolina ...... 149,744 lence cover a wide spectrum of technical inter- Mississippi State ...... 135,000 Florida ...... 71,864 est and are designed to attract abroad repre- Mountain: $ 738,544 sentation from the industry and research Arizona ...... $ 503,530 community. Member companies may also par- Arizona State...... 100,914 Colorado State ...... 84,000 ticipate in SRC activities by assigning an em- Texas A&M ...... 50,100 ployee to participate in the management of the Pacific: $3,510,240 SRC program at Research Triangle Park. In- Stanford ...... $1,511,990 Berkeley (UC)...... $1,350,000 dustry assignees may also become research- Santa Barbara ...... 450,000 ers in residence, spending at least 3 months Southern California ...... 101,943 to a year, working in the university laboratory UCLA ...... 96.307 Total ...... ’...... $13,617,409 with one or more of the university research- SOURCE: Semiconductor Reseach Corp. Reasearch Triangle, NC. ers. This may foster technology transfer in ways that are not accomplished through the dissemination of reports, newsletters, and con- In its short span of operation, SRC provides ference results. an example of a joint industry approach in the management and coordination of information There is no question that SRC has provided technology R&D and in the establishment of additional research opportunities for the uni- new relationships between industry and aca- versity community and that these opportuni- demia for the conduct of research efforts.Thus ties have reached a range of institutions. Re- far, SRC member companies have been able search results have been freely disseminated. to agree on research priorities. Ongoing re- The ownership rights to the patents are held search projects have focused onVLSI circuit by the universities. So far, only one patent has processes and Computer-Aided Design, aimed resulted from SRC-sponsored research. Sev- at commercially relevant results over a 3-to eral researchers have indicated their apprecia- 5-year period. In developing a list of potential tion of lack of bureaucratic hassle in the SRC research topics for longer term research (e.g., contracting process, and find the yearly re- research needs in GaAs), SRC workshops have ports and reviews helpful. Ch. 6—New Roles for Universities in Information Technology R&D ● 193

The eventual impacts of SRC will be seen pany, MCC began formal operations in Janu- in how well it meets the needs of both univer- ary 1983, with the selection of its chief execu- sities and the semiconductor industry. For the tive officer and the development of a plan for member companies, the usefulness of research R&D. A technology Advisory Board of share- results, in both the short and long term may holder representatives provides advice in de- become factors in their continued support. For veloping the research strategies, in evaluating the industry as a whole, new knowledge and new program proposals, and in monitoring ex- new manpower are important, as well as the isting programs. attraction of additional researchers to new With a $50 million to $60 million annual fields of study in the future. How effective is budget, four long-range advanced technology the interface between the university and the programs are expected to cover a 6- to 10- year 64 industry over the long term? A sign of success, time Span. In defining the areas of research, at least interim success, note SRC spokesman, the shareholder companies “came to concen- is the increase in the number of member com- trate on areas in which they believed accom- panies and the continued support of the ini- plishments were necessary to make quantum tial companies who have signed on each of the jumps in the performance of the next genera- three years of operation. 65 tion of computers. ” The programs include: Microelectronics and Computer ● Packaging: A 6-year program to advance Technology Corporation (MCC) the state-of-the-art in semiconductor packaging and interconnect technology, The Microelectronics and Computer Tech- with a focus on technologies compatible nology Corporation (MCC) is an R&D joint- with automatic assembly at both the cir- venture owned and operated by 20 U.S. com- 61 cuit and system level. puter and semiconductor companies. The ● Software Technology: A 7- to 8-year pro- idea for MCC was conceived by William C. gram to develop new techniques, proce- Norris, President of Control Data Corp., and dures and tools that can be used to im- in his view “MCC represents a cooperative ef- prove the productivity of the software fort to develop a broad base of fundamental development process by one or two orders technologies for use by members who will each of magnitude. add their own value and continue to compete ● Computer-Aided Design and Manufactur- with products and services of individual con- 62 ing (CAD/CAM): An 8-year program to im- ception and design. ” While these companies prove CAD/CAM technology and to de- have traditionally avoided cooperation, “in velop an integrated set of tools that will this period of scarce resources, however, and have particular application to complex at a time when this country’s leading position systems and the complex VLSI chips in technology is being challenged by foreign from which they will be built. competitors, refusal to cooperate is no longer ● 63 Advanced Computer Architecture: This 10- tenable. ’ year effort will focus on artificial intelli- Governed by a Board of Directors composed gence, new techniques for database man- of representatives of each shareholder com- agement, human interface with computers, and parallel processing. ‘lShareholder companies include Advanced Micro Devices, Allied Corp., BMC Industries, Control Data Corp., Digital In addition to forming a comprehensive Equipment Corp., Eastman Kodak, Gould, Hank Corp., Honey- well, Lockheed, Martin Marietta Aerospace, Mostek, Motorola, agenda for research, MCC has selected a site National Semiconductor, NCR, RCA, Rockwell, Sperry Corp., for operation and hired staff. While still in tem- Boeing, and 3-M. 62W. C. Norris, “Cooperation for Improving Productivity, ” keynote address, Prepatory Meeting for the White House Con- 84B. R, Admiral, President, Microelectronics Computer Corp., ference on Productivity, San Diego, CA, July 20, 1983. personal communication, May 1984. 63W, C. Norris, “How to Expand R&D Cooperation, ” Busi- ‘i’M. A. Fischetti, “MCC: An Industry Response to the Jap- ness Week, Apr. 11, 1983, p. 21. anese Challenge, ” IEEE Spectrum, November 1983, pp. 55-56. 194 ● /formation Technology R&D: Critical Trends and Issues porary quarters, more than 173 professionals new faculty positions and 75 new graduate have been brought into the operation. Origi- fellowships would be supported. In addition, nally, the staffing plan was to draw senior and there was a commitment of at least $1 million highly trained technical professionals from the a year for maintenance and support for re- participating companies, with only about 25 searchers, and $5 million for purchase of lab- percent expected to come from the outside. In oratory equipment at UT-Austin. After MCC actuality, 40 percent of the professionals have selected the Austin site, an anonymous donor come to MCC from the shareholder companies. made available $8 million, with the proviso There is some concern that MCC will attract that other private sources match that amount. senior faculty members from the universities The university then matched that total, using and put a strain on available manpower re- funds from the Permanent University Fund souces, particularly in areas such as artificial (derived from revenues from oil leases on land 66 intelligence. MCC officials recognize that if owned by the University). The result is 32 new they hire away the university faculty, they will endowed chairs at the University of Texas, 10 compromise the universities’ ability to produce of which are in microelectronics and computer highly trained top-quality graduates-the very sciences. 68 67 people they need for the future. The developments in the academic commu- So far, the MCC strategy appears to be nity, the development of MCC, and the devel- working; of the talent on board, the majority opments in the fast-growing high-technology are from industry, and the remainder from corridor between Austin and San Antonio 69 academia and government. Full operation, ex- have drawn national attention. The potential pected by late 1985, will bring the total num- for economic growth, quality education and ber of professionals to 350. At full strength, cutting-edge research are cited as the real MCC will be looking for the “brightest grad- cause for excitement.70 Texas leaders point out uates, ” and it is expected that many will come that these high-technology initiatives (e.g., from nearby educational institutions–the MCC, the university programs) are just the be- University of Texas and Texas A&M Uni- ginning of the State’s commitment to high versity. technology development. It is recognized that not only the universities, but the entire State Not surprisingly it was these universities, educational infrastructure have to be strength- and their promised commitment to develop a ened and supported over the long term. The major, first-class computer science and micro- improvement of the State’s elementary and electronics program, as well as strong support secondary schools has been addressed by the from the State officials and the business com- Governor as well as MCC’s director, Admiral munity, that led to the decision to locate MCC Bobby Inman and other leaders in industry, headquarters in Austin, after conducting a who are concerned that without improvement search of 57 cities in 27 States. It has been Texas public schools represent a deterrent to noted that few cities in Texas—or anywhere recruiting engineers and other highly trained else—could put together the incentives that specialists. Among the recommendations of a were offered. The University of Texas at Aus- special panel headed by Texas industry leader, tin promised to construct a $20 million office- H. Ross Perot, are increased teacher salaries, laboratory facility, to be leased to MCC. Thirty ea~e the OTA Case study on Artificial Intelligence. 68A11 32 ch~rs me timed at strengthening the university’s @Tin a recent interview, Admiral Inman discussed this issue. science and engineering programs. Eight disciplines are thef~ “I have a standard rule that I will not recruit from universities. cus of this effort: chemistry, mathematics, molecular biology, If I am approached by someone on a faculty, my requirement physics, computer engineering, manufacturing, systems engi- is that they go up the chain and say they are going to leave neering, materials science, and microelectronics. to go to industry. I can’t have it both ways–to encourage the ‘gFor examples of recent economic development, see J. R. Lint+ production of additional topquality graduate students, and to backer, “Letter from Austin: Texas Cash Fuels Electronics hire away university talent. ” See: J. A. Turner, “Big-Spend- Boom. ” Electronics, June 15, 1983, pp. 95-96. ing U. of Texas Aims for the Top in Computer Science, ” Chron- ‘“J, Kraft, “The Japaning of Texas, ” 14’ashington Post, Apr. icle of Higher Education, Apr. 4, 1984. 17, 1984. Ch. 6—New Roles for Universities in Information Technology R&D ● 195

State aid to equalize school spending among noted earlier, MCC originally intended to draw rich and poor districts, and strengthened cur- its staff principally from the member compa- riculum requirements—at a cost estimated at nies, thereby speeding technology transfer. nearly $1 billion in new taxes. Since recruitment has drawn more heavily on outside sources, MCC will have to find other The ultimate test for MCC will be its ability approaches if it is to accomplish this goal. to draw sufficient talent to conduct the R&D necessary to keep its member companies inter- While it is too soon to assess the impacts nationally competitive. MCC officials and cor- of the MCC joint venture, and related activi- porate sponsors are confident that this can be ties at the University of Texas and Texas A&M, accomplished. Some observers are less confi- they do provide an example of how academia, dent that MCC will be able to transfer its tech- business, and government can join forces to nology to individual corporate efforts. As create new institutional arrangements.

Chapter 6 References

“Academe and Industry Debate Partnership, ” Brown, T. L. “University-Industry Relations: Is Science, vol. 219, No. 4481, January 1983, There a Conflict. ” pp.150-151. Burger, R. M. and Surnney, L. W., “An Update on “The Academic-Industrial Complex, ” Science, vol. the semiconductor Research Corporation, ” QIE 216, May 28, 1982, pp. 960-961. (Quab”tyin Electrom”cs) Conference Proceedings, “Artificial Intelligence (I): Into the World, ” Sci- 1983. ence, vol. 223, February 24, 1984, pp. 802-805. David, E. E., “Supporting Research with a Com- “Cooperation is the Key: An interview with B. R. mercial Mission, ” Change, vol. 14, September Inman, ” Communications of the ACM, vol. 26, 1982, pp. 25-29. No. 9, September 1983, pp. 642-645. David, E. E., “The University-Academic Connec- “The Challenges: Designing the Next Generation, ” tion in Research: Corporate Purposes and Social IEEE S’ectrurn, November 1983. Responsibilities, ” Journal of the Patent Office Probable Levels of R&D Expenditures in 1984: Society, vol. 64, April 1982, pp. 209-218. Forecast and AnaZysis (Columbus, OH: Battelle Dineen, G. P., “Why Cooperative Research is a Memorial Institute, December 1983). Viable Strategy at Honeywell. ” Presentation at “Texas Uses Oil to Fuel Research, ” Science, vol. Conference on Cooperative Research Ventures, 220, April 22, 1983, pp. 390-391. sponsored by Control Data Executive Forum. “Tomorrow’s Computers: The Quest, ” IEEE Spec- Apr. 4-5, 1984, New York City. trum, November 1983. Dorfman, Nancy S, “Route 128: The Development “Semiconductor Research Co-op Eyes 4 Megabit of a Regional High-Technology Economy, ” Re- Memory Chip,” Ekctrom”c News, July 18, 1983. search Poh”cy, vol. 12, No. 6, December 1983, pp. “Weighing the Social Costs of Innovation, ” Sci- 299-316. ence, vol. 223, Mar. 30, 1984, pp. 1368-1369. Fowler, D. R., “University-Industry Research Re- Ashford, N. A., “A Framework for Examining the lationships,” Research Management, vol. 27, Effects of Industrial Funding on Academic Free- No. 1, January/February 1984, pp. 35-41. dom and the Integrity of the University, ” Sci- Fowler, D. R. “University-Industry Research Rela- ence, Technology and Human Values, vol. 8, tionships: The Research Agreement, ” The Jour- Issue 2, spring 1983, pp. 16-23. nal of College and University Law, vol. 9, No. Brademas, J., “Graduate Education: Signs of 4, 1982-1983, Pp. 515-532. Trouble, ” Science, vol. 223, No, 4639, March Giametti, A. B., “The University, Industry and 1984, p. 881. Cooperative Research, ” Science, vol. 218, De- Branscomb, L. M,, “The Computer’s Debt to Sci- cember 1982, pp. 1278-1280, ence, ” Perspectives in Computing, vol. 3, No. 3, Gray, P. E., “The Role of the University in Global October 1983. Technological Change, ” In GlobaZ Techno]o~”- 196 ● Information Technology R&D: Critical Trends and Issues

cal Change: A Strategic Assessment, June 21- Norris, W. C., “How to Expand R&D Coopera- 23, 1983. Industrial Liaison program of the Mas- tion, ” Business Week, Apr. 11, 1983, p. 21. sachusetts Institute of Technology (Cambridge, Noyce, R., “Competition and Cooperation–A Pre- MA: MIT, 1983). scription for the Eighties, ” Research Manage- Hutt, P. B. , “University/Corporate Research ment, vol. , No. , March 1982, pp. 13-16. Agreements,” Technology in Society, vol. 5, Ploch, M. “Micros Flood Campuses,” High Tech- 1983, PP. 107-118. nology, vol. 4, No. 3, March 1984, pp. 47-49. Inman, B. R., “Research and Development Joint Prager, D. J. and Omenn, G. S., “Research, In- Ventures” testimony before the Subcommittee novation, and University-Industry Linkages, ” on Monopolies and Commercial Law, House Ju- Science, vol. 207, No. 254, January 1980, pp. diciary Committee, Sept. 28, 1983. 379-384. Kraft, J. “The Japaning of Texas, ” Washz”ngton Schmitt, R. W., “Building R&D Policy from Post, Apr. 17, 1984. Strength, ” Science, VO1.220, June 3, 1983, pp. Langfitt, T. W., Hackney, S., Fishman, A. P., et 1013-1015. al., Partners in the Research Enterprise, Schmitt, R. W., “National R&D Policy: An Indus- Um”versity-Corporate Relations in Science and trial Perspective, ” Science, vol. 224, June 15, Technology (Philadelphia: University of Penn- 1984, pp. 1206-1209. sylvania Press, 1983). Shils, E., “The order of Learning in the United Location of High Technology Firms and Re~”onal States from 1865 to 1920: The Ascendancy of Econorm”c Development, a staff study prepared the Universities, ” Minerva, vol. 16, No. 2, sum- for the Joint Economic Committee, Subcommitt- mer 1978, pp. 159-195. ee on Monetary and Fiscal Policy, U.S. Con- Smith, K. A., “Industry-Univeristy Research Pro- gress (Washington, DC: U.S. Government Print grams, ” Physics Today, vol. 37, No. 2, February ing Office, 1982). 1984, Pp. 24-29. Lynn, W. R. and Long, F. A., “University-Indus- Sproull, R. L., “Protectionism and the Univer- trial Collaboration in Research, ” Technology in sities,” Science, vol. 219, No. 4581, January Society, vol. 4, 1982, pp. 192-212. 1983. National Research Council, Rew”tdizing Labora- Thomas, L. “The Value of Basic Science, ” Review tory Instrumentation, the report of a workshop Magazine of the Um”versity of Rochester. of the Ad Hoc Working Group on Scientific In- Turner, J. A., “Big Spending University of Texas strumentation, Mar. 12-13, 1982 (Washington, Aims for the Top in Computer Science. ” Ckom”- D. C.: National Academy Press, 1982). cle of Higher Education, vol. XXVII, No. 6, Apr. National Science Foundation, Unz”versity/industry 4, 1984. Research, Relationslups: Selected Studies, 14th U.S. Congress, Committee on Science and Tech- Annual Report of the National Science Board nology, U.S. House of Representatives, Hear- (Washington DC: U.S. Government Printing Of- ings before the Subcomnu”ttee on Investigations fice, 1982). and Oversight and the Subcomzm”ttee on Sci- National Science Foundation, Um”versity-Industry ence, Research, and Techmdogy: Japanese Tech- Research Relationslups: Mflhs, Reah”ties and nological Advances and Possible Um”ted States Potentials, 14th Annual Report of the National Responses Using Research Joint Ventures, June Science Board (Washington D. C.: U.S. Govern- 29-30, 1983 (Washington, DC: U.S. Government ment Printing Office, 1982). Printing Office, No. 45, 28-3770, 1983). Nelkin, D., “Intellectual Property: The Control of U.S. House of Representatives, Hearings on Scientific Information, ” Science, vol. 216, No. Um”versity/Industry Cooperation in Biotechnol- 14, May 1982, pp. 704-708. ogy, Subcommittee on Investigations and Over- Noble, D. F., America By Design: Science, Tech- sight and Subcommittee on Science and Tech- nology, and the Rise of Corporate Capitah”sm nology, of the Committee on Science, Research (New York: Alfred A. Knopf, 1977). and Technology, June 16-17, 1982. Norns, W. C., “Cooperation for Improving Produc- U.S. Congress, Office of Technology Assessment, tivity, ” Keynote Address, IEEE Task Force on Commercial Biotechnology: An International Productivity and Innovation, San Diego, July Analysis, OTA-BA-218 (Washington, D. C.: U.S. 20, 1983. Government Printing Office, January 1984. Ch. 6—New Roles for Universities in Information Technology R&D ● 197

U.S. Congress, Office of Technology Assessment, U.S. Congress, Office of Technology Assessment, Informational Technology and Its Impact on Technology, Innovation, and ReD”onal Eco- American Education, OTA-CIT-187 (Washing- nonu”c Development: Encoura~”ng High Tech- ton, D. C.: U.S. Government Printing Office, No- nology Development—Background Paper 2, vember 1982). OTA-BP-STI-25 (Washington, DC: U.S. Govern- U.S. Congress, Office of Technology Assessment, ment Printing Office, February 1984). Technology, Innovation, and Regional Eco- Wilson, K. G., “Science, Industry, and the New nomic Development: Census of State Govern- Japanese Challenge, “ Proceedings of the IEEE, ment Initiatives for High-Technology Industrial vol. 72, No. 1, January 1984, pp. 6-18. Development—Background Paper, OTA-BP- STI-21., (Washington, DC: U.S. Government Printing Office, May 1983). Chapter 7 Foreign Information Technology Research and Development

Contents Page International Trends in Information Technology Research and Development 201 International Trade . . 201 Adapting Technology for International Markets 202 Multinational Corporations ...... 206 Technology Exchange Agreements . . . . . 207 Patents ...... 210 Scientific and Technical Literature . . 211 Science and Engineering Students . 212 Implications for U.S. Information Technology R&D Policies 214 Science and Technology Policy Goals ., . 215 Government Role in Information Technology Research and Development 216 Government/Industry/University Institutional Arrangements for Information Technology Research and Development ., 217 Industry Participation in Information Technology Research and Development 219 Conclusions ...... 221 Japan ...... 222 The Size of Japanese Participation in Information Technology Markets 225 Government ...... 226 University ...... 240 Industry ...... 243 France ...... 246 Introduction ...... ,, 246 The Size of French Participation in Information Technology Markets 247 The Political Environment for French Information Technology Research and Development . . . . . 248 Social Environment for French Information Technology Research and Development...... 252 Government ...... 252 University...... 257 Industry ...... 259 The United Kingdom ...... 261 The Size of U.K. Participation in Information Technology Markets ...... 263 Government ...... 263 University...... 269 Industry ...... 270 European Strategic Program for Research in Information Technology ...... 272 Tables Table No. Page 28. Computer Production and Apparent Domestic Consumption of Six Leading Supplier Nations...... , 201 29.U.S. Computer Trade: Origins and Destinations, Flow Value, and Annual Growth ...... 204 30. World Trade in Telecommunications Equipment ...... 204 31. Aggregate Trends in U.S. Telecom Equipment Trade ...... 204 32. Company R&D Performed Abroad by Foreign Affiliates of U.S. Domestic Companies by Selected Industry: 1975 and 1981 ...... , 207 33. International. Technology Agreements...... 208 34. Number of U.S. Patents Granted to Selected Foreign Countries in All Product Fields and in Communication Equipment and Electronic Components, 1963-81 ...... 210 35. Share of Foreign Patenting in the United States for the Three Most Active Countries by Selected Product Fields: 1981 ...... 211 36.U.S. Owned and Foreign Owned U.S. Patents in Information Technologies ...... 212 37. Distribution of Foreign Students With Percentage of All Foreign Students by Field of Study, for Selected Years: 1954/55-1981/82 . . 213 38. Doctoral Degrees Awarded to Foreign Students as a Percent of All Doctoral Degrees from U.S. Universities by Field: 1959-81, ...... 214 39. Average Annual Growth Rates in the Engineering Industry ...... 218 40.R&D Performed in the Business Enterprise Sector by Source of Funds: 1970, 1975, and 1979 ...... 220 41. Percent of Private Industrial R&D in Selected Industries: 1969-79 ...... 220 42. R&D Expenditure by Type of Activity ...... 224 43. Japan Development Bank Loans for Development of Technology...... 240 44. Department of Trade and Industry R&D Programs ...... 266 Figures Figure No. Page 35.U.S. Computer Trade Imports by Source; Exports by Destination...... 203 36. Sources of U.S. Imports and Destinations of U.S. Exports of Telecommunications Equipment...... 205 37.U.S. Bilateral Trade Position in Telecommunications, With Selected Countries . . . 206 38. Share of Foreign Patenting in the United States for the Three Most Active Countries in 1981 ...... 211 39. index of International Cooperative Research by Country ...... 212 40. Estimated Ratio of Civilian R&D Expenditures to Gross National Product for Selected Countries...... 217 41. Trends in the Production Composition Ratio of Major Consumer Electronics Equipment ...... 225 42. Japanese Share of World Production of Consumer Electronics Products in 1980. . 226 43. Japanese Information Technology Industry Sales ...... 227 44. Integrated Circuit Relative Production Share ...... 228 45. Japanese Government Organization for Information Technology Research and Development...... 229 46. Japanese Government Support for Information Technology ...... 233 47. Cooperation Between Research Participants for the Fifth-Generation Computer Systems Project ...... 234 48. Concept Diagram Showing How Research and Development Are to Progress in the Fifth Generation Computer Systems Project ...... 235 49. Chart Showing the Recommended French Government/Industry/University Cooperation Relationships for Information Technology R&D ...... 251 Chapter 7 Foreign Information Technology Research and Development

International Trends in Information Technology Research and Development

Several trends demonstrate that the United and activities relevant to U.S. R&D efforts. States is experiencing greater international in- First, they indicate a growing number of links terdependence in the area of information tech- between other nations’ R&D efforts and those nology research and development. They in- of the United States. Second, these trends clude: 1) the large and growing world market point to a growing participation of foreign na- for computer and communications products; tions in information technology innovation 2) the increasing adaptation of information and markets, which has led to a relative de- technology products and standards for inter- cline in the U.S. market share. Thus, the national markets; 3) the growing number of United States, which in the past has developed multinational information technology firms; policies for its internal markets that were 4) the increasing number of international tech- largely unaffected by foreign manufacturers, nology exchange agreements; 5) the increas- may now need to take greater account of for- ing percentage of U.S. information technology- eign information technology research and de- related patents granted for foreign inventions; velopment efforts. 6) the greater utilization of foreign contributions in U.S. scientific and technical journals; International Trade 7) and the growing number of foreign students World trade in computer products is grow- enrolled in technical and scientific programs ing rapidly. For each of the major supplier na- at U.S. universities. tions, overseas shipments are a steadily ris- These trends indicate two significant facing share of both total output and consump- tors, both of which make foreign organization tion. Table 28 shows this trend towards a glo-

Table 28.—Computer Production and Apparent Domestic Consumptiona of Six Leading Supplier Nations

1982 Percent 1982 Percent product ion change ADCa change ($ million) 1981-82 ($ million) 1981-82 United States...... $33,550 b 12.30/o $26,888 15.3’%0 Japan ...... 7,1 79C 21.0 6,276 10.2 France ...... 3,834d –5.0 4,720 8.4 West Germany...... 3,511 7.5 3,789 5.3 United Kingdom ...... 1,929d 11.8 2,898 NA Italy ...... 1,076 11.0 1,343 3.2 Total ...... $51,079 10.80/0 $45,914 NA Estimated share of world total (percent) ...... 89 80 aAPParent Domestic consumption (ADc) iS production ITIi I_IUS exports Plus ‘m Ports bEstimated by Bureau of Industrial Economics CDOeS not Include parts ‘Preliminary. SOURCE U S /ndustr/a/ Out/ook, 1984, Bureau of Industrial Economics, U.S Department of Commerce, 1984

201 202 . Information Technology R&D: Critical Trends and Issues

bal computer market as it has evolved over the Adaptations of Technology for last few years, and figure 35 illustrates ma- International Markets jor sources and destinations. This trend results in part from the rapid rise in demand for The growing international trade in informa- computer-related products in the developing tion technology products has led to increased world, a steady demand in traditional markets efforts to develop international standards for for products that incorporate information information technology products in order to technology, and increasing overseas activities allow access to foreign markets and to allow of multinational subsidiaries.1 interconnections of services. For example, following a recent meeting of the Commission of For the United States, increased global par- European PTTs (CEPT), European countries ticipation in the information technology mar- agreed to develop technical standards not only ket has meant a rapid rise in imports and a for basic equipment such as telephone hand- decreasing world market share. During the pe- sets, but also for videotex systems and other riod 1978-82, U.S. imports of computers and sophisticated data communications systems. computer-related products rose by approxi- The CEPT program will also suggest other mately 30 percent. (See table 29.) areas where national telecommunications prac- The telecommunications market is also be- tices might be standardized. This could even- coming internationalized as equipment man- tually lead to a unified European network of ufacturers look beyond maintaining tradi- approximately 400 million subscribers. The tional markets (the national telecommu- European Program for Research and Develop- nications service monopolies, or PTTs) toward ment in Information Technology (ESPRIT) expanding international trade.2 Table 30 illus- has a group working on international stand- trates this trend and summarizes the current ards specifically designed to enable various positions of the United States, the United European-manufactured products to commu- Kingdom, France, and Japan. nicate with each other. The internationalization of the telecommu- In markets where standards do not exist, in- nications market, as in the case of the com- formation technology products, such as computer market, has weakened the relative U.S. puter software, must be tailored for interna- position in telecommunications trade. Al- tional sale. Because personal computer though U.S. exports have increased at a rate hardware has proliferated worldwide without of 13 to 18 percent per year, a continuing in- a parallel growth of indigenous software com- crease in foreign imports (24 to 30 percent a panies, many American software companies year) has diminished the U.S. trade surplus are developing products for the international (table 31). Japan supplied about 50 percent of market. U.S. imports, resulting in a U.S. trade deficit 3 For example, Lotus Development Corp. has with Japan of $250 million (figs. 36 and 37). been tailoring its software packages to the lan- —. —.. .——.. guage and idioms of other nations. The Lotus ‘High Technology Industries: Profiles and Outlooks: The International Character Set enables the pro- Computer Industry, U.S. Department of Commerce, Interna- tional Trade Administration, 1983, p. 22. gram to generate different currency signs and ‘High Technolo~ Industn”es: Profiles and Outlooks, The Tel- different versions of international day and ecoxnmuxu”cations Industry, U.S. Department of Commerce, In- date displays. Although the cost of converting ternational Trade Administration, 1983, p. 18. ‘Although the French, the British, and the Japanese are in- programs for international markets can be creasing their participation in information technology markets, quite high, Lotus Corp. reportedly believes particularly in the computers and telecommunications areas, that the return on its investment will also be the degree to which this trend is linked to information technology R&D remains unknown. The traditional skills needed for substantial. They expect that international success in the marketplace range from basic research, to ap- sales will eventually generate between 30 and plied R&D, to production and distribution, and to marketing 40 percent of the company’s income.4 skills; it is therefore difficult to attribute success in the mar- —.— — ketplace solely to R&D efforts or to any other single factor. 4Mich~el Schrage, “Firms See Boom in Software, ”The Wiwh- See ch. 2 for a more complete discussion. ington Post, Mar. 4, 1984, p. H, 4. —

Ch. 7—Foreign Information Technology Research and Development ● 203

Figure 35.—U.S. Computer Trade (SIC 3573) Imports by Source; Exports by Destination

United Kingdom 15.20/o Other countries 25.80/0

I —

NOTE: SIC 3573 includes: Computing equipment (equipment, peripherals, and services). SOURCE: High Technology Industries: Profiles and Outlooks: The Computer Industry, International Trade Administration, U.S. Department of Commerce, 1983. 204 ● Information Technology R&D: Critical Trends and Issues

Table 29.—U.S. Computer Trade @W 3573): Origins and Destinations, Flow Value, and Annual Growth 1981-82, percent of value) (millions of U.S. dollars)

1982 Imports 1982 Exports Japan ...... $ 729 (+88.20/o) United Kingdom ... .$1,374 (+15.3%) Canada ...... 404 (+ 0.0%) United States Canada ...... 1,103 (–11.2%) Hong Kong ...... 151 (–21.60/o) Imports Exports West Germany . . . . . 958 (– 6.20/o) Mexico ...... 123 (+27.00/o) 1978 ...... $ 755 $4,194 France ., ...... 841 (+ 7.0%) United Kingdom . . . . 90 (+ 12.2%) 1981 ...... 1,646 8,493 Japan ...... 777 (+ 8.30/o) West Germany . . . . . 79 (+12.90/o) 1982 ...... 2,295 8,957 Netherlands ...... 380 (+ 14.0%) Spain ...... 78 (+25.3%) 1983 ...... 4,100 10,300 Australia...... 344 (+ 0.0%) France ...... 64 (– 2.6%) 1984 ...... 6,470 12,360 Italy...... 298 (– 4.1%) Total ...... $2,140 (+29.9%) Growth Total ...... $9,040 (+ 4.5%) (1978-82) . +29.8% +21.2`% NOTE: SIC Code 3573 includes: Computing equipment (equipment, peripherals, and services). SOURCE: H/gh Techrro/ogy /mWstr/es: Profiles arrd Outlooks: The Computer Industry, International Trade Administration, U.S. Department of Commerce, 1983.

Table 30.–World Trade in Telecommunications Equipment (SIC 3661) (millions of U.S. dollars)

1977 1981 Principal producer countries Imports Exports Balance Imports Exports Balance Japan ...... $ 24 $ 363 +$ 339 $46 $ 911 +$ 665 Sweden ...... 36 458 +422 65 776 +71 1 West Germany ...... 104 562 +458 128 809 +681 Netherlands...... 126 228 + 102 128 398 +270 France ...... 57 168 +111 86 320 +234 United States ...... 129 257 + 128 494 653 + 159 Canada ...... 93 80 –13 143 298 + 155 United Kingdom ...... 91 247 + 156 235 331 +96 Belgium/Luxembourg ...... 76 248 + 172 118 262 + 144 Italy ...... 50 97 +47 101 143 +42 Total ...... 786 2,708 + 1,922 1,544 4,901 +3,357 NOTE: SIC Code 3881 includes: Telephone and telegraph apparatus. SOURCE: High Tec/rno/ogy hrdusfries: Prof//es and Outlooks: The Telecommunications Industry, International Trade Administration, U.S. Department of Commerce, 1983.

Table 31.–Aggregate Trends in U.S. Telecommunications Equipment Trade (SIC 3661) (millions of U.S. dollars)

1977-83 1972 1977 1979 1980 1981 1982 1893 growth rate Exports ...... $76 $257 $448 $557 $653 $725 $850 +22.1 “/0 Imports ...... 86 129 319 421 494 635 790 +35.3% Balance ...... –10 + 128 + 128 + 136 +159 +90 +60 – 11.1% NOTE: SIC Code 3881 includes: Telephone and telegraph apparatus. SOURCE: U.S. /ndustrla/ Out/ook, 1983, Bureau of Industrial Economics, U.S. Department of Commerce, 1983.

International marketing is also an impor- West Germany. Like Lotus, Microsoft has tant component for Microsoft, a U.S. software revised many of its software programs for company whose overseas market accounted foreign use. The company has closely tailored for approximately one-third of its estimated its software products for the Japanese mar- $75 million 1983 revenue. Microsoft already ket by offering phonetic Japanese versions of has development operations in Japan and sub BASIC, and it has translated its Multiplan sidiaries in the United Kingdom, France, and program (business applications program) and Ch. 7—Foreign Information Technology Research and Development Ž 205

Figure 36.—Sources of U.S. Imports (1982) and Destinations of U.S. Exports (1982) of Telecommunications Equipment (SIC 3661)

Japan 45.20/o Netherlands

West Germany 2.4%

United Kingdom 2.9% Canada . . \ 22.50/o \ /“

-! ----- Kingdom I /

43.0% I 4.60/o

NOTE: SIC 3661 includes: Telephone and telegraph apparatus. SOURCE: High Technology /ndustries: Profiles and Outlooks: The Te/ecornmunicatio~s Industv, International Trade Administration, U.S. Department of Commerce, 1983 206 ● Information Technology R&D: Critical Trends and Issues

Figure 37.– U.S. Bilateral Trade Position in Amdahl and Trilogy Systems, manufacturers Telecommunications with Selected Countries (1981) of plug-compatible mainframes, have opened facilities in Ireland, intended in part to supply the Common Market. Japan Can U.K. Ger Neth Mex

+ 150 Many foreign firms operate subsidiaries in the United States and other foreign countries. + 100 For example, Japan’s NEC Corp. has estab-

+50 lished three subsidiaries in the United States and has won major contracts to supply U.S. (Mil $) O manufacturers with Japanese technology. In

-50 addition, NEC has subsidiaries in Germany, the United Kingdom, and countries in Africa -100 - Imports Exports and South America. – 150 This growth of international activity has not – 200 only led to increased trade in information technology products, but has also encouraged the performance of R&D by firms in various nations. After establishing foreign subsidiaries, SOURCE: High Technology Industries: Profiles and Outlooks: The Tebcommuni- cations Industry, International Trade Administration, U.S. Department many companies find that R&D is necessary of Commerce, 19S3. to support local manufacturing when the requirements or standards for the foreign marketplace are significantly different. its MSX operating system (home computer program) for the Japanese market. U.S. companies are performing an increasing amount of research and development Multinational Corporations abroad. Since 1975, total R&D conducted by U.S. subsidiaries overseas has more than dou- Rising innovation costs and the accompany- bled and in 1981 amounted to $3.2 billion-9 ing size of financial risks, as well as increas- percent of total U.S. private R&D funding. ing equipment costs, have intensified the need Table 32 illustrates the increasing amount of for expanding production and have forced electronics R&D which is performed abroad by many manufacturers beyond the limitations foreign affiliates of U.S. companies. In con- of domestic markets. For a variety of reasons trast, in 1979, total expenditures for elec- including tariffs and other forms of protective tronics research and development performed legislation that place imported products at a by U.S. affiliates of foreign companies in- competitive disadvantage, multinational firms creased to $148 millions have attempted to capture specific foreign markets through the establishment of foreign Many governments actively encourage for- subsidiaries. eign subsidiaries not only to establish production facilities, but also to conduct R&Din their Many U.S. firms have opened production nations. The United Kingdom has, for in- and R&D facilities in foreign nations. Digital stance, implemented a series of incentives for Equipment Corp., for instance, operates six foreign firms to innovate. In addition to pro- plants in Europe and three more in the Far viding financial incentives for establishing East. Hewlett-Packard, Wang, Data-General, manufacturing facilities, the United King- Datapoint, and Texas Instruments are U.S. dom’s Support for Innovation Program (SFI) minicomputer manufacturers that also operate foreign production facilities. Apple Com- ‘Science Indicators, 1982, National Science Board, National puter has plants in Ireland and Singapore. Science Foundation, 1983, p. 25. Ch. 7—Foreign Information Technology Research and Development . 207

Table 32.—industry R&D Performed Abroad by Located in Munich, close to a number of elec- Foreign Affiliates of U.S. Domestic Companies by tronics firms, the ECRC will begin operations Selected Industry: 1975 and 1981 (millions of U.S. dollars) with an initial capital investment of $655,000 and a staff of four researchers from the three Percent Industry 1975 1981 increase firms. The number of researchers is expected Food and kindred products. . . 23 66 187 to reach 30 to 35 by 1985, and approximately Chemicals and allied products 269 651 142 50 within 2 years. Industrial and other chemicals ...... 85 275 124 Drugs and medicines ...... 184 376 104 Technology Exchange Agreements Stone, clay, and glass products ...... 7 15 114 Table 33 illustrates some of the recent tech- Primary metals ...... 9 b nology exchange agreements between U. S., Fabricated metals...... (a) 26 N—A Machinery ...... 331 585 77 European, and Japanese companies. Such Electrical equipment ...... 245 455 86 technology exchange is seen by firms as a way Electronic components . . . . 7 47 571 to spread the risk in large projects, to enter Transportation ...... 412 893 117 Motor vehicles and other international markets where politics or nation- transportation equipment . 373 791b 112 al market specifications hinder entry into do- Aircraft and missiles ...... 39 lo2b 161 mestic markets, and to allow competitors to Professional and scientific instruments ...... 49 101 106 pursue a dominant position in a specific 6 Other manufacturing market. industries ...... 105 147 40 Nonmanufacturing industries . 4 12 200 An example of a technology exchange agree- Total ...... $1,454 $3,157 117 ment is the reciprocal development and mar- alncluded in the other manufacturing industries 9r0up keting agreement for office telecommunica- bEstimated tions equipment between AT&T and Ing. C. NA—Not available SOURCE: Science /rrdicafors, 1982, National Science Board, National Science Olivetti & Co., a major European supplier of Foundation, 1963 office automation equipment. In accordance with this agreement, AT&T will increase its offers grants of up to 33 1/3 percent towards stake in Olivetti over the next 4 years to the cost of significant research and develop- acquire 25 percent ownership in the company. ment of high technology products. About 200 The arrangement gives AT&T access to Oli- information technology firms are located in vetti equipment such as workstations, word “Silicon Glen” in Scotland. U.S. firms there processors, typewriters, and data processing include IBM, Honeywell, NCR, Hewlett-Pack- systems for domestic marketing. In turn, Oli- ard, Digital Equipment, and National Semi- vetti will market AT&T communications con- conductor. trollers for voice, data and networking applica- 7 Multinational information technology cor- tions, and a variety of micro-computers. porations are also forming cooperative inter- Other technical partnerships include those national research and development arrange- of LM Ericsson of Sweden with Honeywell, ments. This new type of international and Atlantic-Richfield in the United States arrangement is exemplified by the recent with Thorn EM I in the United Kingdom. In establishment of Europe’s first multinational Italy, Italtel is cooperating with Telettra and research institution for information technol- the U.S. company GTE in the public switch- ogy. Following an agreement made last Sep- ing field. The British firm, ICL, has links with tember, Europe’s three largest computer man- Mitel, and Plessey (another U.K. company), ufacturers, France’s Compagnie des Machines Bull SA, Britain’s International Computers Ltd. (ICL), and West Germany’s Siemens AG, —— — —— “’Bulls on Skis, ” The Economist, Feb. 4, 1984, p. 76. will operate a jointly run and jointly financed ‘In addition to its technical exchange agreements with Oli- European Computer Research Centre (ECRC). vetti, AT&T has also made exchange agreements with Philips. 208 . /formation Technology R&D: Critical Trends and Issues

Table 33.—international Technology Agreements

Sector/partners a Date Technology Agreement Communications: Hitachi and Western Electric 1981 Communication equipment Patent exchange—technological agreement Fujitsu and Ungermann-Bass (CA) — Local networks Industrial and commercial agreement NTT and Hughes — Telecommunications by satellite — Sperry-Univac and Mitsubishi — Local networks and Technological agreement communication and data processing system Motorola and NEC 1982 Portable paging systems Agreement for manufacture and commercialization in Japan Exxon Office Systems and — Telecommunications equipment Technological agreement Mitsubishi ATT and Philips 1983 PBX Commercial agreement Honeywell and Ericsson 1983 Telecommunications and office Technological and commercial automation agreement GTE and Italtel 1982 PBX Technological and commercial agreement Plessey and Stromberg Carlson 1982 PBX Purchase of Stromberg-Carlson by Plessey

General Instruments and Thomson 1983 Videocommunication and Technological and commercial teledistribution by cable agreement Micro V and Jeumont-Schneider 1982 PBX Technological and commercial agreement with partial acquisition of Micro V and creation of a joint subsidiary

Data processing: Honeywell and NEC 1984 Main frame computers Technological agreement Exxon Office Systems and Toshiba — Office automation Technological agreement TRW and Fujitsu — Data processing Joint venture now controlled 100 percent by Fujitsu Sperry and Mitsubishi 1982 Office automation Technological agreement Vertimag and Teijin (Osaka) — High-density magnetic memory Technological and commercial agreement Amdhal and Fujitsu 1982 Computers and peripherals Take-over by Fujitsu Drexler Technology and Toshiba — Smart card Technological and industrial agreement Olivetti and Docutel 1982 Office automation Control of Docutel by Olivetti Olivetti and Stratus 1982 Minicomputers Control of Stratus by Olivetti Nixdorf and Auragem 1983 Minicomputers Control of Auragem by Nixdorf Philips and Micom 1977 Office automation Purchase by Philips ATT and Olivetti 1983 Data processing, office ATT acquires 25 percent of Olivetti automation, communications Fortune and Thomson 1982 Microcomputers Thomson acquires 17 percent of Fortune Tandy Corp. and Matra 1982 Microcomputers Technological and commercial agreement Matra and Tymshare 1982 Terminals Commercial joint venture in the United States Cii Honeywell Bull and Trilogy 1980 Main-frame computers Technological agreement with Cii Honeywell Bull having minor share in Trilogy Cii Honeywell Bull and Honeywell — General data processing Technological and commercial cooperation AMD and IBM — Computer-aided design Commercialization by IBM of Catia Ch. 7—Foreign Information Technology Research and Development ● 209

Table 33.—lnternational Technology Agreements—continued

Sector/partnersa Date Technology Agreement Electronics and components: /rite/ and NEC 1982 VLSI microprocessors and circuits Technological agreement Texas Instruments and Fujitsu 1979 Integrated circuits Technological agreement Hewlett-Packard and Hitachi — Integrated circuits Technological agreement Western Electric and NTT — Integrated circuits Technological agreement IBM and NTT — Integrated circuits Technological agreement Zilog and Toshiba — Microprocessors Technological agreement 1982 Integrated circuits Production and commercial Western Digital and Siemens agreement United Technologies and AEG 1982 Custom-made semiconductors Production and commercial Telefunken agreement Philips and Signetics 1975 Components Purchase by Philips Philips and Magnavox 1977 Consumer electronics Take-over by Philips Philips and Sylvania 1980 Consumer electronics and tubes Purchase by Philips

GCA and Matra 1982 Microelectronic equipment Technological and commercial agreement Harris and Matra 1981 Components and integrated Technological, industrial, and circuits commercial agreement Motorola and Thomson 1978 Components and integrated Technological, industrial, and circuits commercial agreement /rite/ and Matra 1981 Integrated circuits Technological, industrial, and commercial agreement Thomson and RCA 1971 Color tubes Technological agreement Rhone-Poulnec and Siltec 1983 Silicon Joint venture Rhone-Poulnec and Dysan — Magnetic disks Technological agreement Sagem and Motorola — Bubble memories Technological agreement Thomson and Diasonic 1983 Medical instrumentation Technological and commercial agreement aThe technologically dominant partner is italicized. SOURCE: Office of Technology Assessment. Compiled from Research and Deve/opmertf /n Electronics.’ USA-France, 1982/7983, French Telecommunications Council, 1984. now owns Stromberg-Carlson. In addition, U.S.-manufactured computers. Moreover, France and the United States have arranged when Amdahl had difficulties raising capital joint ventures whereby American semiconduc- for expansion, Fujitsu bought 40 percent of tor firms have exchanged technical know-how Amdahl Although Amdahl might have been for access to the French markets-particularly sold to another American firm, Amdahl man- to the French telecommunications market agement preferred to sell its stock to Fujitsu (normally well protected by the French PTT). in order to facilitate technology exchange These joint ventures in which the French part- agreements, which involve cross-licensing, ners hold controlling interests include Thom- financing, and information exchange.8 Fujitsu son and Motorola, Saint-Gobain and National Ltd. has also recently agreed to supply Texas Semiconductor (in a firm named Eurotechni- Instruments with gate array technical know- que), and Matra and Harris. how; Texas Instruments will produce the Jap- anese gate arrays and ship them back to U.S. companies also have technical ex- Fujitsu. change agreements and other business relationships with Japanese firms. Intel Corp., for Other joint technical agreements between example, has a 5-year cross-licensing, cross- American and Japanese information technol- compatibility, and technology exchange agree- ogy firms include Sperry’s high technology co- ment primarily in the area of controllers and operative agreement with Mitsubishi, which peripheral equipment with NEC Corp. Amdahl covers joint activities in manufacturing, re- Corp. currently uses a semiconductor chip de- 8Patricia Keefe, “Many U.S. Firms Have Japanese Ties, ” veloped and manufactured by Fujitsu in its ComputerWorld, May 2, 1983, p. 73. 210 ● Information Technology R&D: Critical Trends and Issues

search and development, and marketing of Table 34.-Number of U.S. Patents Granted to computer systems. Mitsubishi also has a joint Selected Foreign Countries” in All Product Fields and in Communications Equipment and Electronic agreement with IPL Systems to develop an Components (1963-81) IBM-compatible processor. The technical exchange agreement between the two firms com- Communications equip- bines Mitsubishi’s” computer-aided design and ment and electronic Country of inventor All fields components large scale integration technology with IPL’s United States . . . . . 865,124 101,914 design expertise. Under the agreement, both Foreign ...... 369,519 41,242 firms are granted the right to market the West Germany . . 91,359 7,850 jointly developed products. Mitsubishi and Japan ...... 77,450 13,013 United Kingdom . 51,138 5,976 Westinghouse have also arranged a joint ven- France ...... 35,244 4,455 ture to design and manufacture integrated cir- Switzerland . . . . . 21,622 1,258 cuits. In addition to technical exchange agree Canada...... 20,241 1,773 Sweden ...... 13,368 1,007 ments between private firms, the U.S. Depart- Italy ...... 11,958 847 ment of Defense encourages Japan to trans- Netherlands . . . . 11,103 2,701 b fer defense-related electronics technologies to U.S.S.R...... 5,111 454 Belgium ...... 4,459 360 the United States. In 1981, for example, the Austria ...... 4,080 273 U.S. Government asked Japan to provide ad- Australia ...... 3,585 198 vanced very large scale integration (VLSI) Denmark ...... 2,520 161 Mexico ...... 1,075 21 technology to enhance air and antisubmarine Other foreignc. . . 15,206 895 defense capabilities. Total ...... 1,234,643 143,156 %ountrles were selected on the basis of being in the top 10 of at least one of the Standard Industrial Classifications. blndicates ranking among the top five foreign countries in this PafllCLIlar Prod- Patents uct field. cother foreign Includes patents granted to foreign COUntrieS not shown A large number of U.S. patents are granted separately. SOURCES: Office of Technology Assessment; compiled from information in Of- to foreign individuals and corporations (see fice of Technology Assessment and Forecast, U.S. Patent and table 34). Foreign patenting activity in the Trademark Office, Indicators of the Patent Output of U.S. /ndustry /V(fW3+31), 19S2; in Science Mlcafors, 19S2, National Science Board, United States has been related both to in- National Science Foundation, 19S3, creased foreign inventive activity and to a growing interest in the U.S. information tech- Table 35 and figure 38 show that Japan has nology market. Moreover, studies have shown the largest number of foreign U.S. patents in that foreign patenting activity in the United communications equipment and electronic States by selected OECD countries correlates components, although West Germany had significantly with industrial R&D in those been the foreign leader in this field through- countries. This correlation is especially high 9 out the 1960s and mid-1970s. Since 1970, Ja- in the electrical and electronics industries. pan has doubled its patent activity in commu- Foreign patenting in the communication nications equipment and electronic comp- equipment and electronic components cate- nents, food and kindred products, primary gory was as much as 40 percent of the total metals, and professional and scientific in- number of U.S. patents granted during 1979- struments.11 Data for the period 1970-81, pre- 81, while the percentage of U.S. owned foreign sented in table 36, show that the percentage patents in the same category was 13 percent.l0 of U.S. patents in information technology —.—— areas decreased more than 20 percent while 9Keith Pavitt, “Using Patent Statistics in Science Indicators: the Japanese share of U.S. patents increased Possibilities and Problems, ” Z%e Meani”ng of Patent Statistics, 12 National Science Foundation, 1979. by over 2oo percent. *’The fields that have relatively high percentages of U. S.- owned foreign patents are the areas corresponding to U.S. direct investment and research activity abroad. It is possible that *’Science Indicators, 1980, National Science Board, National U.S. laboratories abroad supported R&D that resulted in Science Foundation, 1981, p. 21. patented innovations. Sa”ence hh”cators, 1982, National Sci- IZThe Office of T~hnology Assessment and Forecast, U.S. ence Board, National Science Foundation, 1983, p. 14. Patent and Trademark Office. Ch. 7—Foreign Information Technology Research and Development ● 211

Table 35.—Share of Foreign Patenting in the United States for the Three Most Active Countries by Selected Product Fields (1981)

Total West United Other Product field foreign Germany Japan Kingdom foreign Percent of foreign Chemicals, except drugs and medicines . . . . . 100 28 27 11 34 Drugs and medicines ...... 100 23 22 14 40 Nonelectrical machinery ...... 100 26 27 9 38 Electrical equipment, except communications equipment ...... 100 21 37 8 34 Communications equipment and electronic components ...... 100 18 44 9 29 Motor vehicles and other equipment except aircraft ...... 100 26 34 9 31 Aircraft and parts...... 100 28 42 10 21 Professional and scientific instruments . . . . . 100 22 43 8 27 Number of patents Chemicals, except drugs and medicines. . . . . 5,338 1,520 1,452 566 1,800 Drugs and medicines ...... 1,288 300 288 182 518 Nonelectrical machinery...... 8,166 2,088 2,240 731 3,107 Electrical equipment, except communications equipment ...... 2,541 535 952 202 852 Communications equipment and electronic components ...... 3.027 534 1,338 279 876 Motor vehicles and other transportation equipment except aircraft ...... 1,652 429 563 151 509 Aircraft and parts...... 777 216 323 75 163 Professional and scientific instruments . . . . . 4,100 892 1,760 329 1,119

SOURCE: Office of Technology Assessment and Forecast, U S Patent and Trademark Office, Indicators of Patent Output of U.S Industry (1%53-81); in Science Indicators, 1982, National Science Board, National Science Foundation, 1983.

Figure 38.—Share of Foreign Patenting in the United from 45 to 37 percent in the United Kingdom states for the Three Most Active Countries (1981) and from 32 to 28 percent in France. Overall, Communications equipment from 1971 to 1981, U.S. patenting in Canada, and electronic components Japan, and in the European Economic Com- munity declined approximately 40 percent.

Scientific and Technical Literature U.S. utilization of foreign engineering and technical literature is growing. Between 1973 and 1980, U.S. citations of foreign research findings in engineering and technology fields increased by 4 percentage points and by 7 percentage points in the field of mathematics. Al- though the U.S. utilization of foreign research has grown, U.S. use of foreign research

SOURCE. Science /rrd/carors, 1982, National Science Board, National Sctence literature is lower than other nations’ use of Foundation, 1983 foreign research literature.14 The number of jointly authored articles by While the foreign patenting activity in the scientists and engineers from different coun- United States has been increasing, U.S. patent activity abroad has been decreasing over —– —- ‘3Sa”ence In&”cators, 1982, National Science Board, National the past decade. Over the past 10 years, the Science Foundation 1983, p. 15. U.S. proportion of foreign patents decreased “Ibid, p. 12. 212 ● Information Technology R&D: Critical Trends and Issues

Table 36.—U.S..Owned and Foreign-Owned Figure 39.—index’ of International Cooperative U.S. Patents in Information Technologiesa Research by Country (Percent) Percent ownership U.S. patent Percent o 5 10 15 20 25 30 35 40 45 50 ownership 1970 1981 change I I I I I [ I I I I 1 United States ...... 76 58 –23.7 United Japan ...... 6 19 +216 States United Kingdom ...... 4 4 0 France ...... 3 4 +33 Japan West Germany ...... 5 8 +60 alnformation technologies included here comprise SIC 357—Office c0mPutin9 and accounting machines; and SIC 365-367—communications equipment and U.S.S.R. electronic components. SOURCE: The Office of Technology Assessment and Forecast, U.S. Patent and Trademark Office. Information technology numbers were calculated France from data developed under the support of the National Science Founda- tion, Science Indicators Unit. tries is also increasing. International co- Canada authored engineering and technology-related United articles as a percentage of institutionally co- Kingdom authored articles have risen from 13 percent West in 1973 to 16 percent in 1980. In 1980, more Germany than 40 percent of all jointly authored articles in mathematics were international collabora- lobtained by dividing the number of all articles which were Written by sCif3n- tive efforts.16 Moreover, the United Kingdom, tists and engineers from more than one country by the total number of articles jointly written by SIE’S from different organizations regardless of the country France, and West Germany had a greater per- involved. centage of internationally co-authored articles NOTE: Based on the articles, notes, andreviews in over 2,100 of the influential journals carried on the 1973 Sc/errce Cltatforr Index Corporate Tapes of (as a percentage of all institutionally co-authored the Institute for Scientific Information. articles) than the United States. Japan and the SOURCE: Science Indicators, 1982, National Science Board, National Science United States had the lowest percentages of Foundation, 1963. internationally co-authored articles. (See figure 39.) remain in the United States (if permitted), a large number of them return to their native Science and Engineering Students countries. Although the number of foreign graduate The number of foreign students in scientific science and technology students in U.S. univer- and technical fields in U.S. universities is in- sities has been increasing, the number of U.S. creasing. In mathematics and computer sci- students enrolled in technical programs at for- ence the number of foreign students enrolled 16 eign universities has been decreasing. The in U.S. universities was 22,620 in 1981 -82. number and percent of U.S. graduate students (See table 37.) Table 38 illustrates the large studying abroad was highest during 1971, but proportion of doctoral degrees awarded to for- now only constitutes about 1 percent of U.S. eign students in mathematics and computer graduate students.18 The decline of U.S. grad- science during 1981. In 1982 non-U.S. citizens uate students studying in foreign universities were awarded 38 percent of the 542 doctorates could be attributed to employment considera- in electronics and electrical engineering and tions and cost of living differences. Currently, 54 percent of the 72 doctorates in computer l7 as other industrialized nations’ technical ca- science. Although some of the foreign engi- pabilities improve, the low number of graduate neering and mathematics students choose to students abroad could inhibit the U.S. ability to keep abreast of the latest foreign research 161bid, p. 31. methods and developments. la~”ence ~~”cator9, 1980, National Science Bored, Nation~ Science Foundation, 1981, p. 240. “Science Inculcators, 1982, National Science Board, National 17’’ Washington Newsletter,”Ehwtrom”cs, Jan. 12, 1984, p. 70. Science Foundation, 1983, p. 29. Ch. 7—Foreign Information Technology Research and Development • 213

Co r-a m

0 000 0 . -Coma). n U)” N-.--- . .

c ) r- d -c

0 0 000 o

0“ NNT-

I 0003 .(=.=

-c

m c--J

0 000000 0 -m -----

0 000

Co “

. m C--J

----.-- m-c-+

o- mm-o

C-J

) n ?) c-o k -J J 214 ● Information Technology D: Critical Trends and Issues

Table 38.—Doctoral Degreesa Awarded to Foreign Students as a Percent of All Doctoral Degrees from U.S. Universities by Field 9-81)b

Field 1959 1963 1967 1971 1975 1979 1981 Science and engineering ...... 14.8 15.5 17.5 18.7 22.1 21.1 22.1 Physical sciences ...... 12.6 14.0 15.7 16.7 22.9 21.0 22.0 Physics and astronomy ...... 14.9 14.5 16.9 18.6 27.6 25.7 26.1 Chemistry ...... 10.5 12.2 14.6 15.6 19.8 19.7 21.2 Earth sciencesc ...... 17.1 19.9 17.2 14.6 22.1 16.6 17.1 Mathematical sciences...... 13.4 15.2 14.6 17.5 24.3 25.5 30.8 Mathematics ...... A NA NA NA NA 26.7 33.7 Computer sciences ...... NA NA NA NA 21.3 26.3 Engineering ...... 24.5 20.8 23.7 29.8 42.1 46.8 51.5 Life sciences ...... 17.6 17.5 19.6 18.2 19.5 16.5 19.8 Biological sciences...... 15.5 16.5 16.2 14.3 14.9 12.1 11.1 Agriculture and forestry ...... 24.9 21.0 32.4 33.6 37.4 35.2 37.6 Social sciences ...... 10.7 11.6 13.5 13.7 13.7 12.9 13.0 Psychology ...... 5.5 4.0 4.4 5.6 5.8 4.0 3.9 Other social sciences...... 15.5 17.6 19.9 19.3 20.2 22.4 24.0 Conscience total ...... 6.0 6.5 7.4 8.0 8.7 10.1 11.0 All fields ...... 11.7 12.3 14.0 14.4 16.2 16.1 17.2

apercen t of those whose citizenship is known. bFiscal year of doctorate, clncludes oceanography. NA—Not available.

SOURCE: Doctorate Record File, Special Tabulations, unpublished data; National Science Foundation, ScierIce Indicators, 1982, National Science Board, National Science Foundation, 1983

Implications for U.S. Information Technology R&D Policies

Given the strong links between other na- these industrialized nations have coordinated tions’ information technology research and de- their R&D efforts at a national level. Indus- velopment activities and those of the United trial and governmental cooperation is viewed States, and economic competition from coun- as a means to achieve common national tech- tries such as the United Kingdom, France, and nological objectives and enhance the competi- Japan in information technology innovation tiveness of the entire national information and international markets, foreign R&D ini- technology industry in global markets. Coordi- tiatives are a concern for the United States. nation for R&D includes efforts at the national For the U.S. Government to participate in in- level to disseminate information on technologi- ternational research and development and, at cal developments, share research results, and the same time, successfully compete with divide research activities among enterprises. these nations’ R&D initiatives, the United Moreover, in coordinating R&D efforts be- States will need to understand other nations’ tween government, university, and industry economic and social goals for technological de- participants, these nations have attempted to velopment, government, and industry roles in link trade competitiveness strategies more information technology R&D, and the signifi- closely with R&D policies.lg cance of targeted national information tech- In the United States, competition in technology R&D programs. nological development among firms is viewed Although the philosophy of industrial competition is prevalent in other economies-par- ‘gWilson P. Dizard, “U.S. International Information Trade, ” ticularly in the United Kingdom and Japan– The Information Society, vol. 2, No. 3/4, 1984, p. 189. Ch. 7—Foreign Information Technology Research and Development ● 215 as a variant of other forms of competition, and activities as models should be carefully such as competition in production efficiency, considered. As a result of major differences in product quality, and marketing. Consequent- the historical, cultural, and economic charac- ly, the United States relies heavily on open teristics of each nation, there appear to be dif- market competition and private initiative to ferences in their respective approaches to spur industrial R&D.20 Moreover, trade com- science and technology policies, as well as gov- petitiveness factors are not always considered ernment and industry participation in infor- in the formulation of U.S. R&D policy. mation technology R&D. Moreover, the United Kingdom, French, and Japanese na- Since other industrialized nations target in- tional information technology research and de- formation technology research and develop- velopment programs differ in overall goals and ment programs at the national level, the ques- organization and vary significantly from cur- tion has been raised whether the United States rent U.S. R&D efforts. should adopt a similar national industrial strategy for technological development. Al- though a number of nations that pursue coordinated industrial policies have relatively Science and Technology Policy Goals weaker overall economic performances than the United States, those that target informa- Although the United Kingdom, France, and tion technology as a national priority may im- Japan each developed science and technology prove their competitiveness. policies in part to strengthen and modernize their economies, each of these nations varies In response to foreign coordinated national in its conception of what technology policy R&D programs, a number of legislative op- should comprehend and what its objectives tions (several of which are modeled on foreign should be.22 In some nations policy is aimed initiatives) for coordinating and targeting in- at strengthening the competitiveness of tar- dustrial sectors have been proposed in the geted industries. Other nations are concerned United States. There is also evidence, pre- with developing information technologies for sented in this report, that the United States social needs or national security applications. may already be responding in ways particu- In other nations, the perception of technology larly suited to its social, governmental, and policy is much broader, and constitutes a part economic traditions. For instance, the govern- of a more general plan of how the economy ment-industry technology transfer activities should be structured in the future. Many of stimulated by the Stevenson-Wydler Technol- the goals for science and technology policy of ogy Innovation Act of 1980 (Public Law 96- the United Kingdom, France, and Japan are 480) (described inch. 2) and the recently form- rooted in history. At the finish of World War ing university-industry cooperative joint ven- II, each of these nations’ societies and econo- tures (described in ch. 6) may be indications mies were severely damaged. These nations’ that the United States is beginning to develop governments therefore perceived a need to ac- indigenous mechanisms to pursue common 21 tively promote the growth of a high technol- technological objectives. ogy industry in order to aid their ailing If, however, the United States is to develop economies. coordinated national industrial strategies in Since World War II and particularly in more response to other nations’ targeted efforts in recent times, science and technology policies the area of information technology, the use of have become increasingly politicized. This re- other nations’ national R&D organizations flects the view that science and technology is linked to nations’ economic well-being (e.g., —— ——— trade, productivity, and employment) and so- ‘“Jack Baranson and Harold B. Malmgren, “Technology and Trade Policy: Issues and an Agenda for Action, ” Bureau of In- cial welfare (e.g., quality of life, education, and ternational Labor Affairs, U.S. Department of Labor, and the training). Moreover, the widespread belief that Office of the U.S. Trade Representative, 1981, p. 5. “See for example, Jan Johnson, “America Answers Back, ” Datamation, May 15, 1984, p. 4057. “Jack Baranson and Harold Malmgren, op. cit., p. 6.

38-802 0 - 85 - 1 5 216 ● Information Technology R&D: Critical Trends and Issues

“the nation that dominates the information French governmental interest and efforts to processing field will possess the keys to world expand advanced information technology re- leadership in the twenty-first century, ”23 has search and production are directed at two ma- caused nations to look to the development of jor goals: strengthening France’s international information technology for their future well- competitiveness and the development of an in- being. Evidence of the movement of science formation technology-based infrastructure for and technology policies into the political arena, the preservation and continued development for instance, can be found in both the recent of French culture and society. Reflecting Thatcher (United Kingdom) and Mitterand French national goals, France’s national infor- (France) political platforms in which informa- mation technology research and development tion technology research and development program, La Filiere Electronique, has as its funding and programs were emphasized. How- long-term goals: to place France on a techno- ever, these two leaders differ in their policies logical level closer to that of the United States for technological development. These differ- and Japan; create a trade surplus in informa- ences are important in understanding and tion technology products; create new jobs; evaluating the roles of these governments in assure a sound technological base; and accel- information technology research and devel- erate the production of information technol- opment. ogy products. The various current approaches to science The Europeans, through the European Eco- and technology policy and the different objec- nomic Community (EEC), are also concerned tives of each nation’s research activities under- about their basic economic survival in world lie the distinct goals of each country’s national markets. Consequently, Europe’s major goal research program. For example, Japan’s con- is to establish a strong technological base cerns lie in developing information technology through collaborative research on various for improving Japanese society (which entails long-range projects that may not be adequate- developing an information-based infrastruc- ly funded within individual nations. The Euro- ture) and improving its world trade position pean Strategic Program for Research in Infor- in information technology products. Conse- mation Technology (ESPRIT), involves the 10 quently, Japan’s goals for its national infor- EEC member countries. ESPRIT’s major ob- mation technology research program, the jective is to keep Europe competitive with the Fifth-Generation Computer Systems Project, United States and Japan in advanced infor- is to develop a fifth-generation computer for mation technology fields. social applications and to develop a technological knowledge base which will enable Japan to maintain and improve the volume of information technology exports that is so vital to Government Role in Information the Japanese economy. Technology Research and Development Like Japan, the United Kingdom is also con- The level of government funding for re- cerned with its economic survival in world search and development varies widely. The markets, as exports also play an important ratios of civilian research and development ex- role in the U.K. economy. The basic goal, penditures to gross national product (GNP) therefore, for the U.K. Programme for Ad- presented in figure 40 show that the United vanced Information Technology is to improve States devotes a lower proportion of its GNP the competitiveness of the U.K. information to civilian R&D than Japan, but a higher pro- technology industries in the world market in portion than the United Kingdom. An exami- order to reverse its negative balance of infor- nation of the annual growth rates of national mation technology trade. research and development expenditures for ‘gRobert E. Kahn, “A New Generation in Computing, ” 1~~~ electrical and electronics industries reveals Spectrum, November 1983, p. 36. that the United States lagged behind most of Ch. 7—Foreign Information Technology Research and Development ● 217

Figure 40.— Estimated Ratio of Civilian R&D the United States in the extent to which each Expenditures to Gross National Product (GNP) for a coordinates and targets these policies to sup- Selected Countries port information technology research and de- 2.6 velopment. Policy measures for R&D support 2.5 include low interest loans, direct subsidies, or

2.4 actual government contracts. Other incentives to stimulate information technology R&D in- 2.3 clude tax incentives, national development 2.2 banks which channel funds specifically to in- 2.1 formation technology industries, public sector 2.0 procurement, and merger/antitrust policies. 1.9 To varying degrees the governments of the ● ☛ ● ☛ 1.8 United Kingdom, France, and Japan have established institutional mechanisms to facili- 1.7 tate coordinated research and development 1.6 policymaking. Coordinated government inter- 1.5 vention on the part of these industrialized na- 1.4 tions is reflected in each of the governments’

1.3 structure which centralizes the responsibility within one or a few government ministries, and 1.2 in the establishment of planning councils. 1.1 Moreover, the coordinated efforts of the Brit- 1.0 ish, French, and Japanese extend much fur-

.9 ther than the U.S. pluralistic and decentralized

o~ approach to R&D support, to the coordination 1961 63 65 67 69 71 73 75 77 78 81 83 of government, industry, and university infor-

Years mation technology research and development Natlonal expenditures excluding Government funds for defense and sPace objectives and activities. R&D SOURCE Scierrce Indicators, 7982, National Science Board, National Science Foundation, 1983 Government/Industry/University Institutional Arrangements for the industrially developed nations during the 1970s.24 Information Technology Research and Development The United Kingdom, France, and Japan singled out information technology as an area A major difference between the institutional for government promotion and support. This arrangements for U.S. information technology contrasts sharply with the U.S. Government, research and development and those of the which at the present time has not singled out United Kingdom, France, and Japan is there- or targeted any specific industry or technol- lationship between government and industry. ogy for government support. Although in mar- In the United States, government tends to be ket economies governments use a more or less viewed as a regulator and enforcer of laws and standard set of policies to support research social policies; U.S. industry generally per- and development,25 other nations differ from ceives its relationship with government as ad- ZtR,easOns ~or the relative declining growth rate Of us. GOV- versarial. Moreover, the U.S. Government, ernment R&D funding during this period include major de- which traditionally avoids involvement in the creases in funding for defense and spacewsearch and develop- private sector, relies largely on private initia- ment, while civilian research and development was held tives for risk-taking, innovation, job creation, constant. Curnmtly, U.S. budget projections reflect increasing R&D funding for defense purposes. and the generation of profits and capital. ‘Slnternational Competitiveness in Electronics (Washington, DC: U.S. Congress, Office of Technology Assessment, OTA- Although this adversarial relationship be- ISC-200, 1983), p. 381. tween the public and the private sectors ex- 218 ● Information Technology R&D: Critical Trends and Issues

Table 39.—Average Annual Growth Rates in the Engineering Industry

United United States Japan Germany France Kingdom 1970-79 1970-79 1971-79 1970-79 1969-78 Research scientists and engineers: a Aerospace ...... 0.3 — –5.3 1.1 –2.6 Electrical and electronics . . . . –0.4 6.7 3.2 4.1 0.9 Machinery ...... 5.1 2.5 6.7 3.9 3.6 Transport equipment ...... 3.3 8.5 3.8 1.8 0.2 Total expenditure: Aerospace ...... – 1.1 — –2.3 2.6 –0.5 Electrical and electronics . . . . 0.1 5.8 5.8 4.7 3.1 Machinery ...... 5.5 5.2 9.8 0.9 1.5 Transport equipment ...... 5.4 11.0 3.9 7.2 –0.3 Government funds: Aerospace ...... – 1.7 — –1.0 1.8 –3.1 Electrical and electronics . . . . –2.2 — 7.1 0.1 9.9 Machinery ...... 1.0 — 10.0 — 1.1 Transport equipment ...... 3.1 — 10.1 6.9 3.0 industry and foreign funds: Aerospace ...... 0.6 — 4.1 5.0 (b) Electrical and electronics . . . . 2.3 — 4.9 7.7 –0.7 Machinery ...... 6.3 — 8.5 — –3.4d Transport equipment ...... 5.9 — 3.3 7.4 – 1.4 a~aPan not in FU1l.Tirne Equivalents. bLarge increases from a very IOW start. cFrom abroad 4.91 per annum. dElectrical and electronics, For R&D purposes the following subclasses are identified with this subgroup: ISIC 3832 and ISIC 638 nec. SOURCE: OECD Science and Technology Indicators, OECD, 1964. ists in most countries, the governments of the To coordinate more fully both their targeted United Kingdom, France, and Japan have his- policies and information technology R&D, the torically participated to a greater degree in in- United Kingdom, France, and Japan have re- dustrial and technological development. For cently adopted major national programs. cultural, historical, and political reasons, these These programs, which have no counterpart industrialized nations often regard govern- in the United States,27 have been established ment-industry relations as a partnership or to pursue research projects cooperatively be- perceive government as an institution for tween government, private industry, and uni- guiding and supporting targeted industries. versities. The scale of funding for all the na- Government-industry coordination has begun tional programs is large and roughly in some nations with the establishment of mul- comparable, representing a major commit- tipartite advisory groups representing government of between half a billion and several bil- ment and private industry.26 Formal and in- lion dollars over the first 5 years.28 This fund- formal advisory councils have become in some ing is magnified because the companies which nations, particularly in Japan, important for- receive government research funds usually are ums for government-industry-academic con- required to match the funds. sultations on industry policy and implementation. Particularly in information technology, these councils orchestrate joint research, de- “Some individuals contend that the Department of Defense velopment, and marketing schemes among in- (DOD) programs, such as VHSIC, are similar to the national formation technology firms and government. research and development programs of the United Kingdom, France, and Japan. However, there are major differences be- ‘Franklin Delano Strier, “On Economic Plannin g, Japan and tween DOD programs and these other nations’ programs. West Germany Have a Better Idea,” The Center Magazine, Jan- 28Trudy E. Bell, “Tomorrow’s Computers-The Teams and The uary/February 1984, p. 36. Players,” IEEE Spectrum, November 1983, p. 46. Ch. 7—Foreign Information Technology Research and Development ● 219

Each nation, in structuring its cooperative as models for future research projects? What research program, has looked to successful are the relative strengths of these new re- organizational examples both domestically search programs versus traditional U.S. re- and in other countries, borrowing organiza- search environments? To what degree will tional concepts and applying them in innova- these new national research programs affect tive ways. For example, the Institute for New U.S. technological development and market Generation Computer Technology (ICOT), share?29 with its central research center, is unique to Japan. ICOT is organized for close coopera- Industry Participation in Information tion of research activities among government, Technology Research and Development industry, and university participants. Japan’s customary approach has been to have each Funding of industrial information technol- participating research institution or company ogy research and development varies from na- conduct research individually. ICOT is also tion to nation. Table 40 represents industrial contracting with outside companies and lab- R&D funding patterns in industrialized na- oratories for some of the research and devel- tions and shows that industry has been a more opment—a technique often used for U.S. de- dominant contributor in Japan (although the fense contracts, but unusual in Japan. Japanese Government also provides a great deal of support through indirect subsidies) The institutional arrangements for the U.K. than in any other nation, including the United Programme for Advanced Information Tech- 30 States. 3’ The U.S. Government, in contrast, nology are also unique, although the Alvey supported approximately half of the U.S. re- Committee closely modeled the program orga- search and development activities in 1970. nization on Japan’s Ministry of International However, by 1979, U.S. industry had increased Trade and Industry (MITI) and the U.S. De- its share of industrial funding to 67 percent, fense Advanced Research Projects Agency approximately as much as the French indus- (DARPA). This new organization is intended try’s 71 percent support of its own research. in part to stimulate the transfer of basic research from academic environments to in- Table 41 illustrates industrial research and dustry. development support for electrical and electronics and computers categories in the indus- The ESPRIT program shares both similari- trialized nations. Japan and West Germany ties and differences with organizations such had the highest proportion of industrial R&D as ICOT and the Alvey Programme. Like in the electrical and electronics category. In these other new institutional arrangements, the computer category, however, the United ESPRIT research is undertaken by teams of States had the greatest proportion of indus- university, government, and industry scien- trial R&D followed by the United Kingdom, tists. In contrast to other nations’ cooperative France, and then Japan. programs, ESPRIT represents a joint venture at the international level in which each project must involve researchers from at least two countries. The success of these national programs in achieving both national aims and research goals remains undetermined. However, these new institutional arrangements raise some interesting questions for the United States. How ‘gIbid. will the cooperative research programs of the 300ECD Science and Technology Inal”cators, OECD, 1984, p. 119. United Kingdom, France, and Japan alter 31 Sa”ence Indicators, 1982, National Science Board, National their traditional research structures and serve Science Foundation. 1983.,* K). 9. 220 ● Information Technology R&D: Critical Trends and Issues

Table 40.—R&D Performed in the Business Enterprise Sector by Source of Funds (1970, 1975, and 1979)

National currency (in millions) Percent Country and source 1970 1975 1979 1970 1975 1979 — — — France ...... 8,322.4 15,616.5 26,260.0 100.0 100.0 100.0 Total domestic ...... 8,007.3 14,393.5 24,460.0 96.2 92.2 93.1 Business enterprise ...... 5,310.0 9,965.8 18,723.0 63.8 63.8 71.3 Government ...... 2,689.0 4,376.8 5,674.0 32.3 28.0 21.6 Private nonprofit ...... 4.6 47.2 58.0 .1 .3 .2 Higher education ...... 3.7 3.7 5.0 — — — From abroad...... 315.1 1,223.0 1,800.0 3.8 7.8 6.9 Japan ...... 895,020.0 1,684,847.0 2,664,913.0 100.0 100.0 100.0 Total domestic ...... 894,193.0 1,683,200.0 2,662,698.0 99.9 99.9 99.9 Business enterprise ...... 876,608.0 1,654,502.0 2,624,843.0 97.9 98.2 98.5 Government ...... 17,585.0 28,698.0 36,807.0 2.0 1.7 1.4 Private nonprofit ...... NA NA 935.0 NA NA — Higher education...... NA NA 113.0 NA NA — From abroad...... 827.0 1,647.0 2,215.0 .1 .1 .1 United Kingdoma ...... 680.3 1,340.2 2,324.3 100.0 100.0 100.0 Total domestic ...... 647.7 1,255.5 2,138.8 95.2 93.7 92.0 Business enterprise ...... 431.2 841.4 1,459.0 63.4 62.8 62.8 Government ...... 216.5 414.1 679.7 31.8 30.9 29.2 Private nonprofit ...... NA NA NA NA NA NA Higher education...... NA NA NA NA NA NA From abroad...... 32.6 84.7 185.5 4.8 6.3 8.0 United Statesb ...... 18,067.0 24,187.0 38,226.0 100.0 100.0 100.0 Total domestic ...... 18,067.0 24,187.0 38,226.0 100.0 100.0 100.0 Business enterprise ...... 10,288.0 15,582.0 25,708.0 56.9 64.4 67.3 Government ...... 7,779.0 8,605.0 12,518.0 43.1 35.6 32.7 Private nonprofit ...... — — — — — — Higher education...... — — — — — — From abroad...... — — — — — — West Germanyc ...... 7,114.0 14,469.0 20,720.0 100.0 100.0 100.0 Total domestic ...... 7,090.0 14,005.0 20,070.0 — 96.8 96.9 Business enterprise ...... 6,146.0 11,397.0 15,650.0 — 78.8 75.5 Government ...... 939.0 2,596.0 4,400.0 — 17.9 21.2 Private nonprofit ...... 5.0 12.0 20.0 — .1 .1 Higher education ...... — — — — — — From abroad ...... 24.0 464.0 650.0 — 3.2 3.1 alg70figure9 forthelJnited Kingdom are from 1989, and 1979 figures are from 1978. bcurrent expenditures plus depreciation onlY.

c1970figures for West Germany are from 1969. NA—Not separately available. NOTE: Details may not add to totals becauseof rounding. SOURCES: OECD Scierrce artd Technology Indicators, VOLBOECD, 1982; Research arrd Development in/ndust~, National Science Foundation, 1981; in Science/n. dicators, 19S2, National Science Board, National Science Foundation, 1983.

Table 41 .—Percent of lndustrial R&D in Selected Industries (1989-79)

United States Japan West Germany United Kingdom France Industry 1970 1979 1970 1979 1969 1979 1969 1978 1970 1979 Six-industry total . . . . 71.2 70.2 61.5 55.6 71.1 72.4 58.4 60.7 NA 61.7 Aerospace ...... 11.8 8.6 – – .1 .6 1.1 5.4 7.8 8.8 Electrical and electronics ...... 19.5 17.6 24.9 23.4 29.3 26.9 20.4 16.0 16.5 20.2 Instruments ...... 5.3 7.8 2.3 2.9 1.6 2.1 3.1 1.9 NA 1.2 Machinery ...... 14.3 6.0 8.7 7.0 }7.0 }16.6 6.6 NA Computers ...... “ 11.5 2.9 2.8 !; 5.2 NA ::; Chemicals groupa . . . . . 20.3 18.7 22.7 19.5 33.1 26.2 22.1 25,6 24.2 23.4 alncludes chemicals and allied products and petroleum refining industries. SOURCES: OECD Science and Technology /nd/caters, vol. B OECD, 1982; Research and Development in Industry, National Science Foundation, 1981; in Science hr- dicators, 1982, National Science Board, National Science Foundation, 1983. Ch. 7—Foreign Information Technology Research and Development ● 221

In addition to direct funding, governments tions of a national government to define the can support or discourage industrial research structure of its information technology indus- and development with tax incentives, fiscal try will have profound but unpredictable influ- and monetary policies which affect interest ence on industrial R&D. For example, France rates and the availability of capital, regulatory has recently nationalized some major informa- policies, procurement practices, patent pol- tion technology firms, whereas the United icies, and antitrust policies.32 Moreover, ac- Kingdom and Japan currently are moving in ——— —.. —. the opposite direction by privatizing their tele- 32Science Indicators, 1982, op. cit., p. 10. communications entities.

Conclusions

Cultural, social, and institutional differences tional organizations such as OECD and the among nations profoundly influence the way ITU have initiated the development of uni- in which technological innovation occurs and form definitions and standards for information underlie the considerable differences in R&D technology products and facilitated the ex- policies in the United Kingdom, France, and change of information on nations’ R&D poli- Japan. Therefore, many of these nations’ suc- cies and activities, the U.S. Government cur- cessful research and development policies and rently does not have a designated agency or endeavors may not be easily transferable or office within an agency to analyze and moni- applicable to U.S. R&D environments. Never- tor foreign information technology R&D pol- theless, cross-cultural comparisons of these in- icies and practices. dividual nations, which have developed differ- The economic importance of industrial com- ent methods of addressing similar public petitiveness and its close reliance on techno- policy issues and technologies, can be useful logical development emphasizes the impor- to an individual society, such as the United tance of developing a program for periodic States, in devising a conceptual framework for mapping of the pattern of technological advan- developing information technology domestic tages and disadvantages relative to foreign and international R&D policies. 35 competitors. Such surveys could help to alert International comparisons also provide a government and industrial representatives to useful method for evaluating the status of U.S. changes in the technological underpinnings of information technology research and develop- their competitive positions. These surveys ment activities and expenditures. However, could also be broadened to encompass non- making comparisons is difficult. Differences technological factors affecting competitive- exist among countries in definitions, concepts, ness or future technological developments. data collection methodologies, and statistical These could include current or prospective reporting procedures.33 These problems are changes in foreign government R&D policies particularly prevalent in the area of informa- that influence the competitiveness of their pro- tion technology .34 Although several interna-

———— ——-—————— gories. Moreover, each nation has its own methods for categoriz- tsscjence ~n~.catom, 1980, National Science Board, Nation~ ing information technology products and research and devel- Science Foundation, 1981, p.4. opment expenditures and activities. 34The U.S. SIC Codes, for example, do not have categories 3%ee S. E. Goodman and M. R. Kelly, “We Are Not Alone: for information technology, but rather information technology A Sample of International Policy Challenges and Issues, ” The products are encompassed within different and separate cate- Information Society, vol. 2, No. 3/4, p. 250-268. 222 ● Information Technology R&D: Critical Trends and Issues ducers in export markets. Such a broader sur- Perhaps the most important observation vey could provide the basis for a more effec- [is] . . . that the general character of changes tive assessment of changes in the current and brought about by science and technology prospective competitiveness of foreign R&D tends, overall, to lead to increased interna- efforts than a narrow focus on technological tional interaction and integration, with cor- capabilities alone. 36 As one analyst stated: respondingly reduced relevance of national borders. The parallel spread and diffusion of Most governments, and most especially technological competence that is eroding the the U.S. Government, are organized to reflect dominance of one or a few nations in science primarily domestic interests and have great and technology makes it imperative that we difficulty in dealing adequately with one of recognize the nature of the underlying changes the consequences of science and technology— taking place as we attempt to develop pol- the gradual blurring of the distinction be- icies to deal with the specific implications of tween domestic and international affairs . . . . any given technology, or to influence the direction of development of technology itself.37 —- “Report by Eugene Skolnikoff entitled “Impact of Science “Bela Gold, “Technological and other Determinants of the and Technology on the International System, ” in “Overview International Competitiveness of U.S. Industries, ” JEh’E of Intimational Science and Technology Policy” hearings before Transactions on En&”neering Management, vol. LM 30, No. 2, Committee on Foreign Affairs, House of Representatives, 98th May 1983, p. 58. Cong., 1st sess., Aug. 2,3 and Sept. 21, 1983, p. 317.

Japan

Given Japan’s minimal domestic natural re- Europe. Following the enactment of the law source base and its high dependence on other on Foreign Capital in 1950, the Japanese have nations for food, energy, and raw materials, signed more than 36,000 licensing agreements the Japanese Government treats science and costing approximately $12 billion.38 Agree- technology policy as a means of spurring over- ments between Japanese and foreign firms all economic growth and enhancing Japan’s were made under strict government supervi- competitive position internationally. Japan’s sion partly to control the outflow of foreign policies focus on maintaining a long-term, high exchange and partly to concentrate technologi- volume of exports in order to gain technological resources into certain key industries. Prod- cal and, thus, market leadership in a broad ucts manufactured with these imported tech- spectrum of high technology, high value-added nologies initially served to develop the products. This perception in Japan has led to Japanese domestic market, bringing about a a consensus in the nation’s government, busi- GNP growth exceeding 10 percent through- ness, financial, and academic communities to out the 1950s. For example, transistor tech- continue strengthening the nation’s technolog- nology imported and commercialized by the ical base. Japanese in the early 1950s provided a foundation for the modern electronics industry. The coordination of science and technology policy to the promotion of economic develop- Although the general trend of importing ment is rooted in Japan’s postwar recovery ef- technologies has been receding, the imports of forts. During the postwar recovery period, var- technologies related to electronic computers ious science and technology institutions and increased by 16 percent over fiscal year 1980, laboratories were established on the assump- with those relating to software accounting for tion that they would help stimulate an eco- over 166 imports. Broken down into sectors nomic recovery. Another major component of ——. -——— this strategy was to import and improve tech- 3%mard Lynn, “Japanese Technology: Successes and Strat- nology from the United States and Western egies,” Current History, November 1983, p. 366. Ch. 7—Foreign Information Technology Research and Development . 223 of industry, out of a total of 2,142 cases, the Therefore, these military developments some- number of imports of foreign technologies dur- times result in expensive products which are ing fiscal year 1980 in the electrical industry so specialized that civilian or consumer ap- amounted to 413 cases.39 plications can be limited. This is often the case In concert with the policy of importing tech- with integrated circuit development in the nology, Japan has sought to develop its own United States. On the other hand, Japan has indigenous research base. The ratio of national succeeded in developing integrated circuit R&D expenditure to national income has risen products solely for commercial application. from less than 1 percent during the first half Taking into account all funds spent on de- of the 1950s to 1 percent in 1957, 1.5 percent fense, the government of Japan still contrib- in 1960, 2 percent in 1971, and 2.36 percent utes significantly less to total scientific re- in 1981. According to official plans, this per- search expenditure than other countries.44 centage will be increased to 2.5 percent by More specifically in the area of information 1985 and to 3 percent by 1990.40 According to processing, the Japanese Government R&D statistics released by the Japanese Prime Min- expenditure in 1979 accounted for 8.2 percent ister’s Office, all Japanese R&D, publicly and in 1979 and 6.2 percent in 1980 of the total privately funded, in the area of information Japanese information technology R&D ex- processing (which includes software and com- penditures.45 In Japan, this government R&D puter systems development only) totaled funding is concentrated in the national univer- Y 158.6 billion ($687 million) in 1979 and sities (13.5 percent), national research insti- Y 164.6 billion ($713 million) in 1980.4142 tutes (13 percent), with as little as 1.5 percent of government funding channeled to private In Japan, a far lower percentage of total re- industrial laboratories. Because government search funds is provided by government (Ja- R&D funding, which is the major supporter pan 27.7 percent, United States 51.1 percent, of academic basic research, is relatively lim- United Kingdom 51.7 percent) than in other 43 ited, reasons for Japan’s perceived ineffective- nations. In part, this difference can be attri- ness in basic research can be clearly under- buted to the high level of expenditure on de- stood. As a result, current improvements in fense research by Western governments (ap- the Japanese academic environments for basic proximately 15 percent of total research funds) research as well as increases in funding levels relative to the small amount spent by the Jap- for overall basic research are high priorities anese Government (0.7 percent). In some areas on the Japanese policy agenda. of information technology R&D such as integrated circuit development, low military R&D Because Japanese Government R&D fund- expenditure has helped Japanese industry. In ing is small and for the most part channeled general, military areas demand the highest into university and national research insti- state-of-the-art standards, regardless of costs. tutes, Japanese industry funds constitute approximately 70 percent of R&D activities. Ap- proximately 28 percent of all Japanese indus- — ‘g’’ Import of Foreign Technologies in Japan, ” Science and trial R&D funding is devoted to information Technology in Japan, April 1982, p. 27, technology R&D. Although the Japanese Gov- ‘“’’ Summary of fiscal year 1981 White Paper on Science and ernment does not directly make use of the Technology, ” Science and Technology Agency, Tokyo, Foreign Press Center, 1981. vitality offered by private enterprise, the lack “Barry Hilton, “Governme nt Subsidized Computer, Software, and Integrated Circuit Research and Development by Japanese Private Companies, ” Scientific Bulletin, Office of Naval Re- search Far-East, U.S. Department of the Navy, vol. 7, No. 4, 44These low expenditures are also the result of the Japanese October-December 1982. Governments current large budget deficits. 42All Japanese yen figures are converted into U.S. dollars ac- 46Barry Hilton, “Government Subsidized Computer Software, cording to foreign exchange rates as of June 1, 1984, where and Integrated Circuit Research and Development by Japanese’ +231 = $1. Private Companies, ” Scientific Bufletin, Office of Naval Re- “’’Science in Japan, “ Nature, vol. 305, Sept. 29, 1983, p. 361. search Far-East, vol. 7, No. 4, October-December 1982. 224 ● Information Technology R&D: Critical Trends and Issues of government funding has intensified compe- the Japanese privatized R&D efforts are high- tition in the area of information technology rely successful. Japan has clearly outstripped search and development. This competitive ef- most Western nations in processing technol- fort is exemplified in Japan’s computer and ogies and incremental engineering-rapidly semiconductor industries which, in order to refining existing designs and ideas by making survive, must develop and efficiently produce them smaller, lighter, faster, and cheaper. Jap- high quality products as quickly as possible. anese engineers, for instance, reengineered the 16 K RAMs with finer features to produce a This “privatized” environment for R&D has 64 K RAM within a 2-year period. given Japan advantages and disadvantages in its information technology R&D efforts. One Examples of Japanese strengths and weak- result of the large percentage of privately nesses in information technology R&D are well funded R&D is the lack of basic fundamental documented in the area of software develop- research activities. Because the R&D is sub- ment. Software development, currently be- sidized mainly by private firms, basic creative lieved to be crucial for future information tech- research, which is high-risk and long-term, is nology development, can be classified into sometimes ignored in favor of cost-efficient, various categories. Japan, with its industrial developmental, applied, commercialized R&D. emphasis on applied R&D, has concentrated This continued preoccupation with R&D efits efforts in production process-control soft- forts that bring quick economic results has ware which has wide commercial industrial ap- resulted in a trend which places less impor- plications and which will reap significant eco- tance on basic, innovative studies. The Japa- nomic benefits, both domestically in terms of nese Science and Technology Agency (STA), the productivity and capacity utilization of infor example, published a list of 15 basic dis- dustry, as well as internationally, in terms of coveries in the fields of recombinant DNA and the benefits of trade and technology transfer. computer technology (superconductivity, op- However, in other categories of software de- tical fibers, lasers, Josephson junctions, tun- velopment such as computer-aided design nel diodes, and transistors). Japan was respon- (CAD), the United States is technologically sible for only two of the breakthroughs listed; more advanced than Japan. Most of this tech- America for nine; the United Kingdom and the nological lead resulted from billions of dollars Netherlands for four. This bias in Japan’s that have been allocated for aerospace and de- overall research expenditures toward applied fense basic research. As a result of this basic research and prototype development is re- research for U.S. defense purposes, the U.S. flected both in government-supported R&D computer simulation models and 3-D design and private sector research expenditure (see programs are among the most sophisticated table 42). in the world. On the other hand, in the area of develop- Currently, Japanese industry is beginning ment, the application of basic research results, to experience some difficulties with its emphasis on appliedborrowed technology. Japan has been slowly catching up to western technologi- Table 42.— R&D Expenditure by Type of Activity cal innovations and has less input from foreign Basic Applied Development basic research patent licenses on which to base 1970 ...... 18.9 28.2 52.9 its refinements. Furthermore, Western firms 1974 ...... 15.0 21.7 63.3 are expressing a disinclination to sell patents 1975 ...... 14.2 21.5 64.3 1977 ...... 16.2 25.1 58.7 to Japan, as they see the reengineered Japa- 1978 ...... 16.6 25.1 58.4 nese products competing with their own prod- 1979 ...... 15.6 25.9 58.5 ucts. Moreover, Japanese firms that have SOURCE: Kagaku Gijulsu Hydron (Indicators of Science and Technology),Kagaku Gijutsu-Cho (Science and Technology Agency) 1981. Note: This table sometimes neglected basic research, have few- covers all R&D, public and private. er technological innovations worth offering Ch. 7—Foreign Information Technology Research and Development • 225

Western companies when they wish to inquire microelectronics. Over the last three decades, about the possibility of cross-licensing. In ad- there have been major shifts in Japanese condition to the fear of the decreasing amount of sumer electronics production. Figure 41 illus- innovative ideas which Japan can buy and per- trates the shift from the production of radios, fect and the fear of being excluded from future to television sets, to audio equipment, and fi- U.S. and other Western nations’ technical de- nally to videotape recorders. This production velopments, national pride is also forcing Ja- progression is particularly interesting in terms pan to put more effort into basic R&D. of technology because it not only illustrates the steady restructuring of an industry to As a result of some of these difficulties, the higher and more complex technologies, but Japanese Government is beginning to place also illustrates the changing position of Jap- more emphasis on basic research activities. anese information technology industry in This movement towards increased basic re- terms of global competition.46 search is reflected in both government and industrial R&D activities as well as in the cur- In each shift in Japanese consumer electron- rent Japanese Government’s institutional ics production, industry has been dependent mechanisms for influencing and funding indus- on foreign technological innovations. For in- trial research. Because the government mech- stance, Bell Laboratories supplied transistor anisms which influence and fund information technology, RCA licenses made Japanese color technology R&D are mostly aimed at promot- television production possible, and Corning ing R&D in the private sector (and many take Glass supplied glass tube technology. Perhaps the form of informal cooperation), it is diffi- more than any other of Japan’s industries, the cult at times to disassociate government and development of Japan’s consumer electronics private sector initiatives and roles in informa- industry is the result of imported technology tion technology R&D activities. However, for that competitive Japanese firms adapted, im- purposes of clarity and comparison, the role proved, and drove costs down. Figure 42 illus- of government, universities, and industry envi- trates the Japanese share of world consumer ronments for the conduct of information tech- electronics productions. The total value of pro- nology R&D will be separately described. Be- duction, second only to that of the United fore discussing these environments in detail, States, reached Y 8,683 billion ($37.6 billion) it is also important to understand the size of —or approximately 150 times that of 1955— Japanese participation in information technol- making electronics one of Japan’s major in- ogy markets. dustrial sectors. ‘James C. Abegglan, and Akio Etori, “Japanese Technology The Size of Japanese Participation Today, ” Scientific American supplement, 1983, p. J, 18. in Information Technology Markets Figure 41.—Trends in the Production Composition Utilizing its basic technology policy which Ratio of Major Consumer Electronics Equipment historically dictated that Japan reengineer imported technological innovations, Japanese industry has developed a very strong position in world information technology markets. In many areas where Japan has managed to capture a substantial percentage of the world information technology market, it can largely be attributed to Japanese industry’s strong capabilities in product development, market” -1951 1955 1960 1965 1970 1975 1980 ’82 ing strategies, and quality control. Years SOURCES: Annual Data on Japan’s Electronics Industry, 1983 Edition, Electronics Industry of Japan, Tokyo, 1983; in James C. Abeggian and Akio Etori, Beginning in the 1950s, Japanese informa- “Japanese Technology Today, ” Scientific American, supplement, tion technology industry efforts focused or 1983. 226 ● Information Technology R&D: Critical Trends and Issues

Figure 42.—JaDanese Share of World Production first to produce 256 K RAMs which will be of Consumer Electronics Products (1980) available in 1984-85. Figure 44 illustrates Jap- anese integrated circuit production relative to U.S. production. Perhaps the most significant step in this technology development sequence is the recent introduction by Japanese firms of one of the fastest supercomputers worldwide. These new computers, manufactured by Fujitsu and Hitachi, represent a major step in Japan’s government-sponsored national effort to build fifth-generation computers.”

Government The large percentage of private R&D funding would indicate the tremendous importance of Japanese industry in the Japanese success in world information technology markets. However, it is a mix of government support, a favorable and stable political structure, as well as freedom from national security expenditures, that have combined with the rather

SOURCES: Japan Electronics Industry Development Association; in Gene Adrian unique Japanese sociology to create a period Gregory and Akio Etori, “Japanese Technology Today: The Electronic of economic growth in the area of information Revolution Continues,” Scientific American, supplement, 19S4. technology. The nickname for the Japanese economy “Japan Inc., ” which was given some Recognizing the primacy of computers and years ago, may be said to be a realistic evalua- telecommunications as growth sectors, as well tion of the Japanese Government and private as conforming to the trend towards more ad- corporations during postwar Japan, when Ja- vanced technology production, Japanese in- pan sought to catchup with the industrially dustry has focused on a technology key to advanced nations. The term “Japan Inc. ” is these areas-integrated circuits. As in the case still used today but in most cases this word of consumer electronics, Japan imported basic appears to reflect a misunderstanding of the semiconductor technology and in the early relationship between the Japanese Govern- 1970s initiated the production of integrated ment and industry. circuits. Although exports were insignificant during the beginning production years, Japa- The Japanese Government does not control nese industry focused on lowering costs and industrial R&D through funding mechanisms improving quality. Integrated circuitry pro- or specific policies that must be adhered to, duction has grown in value terms at approxi- but rather there is a participatory partnership mately 25 percent per year. Figure 43 illus- among different segments of government and trates the growth in the Japanese information industry, based on pragmatic decisions, mu- processing, computer, and integrated circuit tual respect, working within a framework of industries between 1974-81. By 1976, Japan common goals. The councils and industrial accounted for a 40 percent share of the world associations have long been proposing to the market for 16 K RAMs, and in 1978, Fujitsu Japanese Government to increase its research Ltd. was the first to announce the commercial production of the 64 K RAM. Japanese com- 47 Phillip J. Hilts, “Japanese Firms Build Two Fastest Com- panies as well as U.S. manufacturers are the puters, ” The WAirJgton Post, Feb. 7, 1984, p. A,l.

Ch. 7—Foreign Information Technology Research and Development • 227

Figure 43.—Japanese Information Technology Industry Sales

2,535.6

Information processing industry 2,166.9 Computer industry I 2,104.0 669.8 IC industry 2,00(

596.6 1,651.8

1,34U.4 460.2

E ,295.6 1,123.0 412.6

1 ,00( 959.8 933.9 307.0 P 245.3 ! , 275.1 7

618.9 58Q.O 541.2

1974 1975 1976 1977 1978 1979 1980 1981 Years

NOTE: Information Processing Industry sales figures for 1981 not available. SOURCE: Survey on Specified Services, Production Statistics in Computer 1982 Japan Infer. Processing Center. and development support, but government intervention is decided on, it is undertaken funding has only increased 1.4 percent annual- through the development of a consensus ly over the past 10 years. Although the gov- among private enterprise and government. ernment does promote some industrial R&D, This consensual decisionmaking process be- industry has always been a larger investor. tween industry and government, frequently Government-industry relations in Japan have accomplished informally as well as through been broadly discussed, and it is often mis- formal institutional structures, is in many understood that such relations are largely due ways the most important factor affecting deci- to the Japanese Government subsidy of indus- sions on R&D projects and funding for infor- trial R&D. mation technology. Japan’s information technology firms are The major function of the Japanese Govern- fiercely competitive and the government’s role ment is to select, or to guide the selection of is seen as a means for providing an orderly technologies to be targeted, to reduce the eco- framework for coordinating private industrial nomic risks normally associated with develop- development. But when any coordination or ing new technologies, and to assist companies

228 ● Information Technology R&D: Critical Trends and Issues

Figure 44.—integrated Circuit (IC) Relative Production Share

SOURCES: Electronics, newspaper articles; in James C. and Akio “Japanese Technology Today, ” supplement, 1982 to achieve large scale production. The direct (NTT), and the Ministry of Finance (FOC) financial support for R&D provided by the through the Japan Development Bank (JDB). Japanese Government to targeted industries The major Japanese Government organiza- is in many instances less important than the tions directly involved with information R&D fact that the industry has been singled out by are illustrated in figure 45. the government as a “target” sector. There are tangible and intangible benefits which flow to Ministry of International Trade such industries. A targeted sector gains pres- and Industry (MITI) tige and public respect. Private banks are Perhaps the most misunderstood agency more willing to extend credit, customers and within the Japanese Government, the Minis- suppliers will tend to give preferred treatment, try of International Trade and Industry and government officials in various agencies (MITI) has been credited by many in the are also likely to be more responsive to the par- United States as being much more pervasive ticular needs of target-sector companies. than it is in reality .48 Taking into account the In addition to the informal consensual deci- scale of the Japanese economy, the complex- sionmaking, targeting and funding of informa- ity of international markets and the rapid tion technology R&D is accomplished through changes in technology, MITI alone cannot and major formal government institutions: Min- does not completely control industry. A case istry of International Trade and Industry in point is MITI’s failed attempt during the (MITI), Science and Technology Agency 1970s to consolidate the Japanese automobile (STA), Ministry of Education, Culture, and —— Science (MOE), Ministry of Posts and Tele- Tsuruta, “The Myth of Japan Inc.,” July 1983, p. 43-48, and Robert C. Christopher, communications (MPT) which has nominal “Don’t Overestimate Tokyo Industrial Aid,” The New York control over Nippon Telegraph and Telephone Times, Jan. 30, 1984, p. A, 21. Ch. 7–Foreign Information Technology Research and Development • 229

Figure 45.—Japanese Government Organization for Information Technology Research and Development

SOURCE: “Science in Japan, ” Nature, Sept. 29, 19s3. industries into three large groups. Some have ment. Despite this broad legal mandate, the claimed that MITI is on paper at least no more pervasive MITI practice of “administrative influential than the Department of Commerce guidance” by which many policies are imple- in the United States.49 mented depends on no statutory authority. Nevertheless, MITI does have the advantage MITI was established in 1949 with the of broad contacts across information technol- broad charter of shaping the structure of Jap- ogy industries and relies extensively on this anese industry, managing foreign trade and informal practice to influence firms and whole commercial relations, ensuring adequate raw industries in the direction it wants them to materials and energy supplies, and managing 5o take. relationships between particular business and technical industrial sectors and the govern- ‘OIra C. Magaziner and Thomas M. Hout, Japanese hfus- try PoL”cy (Berkeley, CA: Institute of International Studies, ‘g’’Science in Japan, ” Nature, vol. 305, Sept. 29, 1983. 1980), pp. 40-41. 230 ● Information Technology R&D: Critical Trends and Issues

Within MITI’s bureaucratic structure, the cultivation and promotion of data processing Industrial Policy Bureau has played the ma- service industries; and 3) the promotion of jor role in guiding overall industrial develop- computer usage and applications programs de- ment. This Bureau consults with representa- velopment. tives from all industrial sectors and through In addition to these major Bureaus and their these informal and formal meetings sets Jap- respective divisions, MITI policies are influ- anese industrial policy. The major MITI bu- enced by several advisory councils, industry reau involved with information technology is associations, and research associations. Per- the Machinery and Information Industries Bu- haps the most unique aspect of MITI policy- reau. In general, this Bureau oversees and making, these advisory groups are where in- coordinates export, import, production, distri- dustry and government officials develop a bution and consumption of machinery and me- consensus on goals and policies for technologi- chanical apparatus. In addition to information cal development. In these councils and asso- technology, aircraft, automobile, machine ciations members from government, academia, tools, as well as other industries, are within and industry discuss technology trends, mar- the Bureau’s responsibilities. Within the Bu- ket potential, and policy. Problems, ideas, and reau there are several divisions which deal proposals are discussed, and if a general con- directly with information technology. sensus is obtained, it is reflected in govern- The most significant division involved with ment and industrial R&D policies and prac- information technology is the Electronics Pol- tices. In addition to these formal channels, icy Division. Its responsibilities include: 1) there are also a number of informal exchanges planning comprehensive policies for electron- between government and industrial represent- ics equipment industries; 2) the distribution atives. of computers; 3) planning programs on the uti- Within MITI, the Agency for Industrial Sci- lization of computers; 4) conducting surveys ence and Technology (AIST) is explicitly on the utilization of computers; 5) represent- oriented toward research and development of ing the Japanese Government at international technology with industrial applications. In ad- organizations concerning information technol- dition to its responsibilities of planning and ogy matters; and 6) overseeing the Data-Proc- administering policies and programs for re- essing Promotion Council. search and development, AIST operates 16 The Industrial Electronics Division is re- government laboratories, including the Elec- sponsible for exports, imports, production, dis- trotechnical Laboratory (ETL), the major tribution, promotion of consumption, and im- MITI laboratory for information technology provement and adjustment of communica- R&D. In conjunction with ETL, AIST also tions products. These products include com- oversees collaborative research with affiliated puters, laser application devices, radar, elec- laboratories and private companies-particu- tronic measuring instruments, telephone and larly for MITI’s targeted national information telegraph equipment, switchboards, facsimile technology R&D programs. The AIST also ad- equipment, broadcasting equipment, fixed ministers the industrial standards programs. multiplex communication devices, and com- Directly under MITI’s jurisdiction, the Elec- munication wire and cables. trotechnical Laboratory is the largest national Two other divisions, Data-Processing Pro- research organization in Japan specializing in motion and Electrical Machinery and Consum- electronics research. The ETL, with an annual er Electronics, also are directly involved with budget of $40 million, employs approximately information technology. As its name suggests, 730 researchers. ETL’s major areas of research the Data-Processing Promotion Division re- include solid state physics and materials, in- sponsibilities include: 1) the examination and formation processing, energy, standards, and licensing of data processing technicians; 2) the measurements. Similar to DOD facilities in the Ch. 7—Foreign Information Technology Research and Development ● 231

United States, ETL has in addition to Its own internal research, responsibility for advising the government on technology options and monitoring industrial R&D programs. ETL, like other technology in-house research laboratories, does not attempt to compete with industry. As in many U.S. Government research labs, ETL concentrates on identifying research projects and directions (usually high risk) where it could supplement industrial research activities. ETL also oversees the industrial research efforts for MITI coordinated national R&D projects.51 As the Japanese Government began to move Phofo cred(f Embassy of Japan into more basic research activities and new Tsukuba Science City state-of-the-art technologies, many of its policymakers felt that its institutions were ill- National Research and Development Projects -equipped (e.g., too constrained, rigid) for basic, Another effort to stimulate basic long-term pioneering research. Consequently, Tsukuba research activities by coordinating industry Science City, which was begun in 1966, was and government research efforts was begun planned and built by the government for the in 1966 with the initiation of National Re- purpose of centralizing research and educa- search and Development Projects. Perhaps the tional activities. In terms of its concentration most significant aspect that sets Japanese of high level personnel, it in many ways resem- R&D efforts apart from U.S. R&D, these na- bles Silicon Valley in the United States, al- tional projects are directed towards research though in terms of government organization, that is in the Japanese national interest, long- the North Carolina Research Triangle is per- term, high-risk, and precompetitive-research haps a better comparison. Located within that is not directed towards any specific prod- Tsukuba City are 30 of Japan’s 98 national re- uct, but technology that is useful for an en- search institutes. These 30 research institutes tire industrial sector. account for approximately 40 percent of the total research budget and 40 percent of the The initiation of a national R&D project is total number of researchers in Japan. In addi- accomplished through a series of steps. First, tion to these 30 national research institutes, through meetings with government, academic, the Tsukuba Science City accommodates a and industrial representatives, usually within total of 46 research organizations, including various councils and associations, MITI offi- two national universities, six organizations cials derive a consensus on areas for national belonging to government-funded special orga- R&D attention. MITI’s “Vision of MITI Pol- nizations, and eight organizations affiliated icies in the 1980s ” is an example of the con- with other administrative entities. Approx- sensuaI decisions reached among the repre- imately 27 research-oriented private corpora- sentatives. Reflected in these “Visions of the tions have also relocated to Tsukuba Science 1980s” is Japan’s basic national economic phi- City.” losophy which states that Japan should seek — —. — — to ensure its economic survival by becoming “George E. Lindamood, “The Rise of the Japanese Computer a technology-based nation and by making Industry, ” Scientific Bulletin, Department of the Navy, Office maximum use of brain power, which is its of Naval Research Far-East, vol. 7, No. 4, October-December 1982, p. 61, 62. greatest resource to develop innovative tech- ‘*For a detailed discussion of Tsukuba Science City see Justin nology.53 More specifically, the report suggests L. Bloom and Shinsuke Asano, “Tskuba Science City: Japan Tries Planned Innovation, ” Science, vol. 212, June 12, 1981, “’’The Vision of MITI Policies in the 1980s, ” provisional pp. 1239-47 and “Science City in Japan-Tskuba, ” Science and translation, Ministry of International Trade and Industry, Technology in Japan, January-March 1983, pp. 6-11. Tokyo, Japan, Mar. 17, 1980.

38-802 0 - 85 - 16 232 . Information Technology R&D: Critical Trends and Issues that Japan should encourage development ef- MITI Electrotechnical Laboratory. The proj- forts and a switch-over to “forward-engineer- ect was jointly funded at $150 million from ing” in the knowledge-intensive or information government and $200 million from industry technologies. over a 4 year period. The VLSI Research ASSO- ciation and MITI laboratory efforts were By targeting specific information technol- largely generic and provided support for al- ogy areas for national priority, MITI identifies ready existing industry R&D efforts. The net those areas to receive a combination of direct effect of these efforts was the worldwide in- and indirect project R&D support. This sup- troduction of the first 64 K RAM device. The port system for national projects, often termed resultant successes of the Japanese informa- seed money because of its relatively small tion technology industry may signify that ef- amount, initiates basic precompetitive re- forts across public (MITI), quasi-public (NTT), search and leaves to industry detailed product- and private (major corporations) sectors in oriented decisions. Often this seed money is pursuit of a common national technological given to various information technology firms goal is in fact one of the strengths of the Jap- in the form of 50-50 matching grants. anese national R&D projects system.55 Lastly, MITI forms company groups or re- The national information technology R&D search associations to work on a specific na- project which has received the most attention tional project. Sometimes research associa- recently is the Fifth-Generation Computer tions have actually overseen R&D activities Systems Project. Begun in 1979, the project (as in the VLSI project); however, these re- has become an impetus to the initiation of search associations generally coordinate each other major national information technology member’s separate research efforts. Staff of R&D projects in the United Kingdom, France, these research associations usually include em- and Europe. ployees on detail from government and industry, as well as retired industry and govern- In light of the Japanese Government’s goal ment officials. of stimulating basic research efforts, the objective of the fifth-generation computer proj- Between 1966 and 1979, the Japanese Gov- ect is to move Japan to a lead position in in- ernment contributed approximately $400 mil- 54 formation technology areas related to office lion to 16 different national research projects. automation, computer-aided design, computer- A chronological history of Japanese Govern- aided engineering, robotics, and computer- ment support for national information tech- aided instruction. Moreover, the intent is to nology research and development projects is direct information technology development in presented in figure 46. Since the early projects, Japan to specific societal needs. These include: typical amounts committed to national re- coping with an aging society; increasing activ- search projects appear to be increasing and the ity in low productivity areas; increasing ener- scope of the projects is towards more basic re- gy savings; and assisting the transformation search. of society into one in which information plays The VLSI development project exemplifies a key role. The goal of the fifth-generation one of the better known national information project is to develop basic technology and pro- technology R&D projects, largely because of totype systems that can perform functions the subsequent market success of the Japa- such as inference, association, and learning as nese integrated circuit industry. Begun in well as non-numeric processing of speech, text, 1976, the VLSI project involved the formation graphics, and patterns. of a new VLSI research association with seven participating private companies in addition to “For an in-depth analysis of the VLSI projec~ see Kiyanori Nippon Telephone and Telegraph and the %ludcibara, “From Imitation to Innovation: The Very Large —— Scale Integrated (VIM) Semiconductor Project in Japan, ” Al- 64 Leonard Lynn, “Japanese Technology: Successes and Strat- fred P. Sloan School of Management, Massachusetts Institute egies,” Current History, November 1983, p. 370. of Technology, 1982-1983.

Ch. 7—Foreign Information Technology Research and Development ● 233

Figure 46.—Japanese Government Support for Information Technology

1 Sponsors Large Scale MIT I All government funding through consignment payments. 2 These are combined because IS a continuation of (1) Sponsors All government funding through consignment payments 3. FONTAC was aimed at developing a large size computer with IBM systems. Corporate 4 The was established by law 1970 to encourage the development of software by direct and operations are by M(TI Three long term credit banks provide loans to software houses and data services through guarantee fund Total government support unclear, but totaled for FY 1972.1980 (see text for additional material), 5 of (2) Sponsors AIST and Corporate Toshiba, Sanyo. Research and Hoya Glass 6. aimed at developing a new series of computers competitive with IBM’s 370 series a 50 percent subsidy to three computer manufacturer groups Corporate (produced M series), NEC.Toshiba (produced and (produced 7 Sponsor 31 companies, 50/50 develop high efficiency Input-output and terminals 8 Sponsor unclear, 50/50 (government/private). 9. Sponsors Machinery and Information Bureau and Data Processing Division, Corporate large Japanese software companies to an I PA the Systems Development Corp , a number of unspecified smaller firms, to Increase the production and use of software programs This constitutes most software development program to date Results unclear 10 & 11, because 11 IS seen as a of (10), Sponsors Machinery & Information Bureau and Industrial Electronics Corporate two phases: Fujitsu, & Toshiba, OK I, Sharp, Matsushita; above, plus NTT and AlST’s Laboratory staff, formed Phase VLSI Research formed, Phase II Electronic Computer Basic Technology Research Association formed (July 1979). Funding (govern. conditional loan, repayable profits are generated from technologies; Phase 1: Y30 from the government, Y42 from the sector Phase from the government; Y24.5 from the sector 12. Sponsors National Research and Development Program, Corporate Toshiba, Matsushita Electric, Research Association of Optoelectronics Systems (January 1981); Laboratory ed by Optoelectronics Joint Research the Fujitsu Plant, all government funding through consignment payments. Sponsors National Research and Development Program, Corporate Fujitsu, Toshiba, Mitsubishi Government Laboratory Ass/s Laboratory, the for the Development of Speed Computers (December 1981), Laboratory is also Involved, although majority of work be conducted at companies’ own research all government funding through consignment payments. Next Generation Baste Technology Of 48 companies 3 areas; numbers ( ) Indicate number of firms, Area New Materials (33), Area Biotechnology (14), Area Ill: Semiconductor Elements (10), five formed, 3 for Area 1, 1 for Area 11, and 1 for Area Ill; all government funding through consignment payments Sponsor” Machinery and Information Bureau, Corporate Part/c/pal/on Fujitsu, Toshiba, Government Laboratory Ass/stance. Laboratory, at preparatory stages; The Institute for New Computer Technology, an endowed research foundation (April 1982), total funding yet to be determined SOURCE Jimmy W Wheeler, Merit E Janow, Thomas Pepper, and Yamamoto, “Japanese Industrial Development in the 1980’s: Implications for U.S. Trade and Investment, ” Hudson Institute , 1982, for the Department of State.

234 ● Information Technology R&D: Critical Trends and Issues

The Japanese Government established the which each of the participating research insti- Institute for New Generation Computer Tech- tutions and companies conducts its own re- nology (ICOT) in April, 1982 as the center search work. In addition to its own internal organization for coordinating the fifth-gen- research, I COT cooperates with two govern- eration computer project R&D activities. ment laboratories, various Japanese univer- Although the government is funding the in- sities, and independent foreign researchers. itial 3 year R&D stage, eight manufacturers ICOT also contracts with Japanese industries donated money to establish and run ICOT. to make and test prototype software and hard- The consortium of eight manufacturers which ware. The specific universities and companies equally support ICOT and share in the re- involved vary with each individual research search results are: Fujitsu Ltd., Hitachi Ltd., project and participation is not limited to the Matsushita Electrical Industrial Co., Mit- consortium of eight major companies sponsor- subishi Electric Co., NEC Corp., Oki Electric ing ICOT. Figure 47 illustrates ICOT’s orga- Industry Co., Sharp Co., and Toshiba Co. nization for various research projects. These companies, in addition to NTT and The research plans of the ICOT center focus MITI’s Electrotechnical Laboratory, have on seven major areas: sent 42 researchers to the ICOT research center. ● basic application systems, ● basic software systems, Beginning with a staff of 52 and a planned ● distributed function architectures, budget of $450 million over the first 5 years, ● new advanced architectures, the overall research program is scheduled to ● VLSI technology, last for 10 years. In addition to its relatively ● systematization technology, and long-term research, ICOT is unusual because ● development supporting technology. it is a separate neutral organization with a centralized research laboratory. This contrasts Within these seven areas, 26 research projects with the traditional Japanese approach in are to be conducted by teams of university,

Figure 47.—Cooperation Between Research Participants for the Fifth-Generation Computer Systems Project

(lest-model making)

SOURCE: Trudy E. Bell, “Tomorrow’s Computers—The Quest,” IEIEE Spectrum, November 19S3.

Ch. 7—Foreign Information Technology Research and Development ● 235 industry, and government researchers. The 10 within the seven different areas are illustrated year span for these 26 projects is divided into in figure 48.56 three phases. Begun in 1982, the initial phase involves reviewing and evaluating research on Science and Technology Agency (STA) knowledge processing and developing basic The Science and Technology Agency (STA) technology for the second phase. Hardware is responsible for the overall coordination of and software subsystems such as simulators, social needs-oriented science and technology prototypes for language processing, and ex- policy and expenditure in Japan. It is respon- perimental natural language processing sys- sible for the planning, formulating and promo- tems are being constructed for several exper- tion of basic policies pertaining to science and imental systems. The intermediate phase will technology, and for coordination of these pol- attempt to develop subsystems for hardware icies and activities throughout the various and software as well as algorithms and basic architecture. The final stage will attempt to integrate software subsystems, hardware sub- Richard Dolen, “Japan’s Fifth-Generation Computer Project, ” Scientific U.S. Department of the Navy, vol. systems, and applications software in order to 7, No. 3, July-September 1982, pp. 63-97, and “Research and develop the first fifth-generation computer Development Plans for Fifth Generation Computer Systems, ” prototypes. These three phases of research Japanese Embassy, April 1982.

Figure 48.—Concept Diagram Showing How Research and Development Are to Progress in the Fifth Generation Computer Systems Project

study of basic Intermediate system (prototype) ● Basic application system theory and system systern on a trial basis . Basic software system

● Trial production of ❁■ experimental machine ● re - , New advanced arch itectul SatsicGW UJ ● ✤❉▲✤❉❅ Fifth ● Distributed function evaluationn of architecture network $yateql generation architectwe -a *..**.

● VLSI technology ● - ● Systematization ..w ~m and technology knowledge base IN

● Development support systems WanCad afchitao- 1 I Sufq

SOURCE” Fifth Generation Computer Systems Conference, Japanese Ministry of Trade and Industry, Tokyo, Japan, Oct. 19-22, 1961 236 ● Information Technology R&D: Critical Trends and Issues

government ministries. STA has jurisdiction Japan Electronic Computer CO. (JECC) over councils, research institutes and develop- and Japan Robot Leasing Co. (JAROL) ment agencies which mainly concentrate on Assistance is also provided by the Japan technological developments in nuclear energy, Electronic Computer Co. (JECC), which bor- space and ocean development, aviation tech- rows money from the Japan Development nology, and laser technology. Practically none Bank (JDB) and also from private banks. It of STA’S budget directly supports industrial is a jointly owned firm that purchases com- R&D; therefore, information technology R&D puters from participating manufacturers and which is categorized as an industrial area, is leases them to customers. In 1980, the Japan not widely influenced by STA. However, ap- Development Bank provided $263 million to proximately half of the STA budget indirectly the JECC, $218 million in 1981, and approx- supports information technology R&D imately $100 million in 1982. When the JECC through procurement of information and com- was first established, it provided major support munication technology equipment and facil- for the Japanese computer industry; however, ities for its agencies’ activities. as the financial resources of these companies increases, the Japanese computer companies Information Technology have become less dependent on government Promotion Agency (IPA) subsidy and are establishing their own leas- Another government organization aimed at ing operations. developing and disseminating information and The Japanese Government also helped toes- computer systems is the Information Technol- tablish the Japan Robot Lease Co. (JAROL) ogy Promotion Agency (IPA), which was es- which is made up of 24 members of the Japan tablished in 1970 under the Information Tech- Industrial Robot Association and 10 insurance nology Promotion Agency law. Its goal is to companies. JAROL’S objective is the encour- promote the use of computers, encourage the agement of the development and use of robots development and use of programs, and help in small and medium businesses. Like the software firms. It is the only national organization in the field of software promotion in JECC, JAROL buys robots from manufacturers and leases them at low prices to small Japan. businessmen. JAROL also receives most of its Financing for the I PA comes from govern- funds from the JDB and is therefore able to ment subsidies, private corporations, three lease its robots at low prices. Similar to the long-term credit banks (the Industrial Bank JECC, JAROL aims to create a mass market of Japan, the Japanese Development Bank, for robot technology while encouraging pro- and the Long-term Credit Bank of Japan), and duction. from revenues earned by the association itself. One of the more important of IPA’s activities Ministry of Posts and is its credit guarantee programs. Information Telecommunications (MPT): Nippon processing firms and software houses are often Telegraph and Telephone (NTT) in need of funds to develop software programs, The Ministry of Posts and Telecommunica- but have limited property that can be used as tions (MPT) indirectly influences information collateral. The I PA has a system for guaran- technology research and development because teeing such obligations, as long as they are S7 of its administrative guidance over Nippon registered with the IPA. S8 Telegraph and Telephone (NTT). NTT’s 58 There is Some contention over whether NTT should be classified as a nongovernmental or governmental entity. Because NTT receives no direct funding from the Japanese Government some argue that Nil’ is not a government entity. On the other 57An IPA-style credit guarantee system is not uncommon in hand, the U.S. Government has encouraged NTT to open up the United States. However, the American credit guarantee sys- its procurements to foreign suppliers on the grounds that NTT tems tend to be aimed at broad industries, such as housing, is a government entity and therefore is subject to the GATT rather than at narrowly targeted sectors. government code. Ch. 7—Foreign Information Technology Research and Development Ž 237 budget, services, tariffs, and overall policies, product. Because NTT is one of the largest in- as well as appointments of top officials, are formation technology and telecommunications subject to MPT’s review and approval. How- markets in Japan, most electronics firms coop- ever, NTT has not received a government sub- erate with NTT. Competition often develops sidy for over 30 years; in fact, over the last between these firms in efforts to be selected 3 years NTT has returned to the Japanese as partners in new technologies and systems Government on request approximately $2 bil- developments. lion. This contrasts sharply with the idea that NTT divides its procurement procedures the Japanese Government heavily subsidizes practices into three tiers or “tracks” which information technology R&D. have been agreed to in the General Agreement NTT is the domestic public telecommunica- on Trade and Tariffs (GATT) and other bilat- tions monopoly in Japan, although it does not eral trade agreements:59 have any manufacturing capability within the Track 1 (competitive bidding) is applied to organization. NTT, as the owner of virtually products to be procured based on the Gov- all the telephone lines in Japan, remains the ernment Procurement Code agreed to by the most powerful single entity in Japanese tele- General Agreement on Trade and Tariffs communications. NTT accounts for about (GATT). These procurements usually include three-quarters of the Japanese market for tel- off-the-shelf products such as PBXS, data ter- ecommunications and data communications, minals, modems, computers, peripherals, fac- equipment, and services. As a result, NTT simile machines, measurement instruments, with its demands for new equipment and serv- etc. ices has a powerful influence on Japanese in- Track II is applied to equipment that is not formation technology research and develop- available in the marketplace and which re- ment—more so than MITI. quires some research and development. This track generally refers to products for which Typically, new communications products a limited amount of collaboration between destined for NTT use are initiated in one of the supplier and NTT is necessary to tailor its four Electrical Communications Laborator- applications to NTT’s specifications. Track ies (ECL). NTT spends approximately 2 per- II contracts are normally single-company cent of its revenue on R&D (which amounted contracts. to more than $350 million in 1980), mainly at Track III is applied to equipment not avail- ECL labs. This system of labs corresponds to able in the marketplace and which requires Bell Labs although it is approximately one- extensive R&D for NTT use. These are the fifth of the size of its U.S. counterpart. ECL most highly prized contracts and the most tends to do more developmental R&D rather difficult for foreign or small companies to penetrate. NTT seeks a supplier with suffi- than the basic research for which Bell Labs cient research ability to develop a product, has been so widely acclaimed. Much of the re- or to develop one according to an NTT pro- search carried out at NTT’s labs is devoted to totype. achieving the extremely detailed and demand- In addition, there are Tracks 11A and 111A ing specifications that the company requires that allow for new producers to take over ex- when it issues R&D contracts to private firms. pired Track II and III contracts. More often, NTT launches research in col- U.S. and other foreign telecommunications laboration with one or more of the four major equipment manufacturers have suggested that Japanese electronics firms (NEC, Fujitsu, Hit- NTY'S close collaborative R&D activities with achi, and Oki Electric), which actually send several Japanese companies prevent them staff to the NTT labs. The subsidized joint re- from penetrating the Japanese telecommuni- search normally results in NTI's appointment of a preferred supplier from among the re- 5gJack Osborn, outgoing telegram from the U.S. Embassy in searchers when the time comes to purchase the Tokyo, Japm Oct. 21, 1983, 7 pages. 238 • Information Technology R&D: Critica/ Trends and /ssues cations market. With a 20:1 trade deficit in such as data communications services, would telecommunications equipment and a substan- be delegated to spin-off companies operating tial Japanese penetration of certain U.S. mar- at a regional level-somewhat like the pattern ket niches, many U.S. companies would like of the AT&T divestiture in the United States. to balance some of these telecommunications 6o Hisashi Shinto, the new chairman of NTT, trade deficits. Moreover, other motivations believes that transforming the massive bu- for insisting on participating in the Japanese reaucratic Japanese telecommunications mon- market relate to the changing balance in opoly into a partly private company will be American and Japanese research and develop- more profitable, while encouraging greater ment. Just as the Japanese presence in the competition and innovation in the telecommu- American market has a dual purpose-exports nications and information technologies indus- and the transfer of technology-so might tries. It is believed that more Japanese com- American participation in Japanese markets panies will be encouraged to vie both for and R&D activities serve two functions. NTT’s business and for the private telecom- The issue of NTT’s procurement policy can munications market previously controlled by be best viewed within the framework of the NTT and its selected family of suppliers. plans for liberalization of the entire Japanese telecommunications monopoly. Because a Kokusai Denshin Denwa Ltd. (KDD) large amount of NTT'S profits (last year’s prof- The KDD operates Japan’s international tel- it was $1.6 billion) goes to the Japanese Gov- ephone, telegraph, and other related commu- ernment ($600 million) which needs revenue in nications services.61 Divided from NTT in the face of its continuing deficits, NTT can- 1953, it is 90 percent privately owned, with not make a large profit. Another fiscal con- 10 percent of its stock held by NTT. At the straint is the rapidly decreasing rate of growth time of the inauguration of KDD, the inter- in the number of telephone users. More than national telecommunications research group 90 percent of Japanese households now have of the Electrical Communication Laboratories telephones, which means that the traditional of NTT was transferred from NTT and reor- market (revenue) has leveled off. In addition, ganized as KDD’s Research Department. By NTT is in the process of testing and eventually 1969, a development center was created and implementing its ambitious 20-year all-digital the department was renamed Research and Information Network System (INS) program, Development Laboratories. The laboratories, which is expected to cost up to $120 billion located in Meguro, Tokyo, employ more than by 1990. 120 R&D personnel and are composed of 12 To solve these financial problems, Prime special purpose laboratories and three Minister Nakasone is supporting a major four divisions. step reform plan recommended last year by a A new laboratory being built in the Nerima special study group commissioned to study suburb of Tokyo, will be completed in 1985. the Japanese Government. The plan, now This R&D reinforcement will permit further under discussion, would first convert NTT to research and development concentration in an incorporated government-owned entity. such fields as switching for integrated digital NTT would then be free, however, to set its networks, fiber-optic submarine cable trans- own management and personnel policies. mission, satellite digital transmission, optical- Under the proposed plan, many operations, memory disks, system conversion techniques, wideband video, etc. To help in this effort, ‘In 1982 the Japanese exported more than $408 million in KDD expects to add 50 more engineers to its telecommunications equipment to the United States. During R&D staff by 1988. this same period, Japan imported less than $88 million of this same equipment from the United States. Peter J. Harm, “Data 8’Yasuo Makino, “Telecommunications in Japan: Changing Communications in Japan, ” Data Communications, August Policies in a Changing World, ” Z’ekcommum”cati”ons, October 1983, p. 56. 1983, pp. 139-145. Ch. 7—Foreign Information Technology Research and Development • 239

Defense Agency Japan Development Bank (JDB) Because of Japan’s small national market The Japan Development Bank (JDB) is anoth- for defense equipment, Japanese defense reer government financial intermediary used to search and development activities are fairly target industrial development. The Japanese limited. Overall, Japanese procurement of de- Government’s Trust Fund Bureau (which is fense equipment accounts for less than four- the main organization in its Fiscal Investment tenths of 1 percent of total Japanese indus- Loan Program) provides JDB with its main trial production. source of capital, though it can also raise funds Within the Defense Technical Research In- by issuing certain types of bonds. JDB’s prin- stitute there are very small-scale development cipal responsibility has been the extension of projects on electronics equipment (radar, etc.) long term, low interest loans for capital invest- with a budget of ¥ 1.3 billion ($5.6 million) in ment in new industries. In the years immedi- fiscal year 1982, ¥ 0.6 billion ($2.6 million) in ately after its formation, JDB concentrated on fiscal year 1983, and ¥ 3.8 billion ($16.5 mil- loans for the reconstruction of basic manufac- lion) in fiscal year 1984. These projects are turing industries. done in cooperation with private firms. In ad- As a result of the consensus to increase the dition, the Defense Agency indirectly supports support for “knowledge-intensive’ industries, information technology R&D by purchasing the JDB began to target support for what it hardware for testing purposes from private terms “development of technology.” The fund- electronics firms. ing categories in area of development of tech- As a result of economic trends in both the nology are illustrated in table 43. Most of the United States and Japan, as well as a chang- computer funds, as a matter of policy, have ing international security environment, there gone to the JECC, although some software have been growing tensions over trade and de- firms have also received funding. For the other fense issues. In general, some Americans be- funding areas for technology development, lieve that the low level of Japanese military there are two general JDB loan programs. expenditures frees funds for civilian research Both of these loan programs, which resemble and investment while requiring higher taxes MITI’s seed money grants, attempt to stim- and absorbing resources in the United States ulate private investment in specific areas of which in turn provides defense.62 As a result, information technology R&D. The first loan the U.S. Government believes that Japan program, set up under a 1978 law, amounted should increase its military strength as well to ¥ 10 billion ($43.3 million) in 1981. Loans as information technology R&D expenditures from this program must be directed toward for military applications. Moreover, U.S. com- specific project areas designated by cabinet or- panies argue that an inequality exists between der. Should a designated project area be over- United States and Japanese trade: no manu- subscribed (as happened with semiconduc- factured U.S. civilian product (with the single tors), JDB can force larger firms that have bet- exception of airplanes) has captured as much ter access to private financial markets to uti- as 10 percent of the Japanese market, while lize those markets, while JDB loans are approximately 14 percent of Japanese defense preserved for the smaller firms. equipment is purchased by the United States.63 The other technology development loan program was established by the bank itself and

——62 — not designated by specific laws, though it still David Denon, ‘‘Japan and the U.S.—The Security Agenda, ” falls within the broad policy guidelines of the Current History, November 1983, p. 355, BqS~phen J. SOl~Z, “A Search for Balance,” ForeignAffm”rs, government. This part of the JDB budget to- p. 75. taled ¥ 44 billion ($190.5 million) in 1981. 240 ● Information Technology R&D: Critical Trends and Issues

Table 43.—Japan Development Bank Loans for Development of Technology (in billions of yen)

Fiscal year Fiscal year Fiscal year Fiscal year 1977 1978 1979 1980 New loans ...... ¥71.2 ¥129.0 ¥108.5 ¥96.4 $457” Development of electronic computers...... 38.2 55.3 47.1 55.4 262 Domestically-manufactured computers . . . . 35.5 53.5 45.0 54.0 256 Computer manufacturing plants ...... 0.4 0.2 0.4 0.6 3 Data processing systems...... 2.3 1.6 1.7 0.8 3 Use of high technology in certain electronic and machinery industries...... 8.3 7.8 10.2 14.5 69 Electronic industry ...... 3.8 2.1 7.0 12.0 57 Machinery industry ...... 4.5 5.7 3,2 2.5 12 Development of domestic technology ...... 24.7 65.9 51.2 26.5 126 Development of new technology ...... 20.4 57.4 40.9 22.6 107 Trial manufacturing for commercial use . . . 0.9 4.0 1.2 0.3 2 Development of heavy machine...... 3.4 4.5 9.1 3.6 17 aln millions of dollars. SOURCES: U.S. Facts arrd Figures About the Japan Deve/oprnent Bank, Japan Development Bank, 1981, p. 28; and Jimmy W. Wheeler, Merit E. Janow, Thomas Pepper, and Midori Yamamato, “Japanese Industrial Development Policies in the 1980’s: Implications for U.S. Trade and Investment,” Hudson Institute Inc., 1982, for the U.S. Department of State.

These loans are devoted to new domestic tech- ing risks, coordinating private investments, nologies and initial manufacturing efforts for and processing information.”64 commercialization of these new technologies. Firms that believe that they have developed University a process or technology falling within the broad parameters established by the cabinet The Ministry of Education, Culture, and Sci- must apply in order to be considered for loans; ence (MOE) funds research at three research JDB does not solicit customers. The firm’s institutes, six National Institutes (for joint use proposal is submitted to a council of scientific by universities), and 93 national universities. advisors, which evaluates the proposal. If the Among the various national universities in technology is approved, the applicant then Japan, the big seven are the universities of faces an evaluation of credit worthiness and Tokyo, Kyoto, Osaka, Tohoku, Hokkaido, of the financial characteristics of its loan ap- Nagoya, and Kyushu. In the area of informa- plication. If the applicant is a large company tion technology, Tokyo University has activ- with well established financial links, it must ities in several departments: information concurrently seek private financing, because science (in the Faculty of Science), information JDB will provide only partial funding. If the engineering and precision engineering (in the applicant is small, and has relatively weak fi- Faculty of Engineering), as well as a large cen- nancial links, or if the project is large-scale or tral computer center. In most of the other uni- viewed as a high priority for the nation, then versities, there is just one department (usu- the JDB may take a lead role in putting to- ally in the Faculty of Engineering), as well as gether a consortium to finance the project. a sizeable central computer center. Kyoto Uni- Finally, by general agreement, the JDB only versity claims to have the oldest information finances the first plant in a new area. Its role engineering department and is considered a is to help launch new technology, not to pro- close rival to the University of Tokyo. Other vide low cost financing for the expansion of national universities that have significant industry. The small size of the JDB loans indicates 6tEi~uke S&&ibma, Robert Feldman, ~d YUZO H~ada! that the importance of government’s financial “The Japanese Financial System in Comparative Perspective,” a study prepared for the use of the Joint Economic Committee mediary role “stems not from outright control (Washington, DC: U.S. Congress, U.S. Government Printing or from overall size, but rather from socializ- Office, Mar. 12, 1982), p. 11. Ch. 7—Foreign Information Technology Research and Development ● 241 computer science departments include Tsuku- inexperienced persons-recruiting them before ba University and the Tokyo Institute of Tech- they go on to postgraduate work. nology. Among private institutions, only Keio Recently, there has been concern that the University and Waseda University are consid- environment for information technology basic ered to have sizeable computer science research has generally not been adequate at programs. Japanese universities. In addition to the in- Higher education has become an important adequate number of graduate students, two social investment in Japan. Between 1960 and other causes for the small amount of informa- 1975, the numbers of students in higher edu- tion technology basic research activities in cation multiplied by more than three times (to Japanese universities have also been cited. 2.2 million), including students in Japan’s jun- Some researchers believe that the heavy em- ior colleges (post high-school institutions con- phasis on rote learning in Japanese schools has cerned with teacher training, technical educa- helped to suppress creativity in the learning tion, etc.). Of the 1.73 million undergraduates process. Moreover, the importance of severe enrolled at Japanese universities in 1981, ap- university entrance examinations has been proximately 334,000 were enrolled in engineer- seen as a deterrence to specially talented or ing studies (exactly six times as many stu- creative students. dents in the natural sciences, including mathematics). Many universities have no Researchers usually cite two indices of suc- science faculty, only engineering departments. cess in originality and successful basic re- These large numbers of engineering students search-the number of Nobel Prize winners indicate the heavy emphasis placed on engi- and the frequency with which scientists’ work neering in Japan. Consequently, Japan contin- is cited by other researchers. Japan has had ues to maintain a large engineering manpower four Nobel Prize winners. The citation index base. However, a small percentage of these devised by the Institute of Scientific Informa- engineering students continue on to graduate tion in the United States includes 19 Japanese study where they could contribute to basic researchers (and roughly half of those worked university research. For instance, in Japan, in American laboratories) among the 1,000 the proportion of graduate research students international scientists accredited with the to the entire undergraduate student popula- most frequent citations in the scientific lit- tion is 3 percent. In the United Kingdom the erature.66 proportion is over 19 percent; in France 22 per- 65 A second major factor affecting basic recent; and in the United States 12 percent. Al- search in university environments is the level though Japan has more undergraduate stu- of government funding. More than 98.8 per- dents enrolled in electrical and electronics cent of R&D expenditures at national and pub- engineering programs than the United States, lic universities was funded by the government Japan has approximately one-fourth as many and only 1.2 percent by private industry.67 In electrical and electronic engineering graduate 1979, universities spent approximately $3.69 students as the United States. Because stu- milion for R&D: national and public universi- dents at the graduate level make a large con- ties spent $2.459 million, and private univer- tribution to basic R&Din university environ- sities spent $1.15 million. The funds for rements, the low number of Japanese graduate search are distributed by the Ministry of engineering and science students has been 66 The citation system however, does not take into account cited as one reason for such a small amount the fact that very few Western scientists are able to read the of university research in Japan. In this context Japanese scientific literature that is published in Japanese. See it is also worth noting that Japanese compa- March 1984 hearings held by the House Science and Technolo- gy Committee on Japanese science and technology information. nies, which prefer to train their own develop- 67Michiyuki Venohara, Nippon Electric Company, Ltd., “Jap- ment engineers, like to recruit young inexpe- anese Social System for Technological Development-Its Merits and Demerits, ” presented on Dec. 8, 1982 in Endohoven, Neth- “The Economist, Aug. 6, 1983, p. 65. erlands, p. 21. 242 • Information Technology R&D: Critical Trends and Issues

Education in the form of formula support and encourage students to obtain work experience project research grants. in private industry. Furthermore, most Japa- nese firms do not generally look to universities Through formula support the government for innovative ideas; the larger, more impor- provides each Department Chairman with tant firms prefer to carry out their own basic three posts, two assistant professors and one research. Some university professors in Japan research assistant or vice versa. In each engi- have accused Japanese firms of suffering from neering or information science department or a “not invented here syndrome” and this has other related department, the chairman is caused a great mistrust between industry and given approximately ¥ 7 million ($28,000) to university faculty. A further factor is that the spend on research. Half of these funds maybe Ministry of Education, which is the direct em- kept by the university to cover administration ployer of university staff, actively discourages costs. As a result, many departments are left direct links between academics and companies. with approximately ¥ 2 million ($8,000). Professors at state universities are forbidden The research project grants, totaling ¥ to consult for private firms because it could 40,000 million ($160 million), also appear to be lead to nepotism in obtaining appointments. insufficient because of their short term and The most effective channel for university-in- small amount. A limitation is that the funds dustry collaboration in Japan is through in- cannot be used for recruiting short-term assist- formal personal links. In comparison with ance because most researchers (as well as most other countries Japanese graduates remain in Japanese workers) enjoy life-time employ- close contact with each other throughout their ment. The Japanese Government began in professional careers. This leads to valuable co- 1982 to award 3 or 4-year support for research operation between academics and industrial- projects, each costing approximately $1 million. ists, particularly in research. This is reinforced Although funding levels for basic research by the role university professors play (as em- are sometimes perceived as being inadequate, ployees of the Ministry of Education) in the the equipment in university laboratories has establishment and implementation of national been described as adequate by one recent research programs, such as the VLSI project American visitor to the University of Tokyo and the Fifth-Generation Computer Systems labs: Project. The presence of academics in the controlling bodies of these projects helps to ex- In general, both in the industrial as well as change research results between companies in the university laboratories, the impression I got was of a lot of equipment, some old, and universities more effectively. some new, mixed in a somewhat random fash- Other cases of informal collaboration include ion. In the university laboratories, in particu- exchanges of researchers between companies lar, space seems to be at a premium. There and industries. For example, the Electronics does not, however, appear to be any short- G8 Department at the University of Tokyo cur- age of new equipment. rently has five visiting researchers from well Research and Development Links known Japanese electronics companies for periods of 1 or 2 years. Also the department re- Between Universities and Industry ceives grants from at least 10 companies to Direct cooperation between universities and be used for purchasing equipment. The ar- industries in basic and applied R&D has been rangement for visiting researchers from indus- relatively limited. Japanese companies do not try has some similarity to the U.K.’S Teaching Company Scheme, but the emphasis seems to ‘Derek L. Lile, “Japanese Laboratory Visits,” %“entific Bul- be on the industrialist working in the univer- letin, Department of the Navy, Office of Naval Research, Far sities rather than on the academic working in East, January-May 1983, p. 48. industry environments. Ch. 7—Foreign /formation Technology Research and Development . 243

Collaboration between companies and uni- tinued innovation. Although there are many versity departments is also aided by the prac- smaller electronic firms in Japan, most of them tice of using universities as “shop windows” subcontract small-scale production or fill spe- for new equipment that is often given or sold cial niches (some in global markets) which re- to them at very low prices. Through univer- quire custom or batch labor intensive produc- sity use of new equipment, firms receive feed- tion technologies. In general, however, major back on the operation of new equipment and information technology firms are more diver- often valuable suggestions for modifications sified and highly integrated than other com- and extensions. Instances of this practice can petitor nations’ industries. For example, Jap- be found in the areas of computers (Fujitsu) anese firms competing in the semiconductor and telecommunication receivers (NEC). markets are significantly larger in total sales Another recent government incentive for in- and assets than their U.S. counterparts-approximately two to four times larger than dustry-university R&D collaboration concerns Texas Instruments and Motorola, and much patents. As government employees, university larger than National Semiconductor, Fairchild, staff in Japan were not allowed to profit from 69 the commercial exploitation of their ideas. and Intel. This has created a disincentive to patenting, In addition to large-scale operations, many while promoting dissemination of results of the major Japanese information technology through open publications. Currently aca- firms are vertically integrated. For example, demic researchers are being encouraged to most computer manufacturers produce semi- apply for patents and at least two universities conductors and several of them are major have set up special offices to facilitate the pat- semiconductor suppliers in the world market. ent application process (Tokyo Institute of This sharply contrasts with the United States Technology and Tohoku University). This where, although most computer manufactur- could be to make research results more attrac- ers have at least some in-house semiconduc- tive to private firms, which can in turn apply tor development and production capability, for licenses in order to market the technology. only a few firms market both semiconductor As the technological level of Japanese indus- chips and computers, and most of them do not tries approaches that of other advanced na- ‘gGene Adrian Gregory, and Akio Etori, “Japanese TWhnol- tions and the innovation of original technology Today, the Electronic Revolution Continues, ” Scientific ogies is more widely demanded by domestic American supplement, 1983, p. J, 22. as well as foreign markets, the need has been voiced by Japanese industrial and government circles for much closer cooperation between industry and university researchers. As a result, the Japanese Government is also encouraging closer research links between universities, industrial firms, and government institutions through the development of research parks, such as Tsukuba Science City.

Industry To compete in global markets, Japanese information technology industries, in general, are large-scale organizations in order to assure Photo credit: Overseas Public Affairs Office, maximum economies of scale and to sustain Electronic Industries Association of Japan the large amounts of capital necessary for con- Semiconductor research 244 ● Information Technology R&D: Critical Trends and Issues

sell the large range of products that are avail- Because of the highly competitive nature of able from Japanese companies. the electronics firms, industrial researchers Another type of structural integration that were at first extremely skeptical about these is believed to give the Japanese a competitive research associations. However, the area of co- advantage is in the area of consumer products. operation is limited to very basic research, and Compared to many American computer man- development efforts are still necessary to de- ufacturers that specialize in their production velop commercial products. The basic research capabilities, Japanese firms which manufac- efforts generally leave many possible avenues ture computers are also large producers of con- for future product development and allow sumer electronics products. Because consumer firms to compete with one another. Because electronics products usually have a large mar- of their wide acceptance and heavy industry ket and generate large sales revenues, Japa- participation, these joint R&D programs have nese firms can utilize these large profits to been useful to Japanese industry. fund research and development projects in other information technology areas. NEC Corp. As a result of structural integration, namely, NEC Corp. is Japan’s largest manufactur- production and development of computers in er of communications equipment with fiscal conjunction with telecommunications equip- 1982 sales totaling $4.98 billion and a pretax ment, and consumer products coupled with in- profit of $204 million. In 1982 sales to NTT tegrated semiconductor development and pro- and the Japanese Government agencies ac- duction capabilities, the Japanese information counted for $888 million, or 19 percent of technology firms may have a strong techno- NEC’S revenues, of which a significant portion logical base from which to continue their in- was for data communications. Foreign pur- novation process. Moreover, because of their chasers accounted for 30 percent of NEC’S vertical integration, Japanese information sales. NEC has a strong position in the world technology firms can draw on cash flow gen- market (including several plants in the United erated by consumer electronics sales to sus- States), and it markets over 14,000 products tain large capital investments and research in over 100 countries worldwide. NEC has also and development costs of basic research. produced the distributed information-processing network architecture (DIMA) on which 60 Japanese information technology firms also to 70 percent of all networks in Japan are now participate in cooperative joint R&D pro- based. NEC’S success in part can be attributed grams. Because there is relatively low technol- to the fact that it is a vertically integrated sup- ogy transfer between Japanese firms (due to plier with strong shares of the Japanese com- lifetime employment of engineers and fierce puter, communications, and semiconductor competition between firms), they often under- markets. Internationally, NEC is known for take identical research projects. The efforts of having installed over half of all satellite earth limited numbers of research engineers in each stations and for its microwave technology. company are therefore spread out across many NEC’S vertical integration activities are best different areas, making it difficult to concentrate on basic research. In response to these summed up by Michiyuki Uenohara: difficulties, the Japanese Government created Combining computers and communica- research associations so that industries could tions (C + C) was conceived as a form of bus- collaborate on research projects centered on iness best suited for our expansion into new different technology areas.70 areas, making the most of technological resources we have as a company that started out as a communications equipment maker. 70See the section on MITI for a description of the industrial It was in 1975 that the concept was put into participation in the research associations. practical business programs, and in 1977, we Ch. 7—Foreign Information Technology Research and Development ● 245

came up with a clear-cut expression of C + C. The “tsu” in the title Fujitsu means com- First of all, we achieved the ability to put on munications and this was the basis of the Fu- the market a digital switching system which jitsu laboratories. However, because of a per- can serve as a link between communications ceived saturation of the communications field, and computer networks. Digital switching Fujitsu decided to diversify and as a result, has made the combination of computers and communications not only possible, but nat- communications equipment now occupies a ural. 71 minor part of Fujitsu’s research operations. Approximately 60 percent of the research ac- NEC’S central research laboratory, located tivities is currently devoted to computer and just outside Tokyo in the city of Kawasaki, computer-related research. The Fujitsu Lab- houses approximately 700 scientists involved oratories, also located in Kawasaki, employ ap- in research on a wide variety of subjects re- proximately 800 people. The semiconductor lated to computers and communications. Spe division consists of 190 people divided into cific research includes projects on GaAs cir- three main subgroups: GaAs devices and circuit development and semiconductor surface cuits, silicon integration, and materials. Be- treatment. cause its research is well advanced in the semiconductor division, Fujitsu is expected to play Fujitsu Ltd. an important role in the Japanese supercom- Fujitsu is Japan’s largest mainframe ven- puter project. dor, and ranks second to NEC in telecommunications equipment sales. Fujitsu is now Hitachi Ltd. ranked sixth in the world, ahead of Honeywell Hitachi is Japan’s third largest company (U.S.), CII-Honeywell Bull (France), ICL (after Nippon Steel Corp. and Toyota Ltd.). It (U.K.), and Siemens (West Germany). With fis- entered the computer market in the late 1950s cal 1982 sales of about $3.36 billion, telecom- and was one of the first Japanese companies munications equipment accounted for $638 to enter into a technology exchange agreement million of Fujitsu’s total sales. Sales to NTT with a U.S. computer manufacturer (RCA in accounted for 37 percent and sales to the Jap- 1961). Since then, Hitachi has made agree- anese commercial market were 27 percent of ments with Intel. During 1978-80, Hitachi also Fujitsu’s total sales. Similar to NEC, Fujitsu completed marketing arrangements for its is well established in the world information computer systems in Europe with BASF technology market, with a solid history of (West Germany), Olivetti (Italy), and St. Go- sales in the United States. Currently, Fujitsu bain (France). is manufacturing and marketing digital PBXs in a joint venture with American Telecommu- Hitachi generated more than $3.3 billion in nications Corp., and microcomputers and ter- fiscal 1982 (out of total sales of $15 billion) minals in a joint venture with TRW, Inc. from information and communications sys- Other joint venture partners include Canadian, tems and other electronics devices. Although West German, and Spanish companies. Al- Hitachi has in the past maintained a strong though Fujitsu ranks second to NEC in position in the consumer electronics market, microwave equipment, carrier transmission, the company is placing more emphasis on in- and automated office equipment, it is a leader dustrial electronic equipment. As a result, in Japan’s fiber optics and optoelectronics re- Hitachi is attempting to evolve into an in- search. tegrated office systems supplier in the United States. Hitachi is concentrating some of its ef- “Gene A~rian Gregory and Akio Etori, “Japanese Technol- ogy Today, The Electronic Revolution Continues, ” Scientific forts on the development and marketing of Amen”can, Special Supplement, 1984, p. J, 44. PBXs. 246 ● Information Technology R&D: Critical Trends and Issues

underway include the development of GaAs, analog circuits for automobile telephones, and semiconductor material research. There is also research taking place in lasers, light-emitting diodes (LEDs) and light detectors, and some preliminary work in electro-optic integration.

Oki Electric Industry Corp. Oki Electric Industry Corp. was one of Ja- pan’s first entrants into the computer market with its transistorized OKITAC 6020 in 1959. Oki-Univac Company Ltd. was formed in Sep- Photo credit: Overseas Public Affairs Office, Electronlc Industries Association of Japan tember 1963 to manufacture various Univac- based machines in Japan. Since that time, Oki A Hitachi engineer feeds a glass plate into an electron beam lithography device which produces a photomask has not engaged in the development of large by drawing LSI patterns on the plate with a sharply computers. However, it remains strong in focused electron beam under high-precision computer control. The device is also capable of directly writing peripherals and terminals. In 1972, Oki estab- patterns on silicon wafers. lished subsidiaries in the United States to pro duce and market communications equipment and computer peripherals, terminals, and com- Hitachi has rapidly increased research and ponents, and has subsequently set up similar development expenditures. Hitachi’s Central operations in Germany and Brazil. Research Laboratory at Kokubunji is one of the five Hitachi research labs and is devoted Oki Electric Industry Corp. had fiscal 1982 to the development of new materials and de- sales of $1.03 billion, of which telecommunica- vices as well as new measurement equipment, tions sales to NTT totaled $185 million and medical engineering equipment, and commu- those to government entities another $94 mil- nications and information processing systems. lion. Oki also exported approximately $120 The laboratory, established in 1942, employs million in equipment, mainly digital printers, approximately 1,200 research and support per- to the United States. In July, the company an- sonnel. Their principal integrated circuit re- nounced that it will build a plant in Atlanta, search project is aimed at developing a 1K bit Ga., to produce mobile cellular radio tele- static RAM for use in the central processing phones for a subsidiary of AT&T. The facil- unit (CPU) and main memory of large comput- ity is expected to produce 1.5 million units a ers. Other projects that are also currently year eventually.

France-

Introduction reduced to between 7 and 8 percent.72 Infor- The historically agrarian society of France mation technology, particularly telecommuni- has changed with tremendous speed since cations, has been a major component of this World War II. In 1945 over half of the French ‘zPierre Aigrain, “Seminar on High Technology in France, ” population was dependent on agriculture for Center for Strategic and International Studies, Georgetown Uni- its income; by 1970 that dependency had been versity, Feb. 9, 1983. Ch. 7—Foreign Information Technology Research and Development ● 247 change. French efforts to expand advanced information technology research and production have been directed toward two major goals: strengthening France’s international competitiveness and the development of an information technology-based infrastructure for the preservation and continued development of French culture and society. Coupled with historic French governmental involvement in industry,73 the comparatively small size of French participation in the world market for information technology, and the unique French social and political contexts for technology, these goals have shaped French information technology research and development into a pattern quite unlike that in the United States. French information technology research and development activities occur in government laboratories, in industrial settings, and in academic environments. The structure, organization, and direction of R&D activities within these communities are all dissimilar, however, from the American experience. The pervasiveness of government intervention in industrial and academic sectors makes it difficult to differentiate the three areas, but for ease of comparison with the American experience, French government, university, and industry information technology research and development environments are discussed separately below. Before discussing these environments, it is important to understand the size of French par- Photo credit’ Scientific Mission, Embassy of France ticipation in information technology, and the and Ministere des PTT, French Government social and recent political environments for the Videotex terminal “Minitel” conduct of information technology research and development. a United States share of 48 percent in 1982. Moreover, if one were to compute the French The Size of French Participation share of the world market using information in Information Technology Markets technology goods and services produced exclusively by French-controlled corporations, In 1982 the French estimated that they con- the share would decline to 4 percent.74 trolled about 5 percent of the world market in information technology. This compares with Slightly over one-half of the French information market is served by foreign suppliers; 73The French Government has traditionally played a large role the United States holds about 22 percent of in the coordination, funding, and direction of the French economy since Jean Baptiste Colbert founded the Academy of Sci- the market, Japan and West Germany 7 per- ences in 1666. French Governments since have changed the cent each. The Netherlands holds 6 percent scope and nature of that involvement, but the traditional mech- ———. -.— - . anisms used by government in industry, including those of the “French telecommunications and Electronics Council, The present French Government, have changed very little. Electronics Industry: U.S.A./France 1982, pp. 10-18.

38-802 0 - 85 - 17 248 ● Information Technology R&D: Critical Trends and Issues and Italy, the United Kingdom, Austria and all candidates had some increased R&D fund- others hold the remainder. Several major sub- ing planks in their platforms. Before losing to sectors of the French market such as main- Mr. Mitterrand, Mr. d’Estaing had designed frame computers are nearly dominated by U.S. a plan for increasing real government R&D manufacturers. In areas of French strength funding 8 percent per year for 5 years begin- within the French market (e.g., in telephone ning in 1980. When Mr. Mitterrand was terminal and switching equipment, in which elected, he more than doubled that goal. the French control about 91 percent of their Mr. Mitterrand’s emphasis on increasing market), French participation in the United R&D spending was part of a larger industrial States and/or world markets is often very policy for France which included companion small. For example, France has less than one- employment and education policies as well as tenth of 1 percent of the U.S. market in tele- planned market programs in several areas of phone terminal and switching equipment. This 78 high technology. The overall government pol- situation may be more due to the structure of icy, designed around the Socialists’ principles the U.S. telecommunications industry than of decentralization, democratization, human- any French inadequacy. In addition, the future ism, and volunteerism, included several French position in the U.S. telecommunica- elements. tions market may improve as both CIT Alcatel and Thomson CSF have recently established The first element of the French Govern- U.S. subsidiaries.” ment’s general policy was the declaration of information technology development as a na- The size of French participation in the mar- tional priority. Thus a major objective was to kets for information technology affects the lev- bring together universities, government label of funds available for R&D. French indus- oratories, and industry to enhance national ef- trial funding of information technology forts in technological development.79 The sec- research and development was reported as ond element was to convince the French people $2.2 billion in 1982, some significant portion of the importance of industry, a particularly of which was funded by the government. In necessary action in a country that has not yet addition to information technology research completely integrated industrial activity and development in industrial settings, the ci- among its values and culture. The third ele- vilian French governmental funding of infor- ment of the French Government’s policy was mation technology conducted in public labora- 76 to create conditions for people to accept more tories was reported to be $0.6 billion in 1982. readily changes in their work environment and social structure (caused by the introduction of The Political Environment for French new technologies), basically through a renewed Information Technology Research social participation. This was viewed as a ne- and Development cessity for the continuous introduction of new technology. The fourth element consisted of It would be difficult to describe French in- the introduction of mechanisms to increase information technology research and develop- dustrial investment. One of the major mech- ment activities without first considering the anisms for attracting investment in major in- context of the present French Government’s dustries has been the nationalization of major policy. The last French presidential election French industrial firms. (1981) marked the first time science and technology were used as apolitical issues.” Indeed, ‘a’’ French Technology Preparing for the 21st Century,” Sa”entific Ame~”can, November 1982, p. F3. 751nterview with Mr. Chavance, CIT Alcatel director, June ‘gRobert Chabbal, “The New Investment in Science and Tech- 1983 and Telephony, July 25, 1983, p. 24. nology in France, “ in Thomas Langfitt, Sheldon Hackney, Al- 7oA.F.P. Sciences, No. 325, Oct. 7, 1982, p. 30. fred Fishnay, Albert Glowaske (eds,), Partners in the Research 77 Pierre Aigrain, “The French Experience in High Technol- Enter@”se, Um”versity-Corporate Relationsin Su”ence and Tech- ogy, ” Center for Strategic and International Studies, George- nology (Philadelphia University of Pennsylvania Press), 1983, town University, p. 2. p. 138. Ch. 7—Foreign Information Technology Research and Development • 249

The first step taken by the Mitterand gov- pagnie Generale d’Electricity, Saint Gobain ernment was to establish a ministerial depart- Pont-a-Mousson, Pechiney-Ugine-Kuhlman, ment for research and technology. One year Rhone-Poulnec, and Thomson-Brandt. The later, the three ministries of research and tech- government also acquired majority shares in nology, industry, and energy were combined Dassult and Matra, and later negotiated con- to create one very large ministry called the trolling or full ownership of three foreign- Ministere de la Recherche et de l’Industrie. owned companies in France, Roussel-Uclaf, This ministry was created in part to stimulate Cii-Honeywell-Bull, and I.T.T. France. interactions and exchanges between govern- The actual effects of nationalization on the ment and industry. information technology industry and on the Another measure taken by the French Gov- French economy as a whole are still uncertain. ernment was the nationalization of major in- This uncertainty is created in part by the dom- dustrial firms and most of the banking sector. inant role the French Government has played Nationalism was viewed as a means of control- in the French economy throughout all of the ling investment and ensuring that the govern- postwar period, and in part by the confusion ment could exert economic leverage to achieve in the transition to Socialist industrial policy. its goals of: expanding employment; trans- Moreover, the effects of nationalization have forming the workplace environment; enhanc- also been obscured by France’s economic de- ing French productivity and competition; cline in the first 2% years of the Socialist gov- directing research and development into areas ernment, which in turn has ultimately weak- of government priority; and recapturing the ened the nationalization efforts to reshape or domestic market by replacing imports with do- control the activities of specific firms or cer- mystically produced products.8o Moreover, tain sectors of the economy.Nationalization government ownership was seen as means to efforts have also been complicated by changes enable those companies to receive ample fund- in three different ministers of the Ministere ing for innovative, 8relatively high risk researchO de la Recherche et de l’Industrie within a 2 and development. Nationalization was also year period. viewed as a mechanism to increase coopera- Most significant to the general Mitterand tion and technology transfer between indus- strategy for the development of technology try and government. was the enactment of the legislation, Law for The nationalization program was imple- Programming and Orientation for Research, mented in several stages over a period of 2 82 which established scientific and technological years. Following several delays, the nation- research and development as national priori- alization program was approved on February ties. Moreover, the law ensured funding for 11, 1984, giving the government control of 5 long-term scientific efforts by stipulating both industrial groups, 39 banks, and 2 financial quantitative and qualitative objectives for the organizations. 83 In the information technolOgy following 5 years. The law stated that between sector, almost every major company has been 1981 and 1985 the percentage of the Gross Na- reorganized to reflect a majority of govern- tional Product (GNP) devoted to research and ment ownership. For example, the government development will increase from 1.8 to 2.5 per- took over the central organizations (but not cent, thereby increasing by 40 percent the necessarily all the subsidiaries) of the Com- spending85 for technological development in 5 years. This process began in 1980 and there ‘OMichael H. Harrison, “France Under the Socialists, ” Cur- rent History, April 1984, p. 155. *“’Pitfalls in France’s Vast R&D Plan, ” Business Week, Nov. “1bid. 23, 1981, p. 94. “Robert Chabbal, “The New Investment in Science and Tech- 82Companies were actually nationalized on Feb. 11, 1982 (Law nology in France, “ in Thomas Langfitt, Sheldon Hackney, Al- 82-155); however, the reorganization plans were not in effect fred Fishnay, Albert Glowaske (eds.), Partners in the Research until January 1983, Enterprise, Um”versity–brporate Relations in Science and ‘3Michael H. Harrison, “France Under the Socialists, ” Cur- Technology (Philadelphia: University of Pennsylvania Press), rent History, April 1984, p. 155. 1983, p. 140, 250 “ Information Technology R&D: Critical Trends and Issues have been 12 percent annual increases in each ● display technology, of the three successive budgets.86 The French ● ergonomics of computerization, Government’s contribution will be matched by ● computer-assisted instruction, industry, which calls for industry to increase ● multiservice communications, its expenditure for research and development ● widebank communications network, by 40 percent in the next 5 years.87 ● design and production assisted by very large scale integrated circuits, In addition to meeting broadly the scientific ● computer-assisted engineering design and and development needs, these increased ex- production, penditures are also for financing national ● voice-processing module, mobilizing programs that are focused on in- ● electrophotographic module, dustrial and government targeted priorities, ● electronic editing, and such as information technology. Although ● computer-aided translation. there are major programs in biotechnology, new sources of energy, machine tools, etc., the In addition to these projects and programs, information technology program is perhaps the plan sets forth mechanisms for state/in- the most ambitious. La Filiere Electronique, dustrial cooperation and outlines efforts which implemented by the Mitterand government in are aimed at alleviating other constraints on 1983, is designed to coordinate and stimulate information technology research and develop- government, university, and industry informa- ment not related to direct R&D funding. These tion technology research and development ef- include manpower programs to correct a per- forts in order to move France into the forefront ceived shortage of engineers and technicians of advanced information technology R&D and and government-sponsored market promotion production.88 Figure 49 illustrates some of the efforts .89 coordination efforts between industry and Because of economic difficulties, there have government for La Filiere Electronique pro- been some funding problems for the Mitter- gram. The 5-year infusion of ƒ 140 billion (ap- and government. Consequently, the plans for proximately $18.7 billion) for R&D is expected an overall increase of 4.5 percent funding for to accelerate the production of information research activities have suffered significantly. technology products by 3 to 9 percent each For most areas of research, this reduction in year, produce a surplus trade balance in infor- funding means that the 1984 research spend- mation technology products, and create 80,000 ing will remain at the same level as in 1983. new jobs. However, the government has stressed that Fourteen national projects for research and it will maintain its commitment to increase development are outlined in La Filiere Elec- funds for high priority areas, including infor- 90 tronique program: mation technology. Other impediments to La Filiere Electronique program and to French ● large scientific and industrial French information technology research and develop- computer, — . . — • building blocks of mini- and micro-com- robotics, electronics, energy, biotechnology, work environments, puting, and cooperation with developing countries. 89 ● For example, the number of people with Level 1 qualifica- consumer electronics systems, tions in information technology (a French Masters degree, ap- 86This increase occurred after 10 years of decreasing spending proximately equal to an American Ph. D.) is expected to fall for scientific equipment and therefore, the need was extremely short of needs by 70,000 for the period 1981-1990 in France. acute. In the French context, this number is quite large; in 1979 it sTRobert Chabbd, “The New Investment in Science ~d T~h- was estimated that 105,000 scientists and engineers were ac- nology in France, ” in Thomas Langfitt, Sheldon Hackney, tively involved in all aspects of French science (energy, phar- Alfred Fishnay, Albert Glowaske (eds.), Partners in the Re- maceuticals, mechanics, aeronautics agriculture etc., as well search Enterprise, Um”versity and Corporate Relations in as information technology). Jean-Pierre Letouzey, Scientific Science and !l’echnology (Philadelphia: University of Pennsyl- Mission, Embassy of Fran~ Sta&ment for the American ASSO vania Press, 1983) p. 140. ciation for the Advancement of Sciences, Mar. 24, 1983, p. 9 “A “filiere” in France is a targeted industry grouping or other (unnumbered). goal around which a government plan for funding, production, ‘David Dickson, “Hard Times Force France to Cut Back Am- investment, education and dissemination assistance has been bitious Plans to Support Science, ” Chrom”cle of Higher Educa- developed, There are currently six filieres in France today: tion, Apr. 21, 1984, p. 10.

Ch. 7—Foreign Information Technology Research and Development ● 251 252 ● Information Technology R&D: Critical Trends and Issues ment may also stem from the proliferation of technology research and development. It can new projects which may in turn dilute avail- only be contrasted to the fact that one of the able funds to inconsequential levels. strong aspects of U.S. information technology research and development has been considered Social Environment for French to be the infusion of small, innovative firms Information Technology Research into the research community. and Development The second French characteristic which ap- The French Government’s desire to push pears to affect the conduct of information tech- France into a technologically based future has nology research and development is the fre- encouraged information technology research quency of what might be termed reengineer- efforts in office automation, microcomputers, ing. Reengineering, or the production from consumer electronics, and telephone terminal scratch, of French versions of an existing prod- equipment. In addition, each of these research uct is very common in French information areas has included a major effort in the human technology. Reengineering efforts maybe at- factors aspects of design and interaction, per- tributed to the French attempt to develop do- haps in recognition of the difficulty expected mestic production capabilities in order to mit- with the assimilation of these technologies into igate United States and Japanese dominance French society. The emphasis placed upon hu- in some niches of the French information tech- man factors engineering has given rise to nology market. French prototypes and products that are gen- The French attention to reengineering is erally esthetically pleasing and easy to use. quite possibly related to a third French char- Many French designs have been adapted else acteristic, strong national loyalty to French- where. For example, it was a French study made products. This adherence to French that suggested amber on black CRT screens products occasionally provides a serious hand- were the most pleasing and produced the least icap to French information technology re- eye strain. search and development. Because the French Three French characteristics stand out as do not manufacture all types of state-of-the- important with respect to information technol- art instrumentation, French scientists and ogy research and development. Risk taking in engineers (who in some cases maybe restricted the French industrial sector does not appear to purchasing French instrumentation) may with the frequency or at the level considered be limited in their research activities. commonplace in the United States. One indication of this is the small number of venture capitalists in France and the unwillingness of Government the traditional banking industry to fund en- There are a variety of French Government trepreneurs. However, there are efforts on the organizations involved in information technol- part of government and industry to increase ogy research and development. A few were the use of venture capital. newly created with the advent of the Mitter- For the conduct of information technology and government, but most have been oper- research and development, the French risk- ating for decades. Although the nationaliza- avoidance characteristic may be translated tion of industries and the government into the general lack of leading-edge, often provision of research and development under high-risk, technological research. Another pos- Mitterrand are often thought of as socialist sible consequence for information technology government actions, the link between French research and development is the prevalence of politics and French research has been long- large organizational environments for the con- standing. Traditionally, France has closely duct of R&D. It is difficult to judge what im- overseen both basic and applied science re- plications this has for French information search through government mechanisms. Ch. 7—Foreign Information Technology Research and Development ● 253

Centre National de la industry have also grown. Closer ties between Recherche Scientifique (CNRS) CNRS and industry seem to be important to the Directeur General of CNRS who has re- The largest and oldest French Government- cently established several agreements between funded research organization is the Centre Na- tional de La Recherche Scientifique (CNRS), CNRS and industry .93 founded in 1939. CNRS supports basic re- CNRS has exchange programs and scientific search in chemistry, physics, earth, atmos- accords with 30 countries, including agree- pheric and ocean sciences, life sciences, engi- ments with the National Science Foundation, neering, social sciences, mathematics, and the National Institute of Health, Massachu- humanities. In 1984, CNRS had a budget of setts Institute of Technology, the University ƒ 7,735 million (almost 24 percent of the pub- of Chicago, and others. lic civil research budget), and employed approximately 25,000 people in 1,350 labora- Ministere des Postes, Telecommunications, tories or within universities, other government et Telediffusion (PTT) agencies, and industry. CNRS has seven labor- The French Ministere des Postes, Telecom- atories devoted to basic research in some munications, et Telediffusion (PTT), through aspect of information technology. The portion the Ministere de la Recherche et de l’Industrie, of the 1984 budget applicable to information is responsible for the provision of all telecom- technology research and development is munications services and equipment, network ƒ 225.2 million, an increase of 13 percent over maintenance, standards development, tele- the 1983 budget. This increase in the budget communications policy, technical assistance for the information technology research (in to former French colonies and other foreign en- contrast to other CNRS research areas which tities, and research and development. The received less or no increases) is significant and PTT’s jurisdiction over telecommunications is largely the result of the recognition of the policy has resulted in increased PTT involve- importance of information technology to the 91 ment in information technology R&D. The French economy and society. PTT often joins the Ministere de la Recherche In line with France’s new efforts in strength- et de l’lndustrie and others in the funding of ening its information technology industry, projects that cross the technological bound- CNRS information technology activities are aries between telecommunications and com- currently coordinated with the Filiere Elec- puters.94 tronique program and are becoming largely de The reach of the PTT through the Ministere voted to applied, industrial research. There are de la Rechereche et de l’Industrie into the in- efforts in software development and program- formation technology research and develop- ming techniques, speech recognition and syn- ment communities is extensive. The PTT thesis, artificial intelligence, and robotics. funds the Centre National d’Etudes Telecom- CNRS draws heavily from the French univer- munications (CNET) much in the same man- sity system for its personnel, unlike industry ner that AT&T funds Bell Labs. The PTT also or most other government agencies, where the funds the Institut National des Telecommu- grandes ecoles supply the researchers and ad- 92 nications (INT), a portion of the Agence de ministrators. l’Informatique (ADI), and all of the Ecole Na- In the past, CNRS ties with industry have tionale Superieure des Telecommunications been relatively weak. However, as CNRS in- ‘sFor example, agreements have been signed with Saint- creases its applied research activities, ties with Gobain, Renault, and Roussel-Uclaf. “French Technology Pre- —— paring for the 21st Century,” Scientific American, November “Discussion of CNRS based on “French Technology Prepar- 1982, pp. F4-F1l. ing for the 21st Century, ” Scientific American, November 1982, “For example, le Project Pilote NADIR (exploration of new pp. F4, F11. uses of Telecom 1 for data and voice transmission) is funded gZ1ntimiew with Ckles Ga.rriques, President, Agence de l’ln- 50/50 by the ITT and the Ministere de la Recherche et de l’ln- formatique, June 24, 1983. dustrie. 254 ● Information Technology R&D: Critical Trends and Issues

States has been a success.95 In 1974, France averaged 12 main telephone lines per 100 inhabitants; in 1981 the figure was 33.96 The number of lines in the national network grew from 7 million in 1975 to 20 million in 1982. In 1975 electronic exchanges were virtually nonexistent in France; in 1981, 70 percent of newly installed switching capacity was electronic. 97 As a result of the recent modernization of the French telephone network, the French have a greater percentage of digital equipment than any other country.ga This, in turn, has spawned the provision of many sophisticated services such as Transpac (public data packet switching network), Transmic (dedicated data transmission), Teletel, and Videophone (video conferencing). These technological possibilities have pushed PTT-sponsored telecommunications and computer research and development in several directions. The three main directions have been satellite technology (the French launched Telecom 1 in the summer of 1983), an integrated services digital network (ISDN), and optical fibers (a wideband, multiservice optical subscriber network experiment is taking place in 1,500 homes in Biarritz). The major Photo credit: Scientific Mission, Embassy of France and Ministere des PTT, French Government research efforts in all of these areas have taken Videotex terminal “Minitel” with CCP card reader. place at the Centre National d’Etudes des Tel- ecommunications (CNET).

(ENST). The PTT’s terminal equipment, switch- Centre National d’Etudes ing and transmission needs are supplied Telecommunications (CNET) through contracts with CIT Alcatel, Thomson The CNET was formed in 1944 to provide CSF, the French Cable Co., and a host of scientific research and technical assistance to others. The R&D for the products manufac- the PTT.99 The CNET is active in applied tured by these firms for PTT use is funded by mathematics, computer science, solid-state the PTT, generally through cooperative efforts physics, and earth sciences. The total budget between the industrial entity and the CNET. in 1982 was about ƒ 1 billion (about $133 mil- The current government-sponsored push in lion); approximately 55 percent is spent on net- information technology research and development follows upon a recent similar effort in ““No Hang Ups for French Phones, ” (ZWecom France, June telecommunications implemented by the PTT. 1982, p. 10. ‘Conseil Econornique et Social, La ‘1’elernatique et L ‘Amen- By most accounts, the research, development, agement du Z’erritoim, Apr. 21, 1983, p. 35. and implementation programs designed in 97P’IT Telecommunications, Bim”tz, The b“ghtwave Commu- 1974 and 1975 to transform the French tele- nications World of the Future, p. 1. ‘81bid. communications system into a technologically ‘gDiscussion based on the group of brochures and pamphlets advanced network on par with the United included in the CNET Dosw”er Presse. Ch. 7—Foreign Information Technology Research and Development • 255 work and service engineering, 25 percent on cember 1979 under the d’Estaing Ministry of components engineering and 20 percent on ba- Industry. INRIA remains under the jurisdic- sic research. tion of the Ministere de la Reserche et de l’lndustrie in the Mitterrand government. The In 1982 2,600 scientists and engineers name of this ministry has had several evolu- worked in six laboratories. Two of the groups tions of late. With the advent of the Mitter- are located in Paris. Paris A performs long- and government the Ministere de l’Industrie term network planning and administers the was changed to the Ministere de la Recherche other five centers. Paris B is the center of basic et de l’Industrie. Currently, the organization research for the CNET. Materials and geophy- is titled the Ministere de l’Industrie et de la sical research are the primary efforts. Applied Recherche. The first change effected by the work in components, transmission, marine Mitterrand government was an effort to com- cables, and satellites is also performed. bine the mission of the Ministry of Industry The two centers at Lannion work on local with that of the Ministry of Research and area networks, ISDN, software, human-ma- Technology. The second change appears to be chine interface, and acoustics (Lannion A) and one of emphasis. INRIA is considered the lead- digital transmission, optical communications, ing research institute in computer science in and components (Lannion B). The center at France. It has several locations. The main re- Grenoble is devoted to microelectronics research center is in Rocquencourt (just outside search, while the laboratory at Rennes is of Paris); another smaller center is located in shared with the Centre Commun d’Etudes de Sophia-Antipolis. INRIA shares facilities with Television et Telecommunications and studies the CNET at Rennes and Grenoble and has a future telecommunications services and their small group in Toulouse. In 1982, INRIA’s integration with broadcasting technologies. budget was ƒ 146 million (about $19.5 million) which funded 409 people, 225 of whom were Since the advent of the Mitterrand govern- scientists and engineers. In 1983, the budget ment, the CNET has assumed a significant was expected to be ƒ 200 million for funding role in the French national industrial develop- of INRIA contracts with industry.l00 ment strategy. Under the plan for the electronics sector (La Filiere Electronique), the CNET INRIA has a three part mission: the conduct laboratory at Grenoble has undertaken proj- of research on experimental computer sys- ects in CMOS technology for very large scale tems; international scientific relations; and the integrated circuits, gallium arsenide, comput- transfer of technology. Each of the missions er-aided design, and artificial intelligence. is guided by industrial needs, at least insofar as those needs are articulated by the Ministere In conjunction with La Filiere Electronique de la Recherche et de l’Industrie. Consequent- research effort, CNET has introduced a pro- ly, the research performed by INRIA is ap- gram to improve its transfer of technology to plied and the bulk of the work can be char- the industrial sector. The CNET owns over acterized as product development. 550 patents which it licenses to French and other companies. The licensing is done virtu- INRIA has eight research programs in areas ally without regard to the royalty potential; such as system architecture, languages, algo- CNET’S 550 patents provide approximately rithms, automation, and man-machine inter- ƒ 20 million in revenue per year. face. Each program conducts three to five projects. In addition INRIA is responsible for L’Institut National de Recherche four of six pilot projects to be undertaken in en Informatique et Automatique (INRIA) connection with La Filiere Electronique program. Those pilot projects are KAYAK (office L’Institut National de Recherche en Infor- matique et Automatique (INRIA) is one of the newest French Government information technology research agency. It was formed in De- IOOINRIA, Dossier Presse. .

256 “ information Technology R&D: Critical Trends and Issues

automation), NADIR (applications of Telecom nominally the job of the Ministere de la Re- 1 satellite capability), SIRIUS (distributed cherche et de l’Industrie, but most often, such systems), and SOL (portable software). coordination is effected by INRIA research Each pilot project has a different configura- staff members who take ideas forward to the tion of funding, personnel, and industry par- Ministry for funding. ticipation. For example, NADIR is financed 50/50 by the PTT and the Ministere de la Re- Agence de l’Informatique (ADI) cherche et de l’Industrie. The project admin- Like INRIA, the Agence de l’Informatique istrative responsibility is shared between the (ADI) was recently formed. Funded by the Agence de l’Informatique (ADI) and the Direc- Ministere de la Recherche et de l’Industrie (75 tion des Affaires Industrielles et Interna- percent) and the PTT (25 percent), ADI’s 1981 tionales, a division of the Direction Generale budget was F 300 million ($40 million). Fund- des Telecommunications within the PTT. The ing in 1982 was F 320 million as was the 1983 actual research is conducted at INRIA with budget. A 20 percent cut is expected for 1984. a mixture of INRIA and CNET personnel. In- ADI was originally designed to be self sustain- dustry personnel are not research team mem- ing though royalties from its research-derived bers, although several industrial representa- patents and the sale of its published studies, tives are participating through provision of but as yet, those items account for less than user specifications.l0l 1 percent of ADI’s revenue. ADI has three ma- SIRIUS is an older project, and as such, its jor areas of activity: research and experimen- original funding sources and location have tation, application development and dissemi- changed. Currently, the project is adminis- nation, and training and education. It also has tered by ADI. The research team includes three support activities: regional programs, in- INRIA, the University of Nancy, several ternational affairs, and economic and legal studies. Sixty professionals manage these pro- smaller divisions of the Ministere de la Re- 103 cherche et de l’Industrie and 15 industrial grams. companies including CAP-Sogeti, Cii Honey- The goal of ADI’s training and education well Bull, and SNCF (the French national program is to produce more computer science railroad).1°2 graduates and to improve the quality and The organization of INRIA is quite different availability of science and engineering educa- from the CNET and in many aspects provides tion. Included in the program is a project to complementary research support. Unlike the put computers in the secondary schools. The CNET which draws heavily from the grandes application development and dissemination ecoles system for its researchers, INRIA re- programs provide a forum for users and pro- cruits from the French university system. ducers of information products to exchange Many project directors at INRIA are French ideas and develop computer applications tai- university professors who come to INRIA for lored to the specific requirements of various the duration of a project. business sectors. Also unlike the CNET, whose mission is to ADI’s research program funds efforts in be the research arm of the state telephone net- computer science in a manner analogous to work, INRIA is not responsible to one central- that of the National Science Foundation; that ly defined set of research requirements. Rath- is, ADI funds but does not conduct the reer, the institution generally must respond to search. This program has approximately a more diffuse set of requirements from indus- F 100 million ($13 million). In addition to the try. The coordination of research projects is four pilot projects ADI helps fund at INRIA,

l’JIThe pilot ~Oj~t NADIR English Language bulletin, May ‘OgAgence de l’Informatique, English language brochure, and 1983. interview with Charles Garriques, President, ADI, June 24, IOZINRIA, ~ggier Pregse Le Project Pilote SIRIUS. 1983. Ch. 7—Foreign Information Technology Research and Development ● 257 partial funding is provided for the RHIN (sys- ity to gather groups representing divergent tem interconnection) and SURF (functional se- interests. curity) projects. These six pilot projects were originally to be undertaken as only a portion of ADI research mission. Longer-term proj- Centre Mondial Informatique ects in new architectures, languages and pro- et Ressource Humaine gramming, human-machine interface, design The Centre Mondial Informatique et Res- aids, automation, computer-aided translation, source Humaine is the other information tech- security, and translation are also planned. Re- nology agency created by the Mitterand gov- cent budget cuts, however, have limited these ernment. The original mandate of the Center research endeavors. was threefold: research in those areas of information technology applicable to microcomput- Centre d’Etudes des Systemes ers; social experimentation in France; and et des Technologies Advancees (CESTA) Third World pilot projects designed to explore computer applications for the dissemination The Centre d’Etudes des Systemes et des 1 5 Technologies Advancees (CESTA) is the new- of medical and education information. ° How- est French agency involved in information ever, the Center’s role in information technol- technology. Founded in January 1982, it is one ogy research and development has not always of two completely new agencies formed by the been clear. Since its inception in November, Mitterrand government in information tech- 1981, the Center has been racked with politi- nology. Under the jurisdiction of the Minis- cal struggles. Exactly what role the Center tere de la Recherche et de l’Industrie, plays or may play in French information technology research and development, however, CESTA’S 40 employees have two missions: l06 technology forecasting and identification of still remains uncertain. employment impacts due to technological advance. The technological scope of CESTA goes University beyond information technology (e.g., there are programs in biotechnology) but the majority Two parallel systems of higher education ex- of its work involves various aspects of infor- ist in France; one is found in the universities, mation technology. CESTA’S forecasting re- the other in the grandes ecoles. Both systems sponsibility takes the form of evaluations of produce scientists, engineers and administra- market, cultural, and social acceptance of tech- tors with relatively little training in the ap- nologically advanced products. When employ- plied sciences and/or the business aspects of ment impacts can be identified, retraining pro- research such as marketing, management fi- grams are developed by CESTA personnel. nance, or accounting. Beyond this similarity, CESTA’S major activities undertaken to fur- the systems have few parallels. ther these goals are the conduct of seminars and the commission of studies and papers on In general, French university training does topics of interest.l04 not usually prepare individuals for careers in government or in the higher levels of indus- Although CESTA resides under the aus- try. University trained scientists are occasion- pices of the Ministere de la Recherche et de ally found in industry and in government re- l’Industrie, the director described his agency search laboratories in instances where the as independent, not administration-linked and drew an analogy between CESTA and the Ten- nessee Valley Authority. He saw this inde- Iobcentw Mond,i~ Informatique et ~ssource Hllmtie, ~’ches pendence as a necessary ingredient to his abil- D’Information, Statuts et orgm”sation. 1%3ee for example, Dray and Menosky, “Computers and a —.— New World Order, ” Technology Review, May/June 1983, and I04CESTA, CESTA, December 1982, and interview with Yves Walsh, “Computer Expert Signs Off From World Center, ” Sci- Stourdze, Directeur CESTA, June 20, 1983. ence, vol. 218, Dec. 3, 1982. 258 • Information Technology R&D: Critical Trends and Issues work is decidedly theoretical or, as in com- iculum at Ecole Polytechnique is based on a puter science, the discipline is so new that a theoretical education in a variety of scientific grande ecole exclusively for the subject is non- disciplines. In addition, students who wish to existent. Although computer science is taught specialize in an area often enter one of the in several of the grandes ecoles, computer sci- other grandes ecoles for additional training entists, particularly software engineers, in after graduation. France have emerged from the university community. As a result, several French univer- The Ecole Nationale Superieure des Tele- sities such as the university of Grenoble, have communications (ENST) is the primary grande recently become sites for information technol- ecole for the production of researchers and ogy research and development activities engineers in information technology. The around which industry has begun to collect. ENST is funded by the Ministere des Postes, This may eventually change the nature of Telecommunications, et Telediffusion (PTT), French research and development which tra- and its curriculum is overseen by the PTT. The ditionally has been institutionally split into school is considered to rank among the top ten three components. In general, the universities of the 150 grandes ecoles in France. The school (and some government organizations) have accepts 70 first year students annually, based conducted France’s basic research, the govern- on performance on the competitive exam. ment and the grandes ecoles have been the More students are added during the second sites of applied research, and industry has and third years of the school’s program. En- been the environment for development activ- try into the second year at ENST can be ob- ities. Although these various institutions have tained after completion of another grandes certainly funded activities in each of the other ecoles education and/or after graduation with components, cooperation among them has a maitrise (roughly equivalent to a U.S. bach- been minimal. elor of science degree) from the university. These additions result in a total student pop- The highly respected grandes ecoles system, ulation of about 540 and the award of 220 de- which produces the French cadre of govern- grees annually. ’07 ment officials and industrial managers, was formed by Napolean to develop an elite group The first and second years of study at ENST of French intellectuals who would be respon- involve mathematics, physics, electronics, sible for guiding France’s cultural, social, eco- computer science, economics, foreign lan- nomic, and political futures. Today there are guage, and humanities. During the third year, a student chooses from eight areas of specialty about 150 grandes ecoles in France of which to study for half of the year. Three months are approximately 10 produce state engineers. The others produce administrators, economists, spent in a work-study project in industry or sociologists, artists, and a host of other pro- government or in a foreign study program. These work-study projects are a unique oppor- fessionals in the sciences, liberal arts, and humanities. All grandes ecoles have entry re- tunity for students to gain applied or indus- quirements. One must take a competitive trial experience. As a result of ENST’S empha- exam which requires between 2 and 3 years sis on applied engineering, the proportion of of study preparation. Typically, the French graduates who work in industry is quite high in comparison to other grandes ecoles. Ap- high school graduate will prepare for this exam proximately 60 percent of the graduates of at the university, within the high school set- ENST go to industry, and the remainder find ting, or in private preparatory institutions. Based on their test scores, students are gen- jobs in the PTT. erally assigned to specific grandes ecoles. The ENST has four laboratories where in- The Ecole Polytechnique has traditionally structional research is conducted: systems and produced the highest level of government offi- communications, computer science, electronics cials in France. As the name suggests, the cur- ‘O’ Ministere des PTT, EA?ST. Ch. 7—Foreign Information Technology Research and Development . 259 and physics, image and sound, and life sci- groups: Bull Systems manufactures main- ences. Occasionally research is conducted in frames; Bull Sems (purchased from Thomson collaboration with government research cen- CSF) manufactures minicomputers; Bull Per- ters and industry; however, the main purpose ipheriques makes disks and printers; Bull of the research is to provide student instruc- Transac (purchased from Alcatel) produces mi- tion, not to further the state-of-the-art.108 crocomputers and office automation products. The collection of groups is now called Bull. The The highly theoretical nature of the transfer of the company to state ownership grandes ecoles training does not always pre- has not only changed the structure of divi- pare government officials to direct and con- sions, and their personnel, it has changed the duct research for purposes of French industrial relationships between management and work- growth. Experience in the industrial commu- er to reflect socialist principles. The work week nity does not appear to be an alternative meth- has been shortened, salary differentials be- od of developing such preparation for govern- tween men and women have been eliminated, ment officials, as the transfer of people from and the number of upper management person- careers in industry to careers in government nel and their salaries have been reduced.log (or vise versa) is quite rare. However, in conjunction with its goal of strengthening the Research at Bull has been reorganized into links between industrial and government re- six working groups, each with 30 scientists search and development, the Mitterand gov- and engineers: advanced systems research; in- ernment has implemented several new pro- tegrated circuits research and technology; cus- grams. These include increasing the number tom design of integrated circuits; standard in- of commercially oriented grandes ecoles, the tegrated circuits; interfaces for technology teaching of marketing, accounting, and fi- use; and discrete components and subsystems. nance throughout the grandes ecoles system, The integrated circuits research and technol- and encouraging more work-study programs. ogy group is mandated to “be a proponent of technology alternatives to our partners” and Industry to [provide] updated competence for the best choices in integrated circuit technology at the The industrial component of French infor- Bull group level. ” Similarly, the custom in- mation technology research and development, tegrated circuits group is to “establish know- just as in the United States, has many mem- how in design of custom VLSI” and “provide bers. Similar to the situation in both Japan support and expertise in our choices of new and the United Kingdom, a few large informa- technologies available outside the company. ” tion technology firms are responsible for the In addition to the six research groups, addi- major proportion of research and development tional studies to determine the feasibility of efforts. Cii Honeywell Bull, Thomson CSF, research in languages, artificial intelligence, CIT Alcatel, and Sogitec, representatives of vector processing, and distributed architec- the spectrum of French industrial experience ture are underway.ll0 in information technology research and development, are described below. Bull’s future position in information technology research and development is unclear. Cii Honeywell Bull It is important to note, however, that should it not improve, the French efforts in informa- When nationalization became operational at tion technology research and development, Cii Honeywell Bull in January 1983, the firm and indeed in other advanced technology was completely reorganized. It took on several divisions of Thomson and Alcatel and re- arranged its internal divisions into four “Wi.i Honey-well Bull, Bilan Social d’Entrepn”se, Exercise 1982 and “Avis du Comite Central d’Entreprise sur le Bilan Social, —. ——— 1982, ” ‘081bid. ““Bull, Corporate Z%chnology, June 17, 1983. 260 ● Information Technology R&D: Critical Trends and Issues

areas, may be diminished. Much of France’s level languages. The basic research effort, technological future is dependent on the avail- while small, is extremely important to Thom- ability of a wide range of sophisticated com- son CSF. The basic research effort provides puters and computer peripherals. As long as Thomson with entree into the basic research the French rely on Bull for these products, a community in France and abroad (Massachu- very important link in the information tech- setts Institute of Technology, Stanford Uni- nology research and development chain may versity, CNRS and several French universities remain uncertain. were mentioned as important research collab- orators). Moreover, a credible basic research Thomson CSF effort helps with the recruitment of scientists and engineers. Because of the excellent reputa- Thomson CSF produces a variety of aero- tion of Thomson CSF’S basic research function nautic electronics equipment, telecommunica- and the small portion of corporate funds it rep- tions equipment for the public telephone net- resents, nationalization is not expected to work, medical devices, electronic components, change the nature of the activities and may and office automation products. Thomson CSF even increase funding. is partially controlled (40 percent) by Thom- son-Brandt, a producer of durable consumer Like Bull, Thomson-Brandt, was national- goods, electromechanical capital goods, lamps ized with the passage of legislation in Febru- and lighting fixtures, and engineering and fi- ary 1982. The effect on R&D activities at nancial services.111 Thomson CSF produces, Thomson CSF has been minimal in compari- sells, and distributes its products through a son with the changes at Bull. Product-oriented network of almost 60 domestic and foreign research has been modified slightly to meet subsidiaries and holding and associated com- some specifications of La Filiere Electronique panies. 112 Its revenues are dominated by the program and state management has caused sale of electronic equipment followed by tele- some difficulties, but personnel and directional 1 3 communications and medical devices. 1 changes for the company have caused almost insignificant disruption. In 1981, Thomson CSF spent over F 4 billion ($530 million), approximately 10 percent of Thomson CSF and Thomson-Brandt com- CIT Alcatel bined revenues, on research and development. CIT Alcatel is a subsidiary of the Compag- Seventy-five percent of the effort was inter- nie Generale d’Electricity (CGE), the fifth nally financed. Thomson CSF performs R&D largest company in France. CIT Alcatel for both Thomson CSF and Thomson-Brandt, (through its 8 French and foreign subsidiaries and its spending represents 25 percent of all and affiliates) represents the public telecom- R&D spending in France. The vast majority munications division of CGE. Another 11 (95 percent) of the research is product-related Alcatel Group members produce office auto- and takes place throughout the company’s mation and professional electronics products subsidiary structures. Basic research takes and provide computer services. Nationaliza- place in the Laboratoire Central de Re- tion has not changed the personnel or the con- cherches. 114 duct of research at CIT Alcatel. It is antici- Basic research at Thomson CSF includes pated that state ownership will improve the relationships between academia and industrial programs in gallium arsenide, molecular beam research and development, although results epitaxy, single mode fiber optics, and machine s are not expected for another 10 years.ll CIT Alcatel’s research and development ac- I I lThom90n.Brmdt, l%nc~pales ~fi”df?s et Participations Francm”ses et Etrangers. tivities are scattered throughout the com- “’Ibid. pany’s subsidiaries. In addition, some research “gThomson-CSF, Rapport Annual des Activities en 1981, ‘]41bid. I IbAlca~l, The Alcatel Group. Ch. 7—Foreign Information Technology Research and Development ● 261 activities take place in association with CGE, McDonnell Douglas, Ford, Daussalt, and CNET, and CNRS, and in conjunction with others in the aircraft, shipbuilding, and auto- other French (e.g., Thomson CSF) and U.S. mobile industries.118 companies (e.g., SEMI Processes, Inc.). The Sogitec has two divisions that are designed company’s director estimated that between 10 to meet individual user needs. The data proc- and 12 percent of the employees (some 17,065 116 essing services division provides software in 1981) were involved in R&D packages and the related hardware for full text Each product division within the CIT Alca- documentation, storage, and retrieval. The tel companies funds the R&D activities it other Sogitec division produces real-time simu- deems needed. Like Thomson CSF, the basic lators and simulation packages for aircraft, research at CIT Alcatel is performed at a cen- helicopters, land vehicles, and ships. The tral laboratory that is shared with all mem- French military has purchased Sogitec prod- bers of the CGE Group. Approximately 6 to ucts for combat pilot training in the French 8 percent of CIT Alcatel’s total research budg- Mirage fighter bombers. In addition, Sogitec’s et is devoted to this central laboratory.117 simulation expertise is currently being adapted for film production, television com- Sogitec, S.A. mercials, and animation. Unlike the majority of French information Some notion of Sogitec’s research intensity technology firms, Sogitec is not a state-owned can be seen in its distribution of personnel. corporation and is representative of a small in- Over 250 of Sogitec’s 550 employees are scien- formation technology firm. Founded in 1964 tists or engineers involved in research and de- by its president, Christian Mons, Sogitec has velopment (50 are located in the United grown to employ about 550 people in three lo- States). Eighty percent of the researchers cations: Paris, Rennes, and Lakewood, Califor- come from the grandes ecoles system, the re- nia. Sogitec also has sales offices in New York mainder from French universities.ll9 and Washington, D.C. Its sales growth has been impressive; 1978 revenues were doubled by 1979. Customers include General Electric, “’Department of State, “WTDR on Sogitec (SIC) Data Sys- “’Ibid. tems,” Oct. 10, 1983, p. 3. “’Ibid. “gIbid.

The United Kingdom

Historically, the United Kingdom has heav- ing a trade deficit in manufactured products. ily relied upon industrial manufacturing for its Moreover, the U.K. ’S needs for various infor- economic well-being. As the post-industrial mation technology products are completely or society arrives with the decline of manufactur- substantially met by imports. As a result of ing as a primary economic activity, the Brit- the importance of exports to the U.K. econ- ish economy has suffered.120 The United King- omy and the increasing importance of infor- dom’s share of world trade in manufactured mation technology to the world economy, both goods in the decade 1963-73 fell from 15 to 9 the U.K. Government and industry have con- percent. For the first time in history the cluded: United Kingdom now appears to be approach- Our basic economic situation dictates that we must become a net exporter of high tech- ‘z”Daniel Bell, The Coming of Post-Industrial Society (New nology, high value-added products; informa- York: BASIC Books, Inc., 1976). tion technology is a prime example of this. 262 • Information Technology R&D: Critical Trends and Issues

Moreover, unless our [the United Kingdom] kets. This is particularly important in the area information technology industry achieves a of information technology where the develop- strong world competitive position, then the ment of user applications of the technologies efficiency of our other industries in the man- for the provision of new and innovative serv- ufacturing services will suffer. Their capac- ices is vital to the marketing of information ity to be advanced users of information tech- technology. nology has a close synergy with the level of the information technology industry itself.121 Reflecting the need to increase applied re- Traditionally, the United Kingdom has been search activities, the current U.K. Govern- one of the world’s leaders in basic scientific ment has taken measures in the opposite direc- research, particularly in the physical and bio- tion of the traditional high level of government logical sciences. Britain holds the highest per intervention in both industry and social pro- capita ratio of Nobel Prize winners in the sci- grams. The more conservative government’s ences which is more than double those for any efforts include privatizing the economy and re- of the other industralized nations. Although storing entrepreneurial initiatives, thus at- the United Kingdom has in the past had the tempting to make industry more independent largest R&D expenditure percentage of gross of government and reducing government in- national product (GNP) outside of the United volvement in the marketplace. Government States, the United Kingdom has been tradi- strategies include privileged credit, deregula- tionally weak in converting and applying its tion measures, and tax benefits to encourage basic research efforts to the production and the growth of small and medium businesses. marketing of new products and services. Privatization measures also entail exposing This difficulty of transferring scientific state-run monopolies to outside competition. knowledge to commercial applications has con- A major example of this introduction of com- sequently created serious problems in the in- petition is the termination of the monopoly of dustrial sector and in the international com- British Telecom (BT) by granting a license to petitiveness of British-made goods. This a major competitor to operate an alternative failure to capitalize on basic research efforts national telecommunications network. It is is documented in a recent publication of the hoped that this recently introduced competi- Central Office of Information, London, “Brit- tion into the provision of telecommunications ish Achievements in Science and Technology”: services will stimulate demand for advanced technology transmission and exchange equip- . . . there have been an array of ‘firsts’ that ment. In conjunction with the privatization apply to the information technology area, in- measures in the field of information technol- cluding radio navigation, computers, optical ogy, the U.K. Government is also promoting fibers, liquid crystal displays, and flat screen televisions, yet in none of these areas is a the marketing or applied aspects of research British manufacturer a principal supplier. ’22 and development by shifting some of its R&D expenditure from government research estab- Thus the current challenge in the United King- lishments to the private sector. In Cambridge dom is to bridge the gap between the creative for example, where a preference for basic re- basic research in information technologies and search has long prevailed, anew view towards the relatively weak state of industrial applica- research is emerging: tion of these technologies in order to create an environment which is more innovative in pur- What has changed is the idea that money suing both domestic and international mar- is dirty and that one must do pure science. Now, as in Cambridge, Mass., or in Califor- nia, people are not adverse to doing research ‘*’The Department of Trade and Industry, “A Prograrnme that could be put to commercial use. This for Advanced Information Technology, ” the Report of the change in the work ethic has helped us.123 Alvey Committee (London: Her Majesty’s Stationary Office, 1982), p. 14. ‘22Central Office of Information, London, British Achieve- IZ3p~p Ld?cmnkr, “Is Britain Reviving?” World Press Re ments in Science and Technology, April 1981, No. 11 l/RP/81. view, September 1983, p. 31. Ch. 7—Foreign Information Technology Research and Development ● 263

Because of radical changes in government 75 percent of the world market, not one is from policy over the past decades in the United the United Kingdom. Kingdom, the continuity of policy which has, for example, aided Japan in its post-war recov- Government ery, has not been there to support industry in its efforts to expand. This complicates any The U.K. Government officially recognized simple characterization of the U.K. Govern- the importance of the development of informa- ment role in industrial policy and suggests a tion technology to the British society, indus- dynamic situation currently regarding U.K. try, and economy by requesting the Advisory Council on Applied Research and Develop- Government initiatives in information technol- 126 ogy. Consequently, it is difficult at times to ment (ACARD), which advises the Prime disassociate government, industry, and uni- Minister and the Cabinet Office on important versity roles in information technology R&D. developments in advanced technology, to ad- However for purposes of comparison with dress the following questions:

United States, Japanese, and French R&D ef- ● Should development and application of in- forts, government, university, and industry formation technology in the United King- R&D environments are described separately dom be stimulated? below. Before discussing these environments, ● Are there constraints on British industry it is important to understand the size of the which supplies and applies information U.K. participation in information technology technology equipment, software, and markets. systems? The Size of U.K. Participation The resultant 1980 ACARD report “Informa- tion Technology” has contributed to present in Information Technology Markets policy formulation in relation to information While the United States and Japan are net technology, with an emphasis on the applica- information technology exporters, the United tion of information technology as a key ele- Kingdom is a net importer of information tech- ment in the future industrial and commercial nology products by F 300 million ($420 mil- success of the United Kingdom as well as on lion) annually.’” In 1980, the U.K. industry the potential significance of information tech- captured approximately 50 percent of its own nology for both society and individuals. F 2.1 billion ($2.94 billion) information tech- The ACARD report recommended the fol- nology market. Moreover, the U.K. informa- lowing: tion technology industry captured 3.8 percent of the world information technology market 1. One minister and one government de- in1980.125 partment should be wholly responsible for information technology. Because of the relatively small size of its na- 2. One government department should be tional markets, the U.K. information industry responsible for the regulation of commu- has been somewhat inhibited. Moreover, it has nications and broadcasting. had difficulty in generating significant export 3. There should be a government commit- markets for its information technology prod- ment to information technology. ucts to balance its heavy imports. For exam- 4. The government should actively pro- ple, of the 19 top semiconductor companies in mote and publicize British information the world, which account for approximately technology. ——— —— ‘z’Probably the nearest United States counterpart toACARD IZ4AlI U.K. pund fi~res are converted into U.S. dollars ac- is the Committee of Advisors, recently created by Dr. Keyworth cording to foreign exchange rates as of Aug. 1, 1984, where L 1 in the Office of Science and Technology Policy in the White = $1.4. House. ACARD, whose members are experts drawn from in- ‘25”A Strategy for Information Technology, ” National Enter- dustry, governmen~ and academia, offers specific recommen- prise Board, 1981, p. 19. dations for government action.

38-802 0 - 85 - 18 264 ● Information Technology R&D: Critical Trends and Issues

5. Central government, local government, development support and enlightened and the nationalized industries should public purchasing.

apply information technology vigo- 34 Action to make individuals more aware rously. of what information technology offers and 6. The Post Office should provide a so enable them to take advantage of the worldwide network. new information technology products and 7. Education, training, and guidance services. should be accentuated in information 4. The provision of a national telecommuni- technology. cations network capable of stimulating, 8. Links between users and suppliers of in- and meeting, demands for new services. formation technology equipment should For implementation of these goals, 1982 was be improved. designated as IT Year in the United Kingdom. 9. Strong British teams should participate A wide range of promotional aids was used to in international fora on regulations and increase the awareness of the general public, standards. industry, and schools, and main procurement 10. Legislation should be introduced to pro- agencies. The major force in IT Year, in addi- vide for better protection of data, and tion to the information technology awareness other legal reforms will be required. campaign, was a major and intensive govern- 11. The Post Office monopoly on use of its ment-industry initiative to encourage re- services should be ended. search, development, and application of infor- 12. Public purchasing should be used to mation technology in order to help strengthen “pull through” development of equip- the overall U.K. economy. ment. 13. The Science and Engineering Research Also during 1982, the British Government Council and the Department of Indus- acknowledged the need to address the field of try should promote research and devel- information technology in a coherent manner opment in information technology. and subsequently made significant changes in 14. All publicly funded information technol- its policymaking structures. A minister with ogy research and development should be special responsibilities for information tech- coordinated. This would involve Depart- nology was appointed, the first such appointment of Trade and Industry, Ministry ment in any nation. The Minister for Indus- of Defense, the Post Office, and the Sci- try and Information Technology has ence and Engineering Research Council. responsibility, under the Secretary of State for Trade and Industry, for all the Department The United Kingdom has taken a wide-scope of Trade and Industry’s activities concerned approach to encourage the development and application of information technology in Brit- with information technology, including those ish industry and society. In response to the related to research and development. More ACARD report, Mrs. Thatcher’s government specifically, the Minister has responsibilities has developed an overall strategy for informa- for information technology, telecommunication technology. The main objectives of the tions, computer systems, microelectronics, U.K. Government’s policy for information electronics applications, robotics, and space. technology are: He oversees British Telecom and the Post Of- fice public purchasing, research and develop- 1. The development of a statutory regula- ment (including the industrial research estab- tory framework favoring the growth of in- lishments) and the British Technology Group. formation technology products and He is also responsible for sponsorship of the services. chemical, mechanical and electrical engineer- 2. The development of new products and ing, and paper industries; and for distribution techniques through direct research and and service trade industries; newspapers, Ch. 7—Foreign Information Technology Research and Development . 265 printing and publishing; and for standards and Mechanical and Engineering; and Metrology quality assurance and firms. Although the re- and Standards, are involved with information sponsibilities of the Minister of Industry and technology R&D support. The Computers, Information Technology are central to the pro- Systems, and Electronics RRB is the major motion of information technology in the supporter of information technology R&D United Kingdom, at times the press coverage with 1980-81 expenditures of F 6 million ($8.4 of ministry activity tends to be exaggerated. million). 128 In keeping with the current U.K. Govern- Department of Trade and Industry (DTI) ment policy, which aims to achieve a profita- As a result of the 1980 ACARD recommen- ble, competitive, and adaptable private sector, dations which suggested increasing coordina- the DTI has recently implemented a variety tion between industry and government, the of programs that are intended to supplement Department of Trade and Industry (DTI) has direct R&D support. Because of the recent re- become the major focus for U.K. initiatives in cession that has inhibited U.K. companies areas related to information technology. In re- from investing large resources in R&D activ- cent years there has been a shift in U.K. Gov- ities, these programs are intended to encour- ernment support from its own research estab- age the private sector to research, develop, and lishments (which still account for a substantial use information technology. The DTI’s sup- part of the nation’s scientific resource) to the port programs generally comprise three private sector. The DTI total expenditure in elements: 1982-83 was approximately F 230 million 1. An awareness program to stimulate inter- ($322 million), of which approximately two- est in the potential of the new technology. thirds was spent in the private sector. DTI al- 2. Consultancy to explain how a particular located approximately F 80 million ($112 mil- technology can be applied to a particular lion) for information technology R&D in 1979- 27 company’s needs. 80.’ This trend is greatly supported by the 3. Support for ensuing projects. U.K. Government because it locates research closer to the point of application (which some These DTI R&D support programs, catego- feel is vital in the area of information technol- rized by the various applications of informa- ogy) and encourages private sector initiatives. tion technology, are presented in table 44. The DTI’s decisions on funding priorities for THE ALVEY PROGRAMME FOR ADVANCED research and development rely on the advice 1NFORMATION TECHNOLOGY of five Research Requirements Boards (RRBs). In addition to industrial R&D support Each RRB is chaired by a Senior Industrialist schemes, the DTI recently initiated a collab- and consists of industrialists, scientists, and orative national information technology R&D government representatives. The RRB’s are program. One of the catalysts to the forma- seen as an effective system for monitoring and tion of the Alvey Programme was the an- developing strategies to ensure that research nouncement of the plans for Japan’s Fifth- support priorities match the demands of Generation Computer Systems Project and Ja- changing technologies and future industrial pan’s invitation to other countries, including needs. The DTI, therefore, works closely with the United Kingdom, to discuss participation industry through the Research Requirement in the program. The scale and cohesiveness of Boards to ensure that research in the national this and other Japanese programs were seen laboratories is directed towards industrial by the U.K. representatives to the Japanese needs. Three of the Research Requirements conference as a major competitive threat. The Boards, Computers, Systems, and Electronics; British also believed that the U.S. industry’s ‘*’John K. Thompson, “ ‘IT’ in Britain, ” speech given at the British Embassy to the Potomac Chapter of the American Soci- ‘z8Statement of Kenneth Baker, before the U.K. House of ety of Information Scientists, June 9, 1982. Commons, Dec. 21, 1982. 266 ● Information Technology R&D: Critical Trends and Issues

Table 44.—Department of Trade and Industry R&D Programs

CAD/CAM — The Computer-aided Design and Computer-aided Manufacture program is designed to promote and accelerate the acceptance and application of CAD/CAM primarily in the mechanical and electrical engineering - 6 million ($8.4 million) (over 3 years from industries. 1981) CAD/MAT — The Computer-aided Design Manufacture and Test Program is designed to encourage the use of CAD/MAT for design testing and production in the electronics - 9 million ($12.6 million) (over 3 years from industries. 1982) FMS — The Flexible Manufacturing Systems Scheme is designed to encourage fiirms to install flexible manufacturing - 60 million ($84 million) (over 3 years from systems. 1982) FOS — The Fiber Optics and Opto-Electronics Scheme is designed to encourage the development, production, -25 million ($35 million) (over 5 years from and application of fiber optics and opto-electronics. 1981) MAP — The Microelectronics Application Project is designed to -55 million ($77 million) (commenced 1978) encourage the application of microelectronics in products and processes in manufacturing industries. MISP — The Microelectronics Industry Support Program is -55 million ($77 million) (commenced 1978) designed to promote the microelectronics components industry, particularly for the manufacture of silicon integrated circuits. ROBOTICS — The Industrial Robotics Scheme is designed to encourage the development and application of robots + 10 million ($14 million) (over 3 years from in manufacturing industries. 1981) SPS — The Software Products Scheme is designed to encourage the development and application of software products and packages. Awareness programs, demonstration projects and other promotional +80 million ($1 12 million) (over 4 years activities commencing 1981)

SOURCE: Office of Technology Assessment.

reaction to the Japanese Fifth Generation ment will be coming together to achieve Computer Systems Project could also cause major advances in technology which none an equal if not greater degree of competition could achieve on their own. The involvement for the U.K. industry. of industry will ensure that the results as they emerge are fully exploited here in Brit- In the light of these factors, the U.K. delega- ain to the advantage of our economy. Infor- tion called for an urgent study into the feasi- mation technology is one of the most impor- bility for a collaborative R&D program geared tant industries of the future and therefore to particular strengths and requirements. This one upon which hundreds of thousands of study, completed by the Alvey Committee, jobs in the future will depend. Collaboration will ensure that the results of the research are outlined plans for a national information tech- widely disseminated particularly to smaller nology R&D effort to improve the United firms which have such an important contri- Kingdom’s competitive position in world in- 129 bution to make to the industry. No one can formation technology markets. John Alvey, guarantee success, but the government is Chairman of the Committee, comments on the convinced that this program will ensure for coordination aspects of the program: British industry secure access to the new technology and to the products and processes This is the first time in our history that we on which our future prosperity depends. shall be embarking on a collaborative research project on anything like this scale. In- The Alvey Program for Advanced Informa- dustry, academic researchers, and govern- tion Technology is a 5 year program funded ——z —-—-— ‘ ’See “A Pmgr amme for Advanced Information Technology, by three government ministries and industry. The Report of the Alvey Committee, ” Department of Indus- Total funding for the program will be approx- try, Her Majesty’s Stationary Office, 1982. imately $525 million over a 5 year period. The Ch. 7—Foreign Information Technology Research and Development . 267

Department of Education and Science (DES) The United Kingdom will also be a major through the Science and Education Research participant in the European Strategic Pro- Council (SERC) will fund approximately 50 gram for Research in Information Technology million ($70 million) for promoting advanced (ESPRIT). Currently, U.K. companies are in- research in academic institutions and the volved in more than half of ESPRIT’s pilot training of necessary manpower. The Minis- projects. The Alvey program is also designed try of Defense (MOD) will fund approximately to complement the ESPRIT program with in- %40 million ($56 million) for research believed terl.inking communications networks and com- to be important to the defense industry and mon parallel research strategies. will contribute its experience in the field of integrated circuits. The Department of Trade THE DEPARTMENT OF TRADE AND and Industry will provide the major portion INDUSTRY’S NATIONAL LABORATORY of the government’s funds, approximately FACILITIES L 110 million ($154 million), and will have the The U.K. national research establishments overall responsibility for management of the involved with information technology R&D program. Industry will fund the remaining are the National Physical Laboratory, the Na- L 150 ($210 million) in the form of 50 percent tional Engineering Laboratory, and the Com- 130 matching funds for each R&D project. puted-Aided Design Centre. The Alvey Program R&D projects are con- The National Physical Laboratory (NPL) centrated in four technical areas, known as has a long distinguished history in computer “enabling technologies. ” These enabling tech- technology. In 1981-82, NPL R&D expendi- nologies, seen as crucial to the development tures were approximately L 22 million ($30 and application of information technology in million). Currently, R&D projects are focused the United Kingdom, include: on data networks, data security, special input devices, and microprocessor applications. The ● very large scale Integration (VLSI) silicon Laboratory also is developing standard net- integrated circuits, work protocols and evaluating cryptographic ● software engineering, methods for data protection. NPL also devel- ● intelligent knowledge based systems ops software for solving engineering problems. (IKBS), and ● man/machine interface. The National Engineering Laboratory (NEL) carries out research, development, de- The research projects in these four technical sign, consulting, and testing in automated areas will be managed by the Alvey Director- manufacturing. The 1982 Information Tech- ate consisting of staff from industry, DTI, nology Year campaign highlighted the Labora- MOD, and SERC. Each of the four technolo- tory’s involvement in robotics and automated gy areas has its own director in addition to a production systems. NEL R&D expenditures director in charge of networks and communi- were approximately L 16 million ($22.4 mil- cations among the various R&D projects. lion) in 1981-82. Key areas of NEL research Each of the research teams will generally be include automated assembly, control and op- organized in small consortia—e.g., two infor- timization of production systems, the devel- mation technology firms, together with a gov- opment of an advanced turning cell, and other ernment research establishment team, and a flexible manufacturing systems. university team. Unlike the Japanese approach of creating a center for research, re- Computer-Aided Design Centre (CADCENTRE) search teams will rely on a data network and is the primary center in the United Kingdom electronic mailbox service that will allow in- l30 Discussion of the U.K. National Research Labs, based on teractive communication among the R&D pro- “Research and Development Report, 1981 -82,” Department of gram participants. Industry, 1982. 268 . Information Technology R&D: Critical Trends and Issues for the development of computer techniques overseas defense electronics market is fairly in design and engineering. In 1981-82 the strong, it is difficult to predict what the long- CADCENTRE R&D expenditures were approx- term consequences of this shift maybe. If the imately L 4 million ($5.6 million). The center market grows for defense electronics, there offers approximately 30 computer software could be a positive effect on the U.K. supplier packages in CAD/CAM and provides a com- industry. On the other hand, because the spin- prehensive range of services including soft- off effect of U.K. military R&D to commercial ware development, consultancy, production products has not been particularly significant, services and the provision of hands-on experi- any increase in attention to military needs by ence in CAD/CAM techniques. Because pri- the limited U.K. electronics industry might vate sector initiatives were encouraged in DTI have the effect of further reducing their civil- sponsored R&D, the DTI agreed to sell the ian-oriented work and reducing their market Cambridge-based CADCENTRE to a U.K. competitiveness. consortium led by the U.K. computer firm In- ternational Computers, Ltd. (ICL) for approx- Department of Education and Science (DES) imately L 1 million ($1.4 million). The newly privatized CADCENTRE employs ICL staff Under the advice of the Advisory Board for Research Councils (ABRC), the Department and DTI management staff. In addition, the of Education and Science (DES) allocates the DTI has agreed to provide some financial sup- science budget to five Research Councils. The port in order to ease the transition from a gov- Science and Engineering Research Council ernment research establishment to a commer- (SERC) has the primary responsibility for in- cially run company. The DTI will be entitled formation technology R&D. The SERC budg- to a royalty based on the CADCENTRE’S 131 et allocated to information technology R&D turnover. was L 5 million ($7 million) in 1979-80, L 8.5 million ($11.9 million) in 1980-81, and L 11 The Ministry of Defense (MOD) million ($15.4 million) in 1981-82.*33 The Ministry of Defense (MOD) is also a ma- SERC, analogous to the U.S. National Sc - jor supporter of applied R&D. Almost half of ence Foundation, accepts competitive bids the United Kingdom’s R&D budget has been from universities for special projects. Unlike devoted to defense. In the 1970s, defense R&D NSF, SERC provides only half the funds for remained relatively constant although all sponsored projects. The rest of the funding other areas of R&D decreased by almost 50 must come from industry, charitable founda- percent. In 1978, the Ministry of Defense ex- tions, or other government departments. penditures for electronics research and devel- 132 SERC also provides research establishments opment were approximately $900 million. with central computing support. The Engi- Beyond the direct support from industrial neering Board of SERC is most involved with research funds, the defense sector has poten- information technology, with funding areas in: tially the greatest possibility for contributing ● device-related research, to civil information technology. There has ● skilled manpower training, been a recent shift in emphasis within the • software technology, database utilization, MOD away from aircraft and towards elec- and system reliability, tronics research and development. Because the ● distributed computing systems. ‘g’ Under the royal~ arrangements, the U.K. Government will SERC operates the Rutherford Appleton Lab- be repaid and could receive a further net amount of L 4.5 million over a 10-year period, assuming forecast revenue levels are oratory (RAL). The lab itself has a consider- achieved by the company. able research program in information technol- ‘s’J. Thyme, “Information Technology in the U. K.: Gover- nment Policy, ” in G.P. SweeIWy (cd.), Informah”on and the 7hns- ‘S9John Thompson, “ ‘IT’ in Britain,” speech given at the Brit- formation of Society, North-Holland Publishing Co., 1982, p. ish Embassy to the Potomac Chapter of the American Society 261. of Information Scientists, June 9, 1982. Ch. 7—Foreign Information Technology Research and Development ● 269 ogy, with efforts in distributed computing to become top-heavy with senior, relatively ex- systems, industrial robotics, computing appli- pensive faculty. The steadily rising cost of re- cations in engineering, electron-beam lithog- search equipment has also affected the avail- raphy, and image processing. able funds for university research. Together, these three factors have resulted in substan- British Technology Group (BTG) tially less spending for university R&D.134 The British Technology Group is an inde- Although university research funding has pendent public corporation established to pro- been decreasing over the last few years there mote the development and application of new have been marked changes in the distribution technology. The British Technology Group of funds—away from basic research towards was formed in 1981 and includes the former engineering and applied R&D. This trend National Research Development Council towards greater emphasis on industrial ap- (NRDC) and the National Enterprise Board plication of research results is exemplified in (NEB). several recently initiated schemes. These schemes, designed to promote high quality re- BTG provides funds for technological inno- search in fields of applied science, are cospon- vation through: sored by SERC and DTI. In one of the more ● joint venture finance under which BTG successful joint SERC/DTI initiatives, the can provide 50 percent of the funds re- Teaching Company Scheme, DTI pays for en- quired for the business in return for a levy gineers to do postgraduate work in industry. on sales of the resulting product or Students work on product development, de- process; sign, and manufacturing processes, but with ● recirculating loans, which are a form of close attention and support of academic staff working capital loan, through which BTG who supervise the innovative aspects of the can help a company to meet specific or- graduates’ work. The program so far has at- ders for innovative products; and tracted great interest from industry, and ● equity and loan finance on venture capi- many firms such as General Electric Co., Brit- tal terms where a company is set up for ish Aerospace, Ferranti, and IBM are partici- the purpose of developing and marketing pating. an invention or new technology; equity Another program aimed at promoting indus- may also be provided in the form of re- deemable preference shares. trial training is the introduction of approximately 150 information technology centers Projects from all sectors of industry are (ITEC centers) throughout the United King- eligible for consideration. The primary consid- dom. Funded by DTI, Manpower Services eration for BTG support is that the proposal Commission, and industry grants, these cen- project must be based on a new invention or ters collaborate with local industries to pro- a genuine technical innovation. vide computer training for unemployed high- school age youths. Approximately 70 percent of the students upon leaving the centers find University employment in a computer-related field. Other Basic research is supported primarily at the programs that offer technical training include university level (approximately 22 percent of Computers in Schools, through which DTI government R&D spending) both by discipline funds 50 percent of the cost of up to two micro- oriented committees of the Science and Engi- computers in each U.K. school, and a Manpow- neering Research Council (SERC) and the Uni- versity Grants Committee (UGC). Recent cuts in university funding have been substantial— 194 Robin B. Nicholson, “Science and Technology Policy in the United Kingdom, ” address given at the Nineteenth Annual 15 percent over the last 4 years. In addition, Meeting of the National Academy of Engineering, Nov. 2-3, lowered enrollment has caused the universities 1983. 270 • Information Technology R&D: Critical Trends and Issues er Services Commission Program that funds thus broadening the ICL product line into information technology training in computer- more powerful computers. It is hoped to ex- related subjects. Approximately half of all the tend this collaboration at a later date to other progr ammers training in the United Kingdom technology areas, including communications is provided by these programs. technology. ICL also recently reached an agreement with a small U.S. company, Three In addition to industrial training programs, Rivers, to manufacture and market worldwide the U.K. Government has encouraged greater a microcomputer designed by Three Rivers. industrial participation in R&D as well as closer industry-university ties through the ini- In 1979, the U.K. Government also invested tiation of science parks. Similar to the Japa- $100 million for the creation of a national semi- nese and U.S. science parks, the U.K. science conductor firm, Inmos. Operating as an inde- parks provide an opportunity for industry to pendent producer, Inmos was established to locate in the proximity of major research manufacture a limited range of products (prin- universities. cipally high-capacity semiconductor memory chips) to sell to large electrical goods manufacturers. Also, to assure indigenous produc- Industry tion of semiconductors, the U.K. Government has convinced several U.S. semiconductor Since the late 1960s, the U.K. Government companies to setup production in the United has played an important role in supporting the Kingdom by offering them grants up to 33% U.K. computer industry. For example, in 1968 percent for R&D costs, under the Support for the government encouraged a series of merg- Innovation Program (SFI). ers that led to the establishment of Interna- tional Computers Ltd. (ICL) and then provided Currently, more than half of the U.K. industrial electronics R&D expenditure is funded by funds amounting to approximately $12 million government. The funding may be in the form annually until 1976. Moreover, the government encouraged ICL’S growth through pref- of cost-sharing grants, procurements, or any erential procurement policies that guaranteed of the multitude of funding schemes. More- almost all large central government contracts over, the pervasive influence of the U.K. Gov- to ICL. ICL became the largest European com- ernment is exemplified in the original Alvey proposal which called for the government to puter company, with a wide customer base in fund 90 percent of industrial R&D activities. the United Kingdom as well as overseas, in- The proposal, however, was finally amended cluding the United States. ICL became so successful that government assistance was with- to the current commitment of 50 percent only after considerable debate, on the grounds that drawn in 1976, and in late 1979 the government sold its 25 percent share in the company. industry would not be committed if it only was required to invest 10 percent of its own re- However, by 1981, ICL was heavily in debt, sources. and the government (although it encouraged private funding and joint research programs) The historically low industrial funding for arranged L 270 million ($378 million) in loans information technology R&D has been attrib- for ICL. uted to the U.K. information technology in- In an effort to become profitable again, ICL dustry’s focus on highly specialized markets. Because the U.K. information technology in- is currently exploring joint activities with non- U.K. firms. In 1982, ICL and Fujitsu reached dustry in some instances fills small but important market niches, it usually does not capture a collaborative agreement. The arrangement mass markets; consequently, the U.K. infor- provides for special access to Fujitsu’s ad- mation technology industry has not always vanced microelectronics technology, and pro- had adequate resources for R&D funding. vides for purchases of semiconductor chips developed by Fujitsu. ICL will in turn market A case in point is the British semiconduc- large Fujitsu computers in Western Europe, tor chip industry. The leading U.K. informa- Ch. 7—Foreign Information Technology Research and Development ● 271 tion technology firms have concentrated on Plessey producing specialized chips known as “spe- cials” where the demand is low and the mar- In 1983, Plessey invested approximately ket relatively small. For example, world sales $225 million in R&D, 15 percent of its sales. in 1980 of custom-built special chips came to Approximately $22 million of the R&D ex- only L 100,000 ($140,000) almost one-twen- penditures were allocated to basic research tieth of the comparable figure for standard activities. Of the 320 researchers in the main 135 Plessey laboratory working on microelectron- chips. Although the leading British electronics, 170 people are working on gallium arsenide ic firms’ demand for these standard chips are and related materials and 150 are working on extremely high, U.K. industry has left the manufacturing of these types of chips mainly silicon. Plessey’s silicon research is geared to companies from the United States and Ja- towards speciality or custom circuits. pan, which together capture approximately 75 Plessey relies heavily on outside contracts, percent of the world market. However, the and approximately 45 percent of its research United Kingdom’s leading firms—Ferranti, is done for the MOD and British Telecom. Of Plessey, and GEC—all plan to expand their the remaining 55 percent, half is conducted for manufacturing capability of standard chips for the Plessey operating divisions and half is con- domestic use. This pattern has also been re- ducted for the head office. Currently, Plessey peated in other information technologies; for is attempting to reduce its reliance on outside example, few British companies produce hard- R&D funding in favor of more operating divi- ware that directly serves the semiconductor sion work.137 and electronics industry (i.e., for testing and production needs). General Electric Co. (GEC) There is some indication that U.K. industry In 1983, the General Electric Co. (GEC) in- may be reassessing its investment policies. vested approximately $900 million in research Current estimates suggest, for example, that and development, approximately 10 percent the volume of R&D has been maintained of its sales. Unlike Plessey, only 25 percent of through the recession on a selective basis with GEC’s research is for the MOD and British substantial advances in areas such as micro- Telecom. Fifty percent of the research is for electronics and corresponding decreases in GEC’s 120 operating companies, and 25 per- metals and traditional engineering. For exam- cent is for basic, speculative research. Because ple, industrial electronics R&D expenditures a large percentage of its R&D is supported by increased form L 279 million ($390.6 million) its operating companies, GEC’s research activ- in 1975 to L 442 million ($618.8 million) in ities are more oriented towards commercial 1978.136 product development. GEC’s key areas of re- Similar to the situation in France and Ja- search include microelectronics, fiber optic devices, software engineering, and custom chip pan, a few large firms are responsible for a 138 large proportion of information technology design. R&D expenditure in the United Kingdom. British Telecom (BT) Several of these major firms are described below. Previously a public monopoly, the U.K. Gov- ernment has recently privitized British Tele- com (BT). Moreover, the U.K. Government has permitted the licensing of other competitors

‘35 Peter Marsh, “Britain Faces Up to Information Technol- ogy, ” New Scientist, Dec. 23, 1982, p. 637. “’Robin Nicholsou “Science and Technology Policy in the “’’’Profile: GEC and Plessey: Two Approaches to R& D,” The United Kingdom, ” address given at the Nineteenth Annual Econom”st, Nov. 20, 1982. Meeting of National Academy of Engineering, Nov. 2-3, 1983. ‘3’’ ’Ibid. 272 . Information Technology R&D: Critical Trends and Issues to provide telecommunications services. It is search expenditure is approximately L 100 hoped that the introduction of competition million ($140 million) annually. BT has tradi- into the provision of telecommunications serv- tionally had strong research programs in four ices will stimulate new demand and provision major areas: of a vast array of new services. Consequently, ● advanced technology (e.g., CAD in large U.K. companies with advanced products are scale integrated circuit design, optical expected to be well placed to take advantage communications systems, gallium arse- of market development and to use the United nide, high speed logic); Kingdom as a springboard for capturing Euro ● transmission (e.g., digital transmission in pean and other world information technology LANs, small earth station satellites); markets. ● customer service/apparatus (e.g., Prestel The privatization efforts, however, have and Viewdata); caused some speculation about the future of ● advanced systems (e.g., microprocessor BT laboratories in much the same way that software development ISDN local con- divestiture has caused speculation about the nection). future of Bell Labs. Currently, BT Labs re-

European Strategic Program for Research in Information Technology (ESPRIT) In an attempt to reverse the decline of Eur- ing sales; a domestic market which, unlike ope’s competitiveness, to ensure a stronger those of the United States and Japan, is frag- technological base, and ultimately to ensure mented into a number of relatively small na- economic and political independence, the Com- tional units; and reluctance by some within in- mission of European Communities proposed dividual countries to subsidize other nations the European Strategic Program for Research who have historically been economic and po- in Information Technology (ESPRIT).139 Al- litical rivals.140 Consequently, the ESPRIT though EEC members hope that ESPRIT will program seeks to: increase the size of research succeed in strengthening Europe’s technologi- teams, optimize the use of human and finan- cal base and international competitiveness, cial resources, and initiate definition and adop- there are concerns as to whether research can tion of European standards for information be effectively undertaken and shared among technology products.141 potential international competitors. ESPRIT attempts to address the problem ESPRIT is designed to address three ma- of the fragmentation of the European market, jor difficulties that currently face the Euro- as well as the divided efforts of the individ- pean information technology industry as it at- ual member nations’ national R&D support tempts to develop new state-of-the-art tech- programs, by linking a significant proportion nologies: the problem of raising adequate of key European engineers and scientists from funds for long-term research and development government, industry, and universities. In this during a period of economic recession and fall- respect, ESPRIT is similar to other national .——— research and development programs such as ‘sgIn 1975, the European Community had a trade surplus in information technology products; however, by 1980, the trade Japan’s Fifth-Generation Computer Project, deficit in information technology products reached $5 billion the United Kingdom’s Programme for Ad- and reached approximately $10 billion in 1982. At present, Eur- vanced Information Technology, and the U.S. ope represents one-third of the world information technology market but accounts for only 10 p-cent of world information Semiconductor Research Corporation (SRC), technology production. For example, in the European domes- which are also partnerships among various tic market, 2 out of every 5 information technology products companies, academic research laboratories, sold are European; 8 out of 10 personal computers sold in Eur- ope are imported from the United States; 9 out of 10 videotape “=David Dickson, “Europe Seeks Joint Computer Research recorders sold in Europe come from Japan. ESPRIT Proposal, Effort,” Science, Jan. 6, 1984, p. 28. COM(83) 258 final, Commission of the European Communities, “’ESPRIT Proposal, COM(83) 258 find Commission of the June 2, 1983, p. 8. European Communities, June 2, 1983, p. 9, Ch. 7—Foreign Information Technology Research and Development . 273 and government agencies. However, unlike because of their perceived importance for fu- these other research and development pro- ture European industrial competitiveness. The grams, ESPRIT also represents an interna- first three of these research areas were selected tional institutional arrangement. in part to develop better enabling or core technologies. Office automation and computer in- Unlike the MITI program in Japan, the ESPRIT program is limited to precompetitive tegration were selected as specific applications areas where information technology is ex- research; it does not intend to develop a com- pected to have a large economic and social im- mercial product as does the Japanese fifth- pact–automation of the office and the factory. generation computer effort. At any point in the research, however, the participating com- The advanced microelectronics project’s ma- panies are free to take the technical results jor goal is to develop smaller, more reliable, gained from the ESPRIT projects and develop and more powerful integrated circuit technol- commercial products on their own. Therefore, ogy so that devices can perform more func- ESPRIT is not seen to be in competition with tions or operations than circuits available national research and development programs today. More specifically, the goal of the ad- or individual companies, but as a reinforce- vanced microelectronics project is to improve ment to make them more effective. the current state-of-the-art process, which is based on three-to-five micrometer structures, Funding for ESPRIT is approximately $1.3 to processes that are based on structures billion for the next 5 years and approximately smaller than one micrometer. The Europeans 2,000 researchers will take part in research activities. 142 The funding be equally shared are hopeful that this advanced microelectronics project will improve Europe’s current among 12 principle corporate partners and integrated circuit trade deficit.143 other participating companies with the Com- mission of European Communities. The 12 The major emphasis of the advanced infor- main corporations participating in ESPRIT in- mation processing project is on information clude: General Electric Co. (United Kingdom), and knowledge engineering, information stor- Plessey plc. (United Kingdom), International age and usage, signal processing, and exter- Computers Ltd. (United Kingdom), Compag- nal interfaces. The research project’s overall nie General de L’Electricity (France), CIT- goal is to develop technological capabilities Alcatel (France), Cii Honeywell Bull, (France), that underlie machine intelligence. Advances Thomson-Brandt (France), AEG-Telefunken in the new types of information processing will AG (West Germany), Nixdorf Computer AG also entail breakthroughs in advanced comput- (West Germany), NmbH Phillips Gloeilampen- er architecture, further miniaturization in fabrieken (Netherlands), Olivetti SPA (Italy), microelectronics, and higher reliability. and Societa Torinese Esercizi Telefoncini The goal of the software technology project (Italy). is to improve software engineering techniques. The ESPRIT program consists of five ma- More specifically, the project’s goals include jor research areas: advanced microelectronics, establishing standardized software interfaces, software technology, advanced information automating the software engineering process, processing, office automation, and computer and disseminating and centralizing software integrated manufacturing. The EEC has cho- research results in a common database so that sen these five areas of information technology individual modules of software programs can ———— be reused where similar functions are re- ‘42Actual funding for ESPRIT is 1,500 million European cur- quired. rency units (ECU). Approximately 1 AU equals 1 U.S. dollar. Of the total European Community countries’ research and development expenditures, 1.7 percent is for EEC research and development activities. Nine percent of the EEC R&D budget ‘43Currently, Europe absorbs 20 percent of the world’s inte- is reserved for industry of which 20 percent is for ESPRIT grated circuit market, although Europe produces only 6 per- funding. cent of the world’s integrated circuits. 274 • /formation Technology R&D: Critical Trends and Issues

The office automation project, one of the two The Commission and the advisory board, specific applications projects, is directed at de- which consists of industry representatives veloping a multimedia interface for all office (mainly from the 12 contributing industries), communication needs, and at developing effi- review research proposals and approve grants. cient electronic filing systems for unstructured Two broad types of proposals are considered information (text, voice, graphics, images, for the different technical projects. Type A, etc.). The project will also examine the cross- which represents the strategic long-term re- cultural interaction of human factors, educa- search activities of ESPRIT, involves large re- tional, sociological, and industrial effects of of- search establishments and large commitments fice automation systems. Moreover, research of resources, both human and financial, as on machine translation (which is of great im- well as clear long-term strategic plans to portance to the European Community) and as- ensure continuity of research and long-term pects of machine-user interfaces, such as in- benefits. Type A projects receive 50 percent tegrated image test speech communication, funding from the European Community, and document creation, and distribution will also the research participant is expected to provide be conducted. the remaining funds. Type B proposals require relatively smaller resources and account for a The goal of the computer integrated manu- significant share of the overall efforts under facturing project is to develop improved sys- ESPRIT. Type B projects could range from tems for automated factories. These systems very long-term, very speculative R&D to will integrate in a common data base comput- shorter-term and more specific R&D. Type B er-aided design, computer-aided manufactur- projects receive at least 50 percent of their ing, computer-aided testing and repair, and R&D funds from the European Community, assembly. Such integration will require further or more if the applicant is from an academic developments in integrated systems architec- institution or smaller business with limited ture, advanced components, real-time based available finance. imaging, and integrated control subsystems mounted on semiconductor chips. One of the first ESPRIT research proposals to be funded is a project to develop advanced Experts from the 12 main industry partners interconnection between very-large-scale inte- as well as outside consultants have developed grated circuits. The research project is a joint a work plan of specific projects in the five proj- effort between Plessy and GEC in the United ect areas. As these specific projects have been Kingdom, Thomson CSF in France, and Tele- solicited. Proposals may defined, proposals are funken in West Germany. Another initial proj- be submitted by research teams from any in- ect, jointly shared between the Polytechnic in dustry or university, although the team must London and the University of Amsterdam, is be composed of nationals from two or more focusing on the development of 11 different EEC countries. This arrangement is meant to aspects of tools and methods for developing encourage international cooperation between machine intelligence. nations and industries, and therefore prevent duplication of research efforts and make op- The Future of ESPRIT timal use of limited financial and human resources. The main criteria for evaluating pro- The initial response to the ESPRIT pilot posals include: technical soundness; contri- phase has been favorable; the 1 year pilot bution to industrial strategy in light of phase of ESPRIT, launched in mid-1983 with ESPRIT objectives; European Community a budget of $20 million and funded 50 percent usefulness; technical, scientific, and mana- by the European Community and 50 percent gerial capability to undertake the proposed by industry, attracted over 200 research pro- project; and proposed activities that will fa- posals. However, only 36 could be selected to cilitate the dissemination of research results. receive EEC matching funds. EEC officials Ch. 7—Foreign Information Technology Research and Development ● 275 were quite surprised not only at the scale of ects. Consequently, there are debates in each response, but also at the apparent willingness country over whether the results of informa- of companies to permit their scientists to work tion technology research and development, as together with few restrictions.144 an important key to future political and economic strength, should be shared with poten- Currently, however, there is still some doubt tial competitors. as to whether ESPRIT will further achieve its stated goals. In the past, the EEC has had International rivalry is also a large indus- some difficulties with other joint projects. In try concern. The lack of an established legal the 1960s, Pierre Aigrain, then President Pom- framework in which companies will be allowed pidou’s chief scientist, proposed a European to collaborate may cause difficulties for Euro- project to strengthen the technological base pean information technology industries. Be- of the computer and communications indus- cause ESPRIT is concerned primarily with try. However, the project was never launched long-term precompetitive research, there may because of resistance from the European tele- not be any conflict with the EEC antitrust communications monopolies to the suggestion laws which are intended to apply primarily to that their own research was insufficient and marketing strategies, rather than product de- the lack of a plan to suggest how a coopera- velopment. However, some companies believe tive research program supported by estab- that as the gap between scientific discovery lished companies and nationalized industries or basic research and commercial application could be beneficial to all the participants.145 narrows, collaborative precompetitive re- Moreover, in the early 1970s, Dutch, German, search will be extremely difficult.148 It is for and French computer manufacturers formed this reason that three of Europe’s largest a joint venture, Unidata. Each company sent mainframe computer manufacturers (Siemens, their top engineers to a joint research and de- ICL, and Bull) have established the European velopment facility. However, the project had Computer Industry Research Center. The Re- many difficulties and some of the participants search Center discourages open collaboration eventually withdrew from the project.146 by initially excluding participation by other companies because: As in past joint European research efforts, the future of ESPRIT depends as much on the If you recognize the fact that you are in a results of political struggles around the re- competitive market, in which companies are structuring of Europe’s economic and indus- fighting against each other, then you must trial base as it does on any judgment of its accept that it is not in the interest to offer technical and scientific merits.147 This is illus- all research results to everyone who might be trated by the first few unsuccessful attempts interested in them, and that at least some in early 1984 for EEC endorsement of ESPRIT, projects will be of character that will forbid the open publication of research results from which became intermingled with broader eco- the beginning. . . . We do abstract research nomic issues ranging from the efficiency of on an international basis where the balance French farming practices to the EEC’s bud- of cooperation and competition should be get procedures. Moreover, it is still unclear as determined by the rules of international com- to how the national information technology re- merce, and market-like research on a national search programs will mesh with ESPRIT proj- basis, where individual companies can adopt the most appropriate strategies for their do- ——-— . — mestic and political environment.149 “’David Dickson, “Europe Seeks Joint Computer Research Effort, ” Science, Jan. 6, 1984, p. 28. “’’’What Hope for ESPRIT?” Nature, Feb. 16, 1984, p. 582. ——...— “’Beth Karlin and George Anders, “Europe Looks Abroad 148This problem maybe particularly acute in research projects for High Technology It Lags in Developing, ” The Wall Street such as office automation and computer-aided manufacturing. Jourzud, Oct. 5, 1983, p. 1. “’Statement by Mr. Heimann of Siemens, David Dickson, “’David Dickson, “Europe Seeks Joint Research Effort, ” Sci- “Europe Seeks Joint Computer Research Effort, ” Science, Jan. ence, Jan. 6, 1984, p. 28. 6, 1984, p. 29. 276 ● /formation Technology R&D: Critical Trends and Issues

Despite the preliminary difficulties in achiev- ments. If ESPRIT does succeed, it is likely to ing agreement on the funding for ESPRIT, the be used as a model for similar cooperative establishment of the controversial Siemens- European Community projects in other fields, ICL-Bull Research Center, and other under- such as telecommunications and biotech- lying political and economic rivalry, ESPRIT nology. may succeed in overcoming these impedi- Chapter 8 Information Technology R&D in the Context of U.S. Science and Technology Policy Contents

Findings ...... 279 Introduction ...... 279 Part I: Background ...... 280 General U.S. Science and Technology Policies ...... 280 Information Technology R&D Policies ...... 286 Part II: Key Issue Areas...... 290 Issue A: Organizationof Government ...... 290 Introduction ...... 290 Dimensions of the Issue ...... 291 Options for Addressing the Issue ...... 293 Issue B: Military/Civilian Balance ...... 294 Introduction ...... 294 Dimensions of the Issue ...... 295 Options for Addressing the Issue ...... 297 Issue C: International Competitiveness ...... 298 Introduction ...... 298 Dimensions of the Issue ...... !...... 298 Options for Addressing the Issue ...... 300 Concluding Thoughts ...... 301

Tables Table No. 45. Policy Tools, Actors, and Goals of Science Policy and Technology Policy ...... 281 46. Tenets of Science and Technology Policy, 1960-84...... 285 47. Forces Affecting Science and Technology Policy Since 1960 ...... 287 48. Federal Government Policies Toward Information Technology R&D ...... 288 49. Department of Defense Funding for Basic Research by Discipline, Fiscal Years 1982, 1983, and 1984...... 296

Figures Figure No. Page 50. Federal R&D Budget Authority for Defense and Nondefense Activities ...... 295 51. Federal R&D Budget Authority for Defense Activities by Character of Work . . . . 295 Chapter 8 Information Technology R&D in the Context of U.S. Science and Technology Policy

Findings

● Information technology is one of the most this area. The advantages of centraliza- dynamic and controversial areas of U.S. tion or coordination are that it could save science and technology due to its rapid money and more effectively focus R&D pace of change, the emphasis placed on in critical areas; the possible disadvan- this technology for economic growth and tages include the establishment of a cum- for national security, and the pervasive bersome bureaucracy and the loss of agen- or “core” nature of the technology and its cy autonomy and flexibility. effects. ● The dominance of the Department of De- ● In this area there is a growing conflict be- fense in information technology R&D has tween policies that emphasize basic re- raised questions: Is military work siphon- search and policies that focus on interna- ing off too much talent from civilian ap- tional competitiveness and applications. plications? Is the military work changing These issues are prominent in information the direction of research in information technology R&D because the lines be- technology in ways that are disadvanta- tween basic and applied research are so geous for the commercial sector or for the uncertain. public? And are existing efforts to trans- ● Interest in coordinating Federal policy for fer technology from military to commer- information technology is intensifying, in cial applications adequate? Evidence cur- part because of foreign government poli- rently suggests that there are growing cies, growing costs for R&D, and grow- problems in this area. ing concern for international competitive- ● Current policies and practices toward in- ness. Although coordination of various formation technology R&D conflict with aspects of Federal policy has been de- the realities of increasing international bated for decades, it is a particularly sa- competition. The situation may call for a lient issue in information technology: many more sophisticated Government role in Government agencies are involved, but monitoring and support of industry and none devote high-level policy attention to research.

Introduction

Earlier chapters documented the rapid themselves continue to change in cost, power, changes occurring in information technology and the variety of functions they can perform. research and development. The technologies At the same time, institutional structures for

279

38-802 0 - 85 - 19 280 ● Information Technology R&D: Critical Trends and Issues

R&D are quickly evolving, international com- Government and the roles of different agen- petition is intensifying, and the technology’s cies in R&D. The purpose of this chapter is to impact on a wide range of social issues and examine these more general frameworks for problems is increasingly prominent. Because policy toward information technology R&D. information technology is pervasive, its effects The chapter is divided into two major sec- cascade through many aspects of society— tions. The first begins with some brief back- from science itself to education, business, and ground on science and technology policy in the defense–and at each point create seemingly United States, and the forces that have af- independent changes and conflicts. fected this policy as it has unfolded over the These changes are bringing increased atten- last few decades. Then, the chapter shows how tion to U.S. policy toward information tech- these broad policies and forces set the context nology R&D. In particular, Japanese and Eu- for and are closely tied to policy toward infor- ropean policies, as noted in chapter 7, have mation technology. In the second section of brought increasing demands for U.S. policy re the chapter, OTA discusses three key areas sponses. that are central to the science and technology policy issues raised by information technology Because the effects of information technol- R&D. The areas are the organization of Gov- ogy are so wide-ranging, any policy to respond ernment, the balance of military and civilian to this technology must consider not only ac- roles, and policy measures to enhance inter- tions within specific issue areas such as man- national competitiveness. power, but also broader issues in science and technology policy, such as the organization of

Part I: Background

General U.S. Science and of technologies, particularly as they relate to Technology Policies international competitiveness. In some recent discussions of industrial policy, technology Historically, science policy has been the policy has sometimes been considered an ele- term used to describe the actions of Govern- ment of, or even a synonym for, industrial ment that affect the funding, organization, policy. performance, and use of science.’ The term has included policy for technology and engineer- Table 45 sketches some of the actors and ing as well as for science. More recently, how- policy tools involved in both science and tech- ever, as technology has played a more promi- nology policy. The two types of policies have nent role in society and industry, many different, yet overlapping, constituencies and experts view “science policy” as inadequate goals. In an area such as information technol- in addressing concerns of technology.2 The ogy, where “science” and “technology” are term “technology policy” has been used in- often commingled, the boundary between sci- creasingly to refer to policy measures much ence policy and technology policy is vague. Re- more directly related to development and use cent statements of science policy (box A) illustrate the priorities of various policymakers, IScience Policy Research Division, Congressional Research and show how science and technology are of- Service, Sa”ence Policy, A Working Glossary (Washington, DC: U.S. Government Printing Office, 1976). ten mentioned together and blurred in the for- ‘See, for example, J. J. Baruch, “The Cultures of Science and mation of policy. Note that although executive Technology,” Science, Apr. 6, 1984. Baruch argues that lump- branch statements of science policy may be ing science and technology policy together is unwise, since the two enterprises have quite different approaches, goals, and the most visible, other actors in the science needs. policymaking arena-particularly Congress Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy ● 281

Table 45.—Policy Tools, Actors, and Goals of Science Policy and Technology Policy

Science Policya Technology Policya Primary policy tools: Primary policy tools Funding of basic research Mission-oriented R&D funding Scientific manpower and education measures Engineering manpower and education measures Science information dissemination Technology transfer mechanisms International exchange programs Limits on international flow of technology and information R&D tax credits Standards and patent policies University/industry research collaboration Primary Actors: Primary Actors: Office of Science and Technology Policy Office of Science and Technology Policy (The White House) National Science Foundation, National Science Foundation National Science Board National Academy of Sciences National Academy of Engineering Agencies conducting basic research, e.g., Department of Mission agencies— e.g., Departments of Defense, Energy, Defense, National Institutes of Health the National Aeronautics and Space Administration Congress Congress University science community Industrial R&D community American Association for the Advancement of Science Industry associations— e.g., Information Industry Association Professional societies—e.g., American Medical Professional societies—e.g., Institute for Electrical and Association, American Chemical Society Electronics Engineers, Association for Computing Machinery Department of Commerce—International Trade Administration, National Bureau of Standards, National Telecommunications and Information Administration Social goals: Social goals: Quality of life Economic well-being Knowledge for knowledge’s sake National security Equity, education Technological leadership a!,sCl~nC~ ~OllCY,, and ,~teChnOIOgY ~OllCY>> are often difficult to sepa~ate, The policy tools, actors, and goals listed under each category are those that tend to be associated with science policy or with technology policy However, in many practical situations, the issues and actors are intertwined. SOURCE” Office of Technology Assessment and the scientists and engineers themselves— science policy has evolved since the 1940s out have a strong (some would say dominant) in- of tension between two fundamental premises: fluence over actual policy. 1. that research should be supported in or- A brief history of U.S. science policy (which, der to push ahead the frontiers of human as noted above, has usually been defined to in- understanding (“science for the sake of clude technology policy) is helpful in order to science”), lay the groundwork for techno- provide a context for the gradual unfolding of logical advances, and train future scien- policy toward information technology.3 U.S. tists and engineers; and 3This chapter’s analysis of science and technology policy is a synthesis of published books, articles, statements and legis- (Washington, DC: GAO, Jan. 30, 1981); Congressional Budget lation; in addition, OTA and its contractor (J. F. Coates, Inc.) Office, Federal Support for R&l) and Innovation (Washington, conducted interviews with several dozen science policy experts. DC: CBO, April 1984); Science Policy Research Division, Con- OTA is indebted to this group (see acknowledgments at the gressional Research Service, National Sa”ence Board: Science front of this volume) for their insights and assistance, although Policy and Management for the National Science Foundation, OTA takes full responsibility for the content of this report. For 1988-1980, January 1983; Frank Press, “Science and Technol- a fuller elaboration of history and issues in science policy, ogy in the White House, 1977 to 1980, ” Science, Jan. 9, 1981, readers should consult, for example: Harvey Brooks,The Gov- pp. 139-145, and Jan, 16, 1981, pp. 249-256; and the annual se- ernment of Science (Cambridge, MA: MIT Press, 1968); W. ries of reports on R&D from the American Association for the Henry Larnbright, Goverm”ng Science and Technology (New Advancement of Science. For a comprehensive view of the role York: Oxford University Press, 1976); Daniel S. Greenberg, The of one key player in science policymaking, seeToward the End- Pofitics of Pure Science (New York: New American Library, less Frontier: History of the Committee on Science and Tech- 1971); A, Hunter Dupree, Science in the Federal Government nology, 1959-79, House Science and Technology Committee (Cambridge, MA: Harvard University Press, 1953); U.S. Gen- Print (Washington, DC: U.S. Government Printing Office, eral Accounting Office, Major Science and Technology Issues 1980). 282 ● Information Technology R&D: Critical Trends and Issues Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy . 283

. . From President Carter’s 1979 message to Congress on Science and Technology: Yet despite the centrality of science and technology in our lives, the Federal government has rarely articulated a science and technology policy, This message sets forth that policy. The thesis is that new technologies can aid in the solution of many of our Nation’s problems, These technologies in turn depend upon a fund of knowledge derived from bade research. The Federal government should therefore increase its support both for basic research and, where appropriate, for the application of new technologies. . . The Federal government’s support of research and development is criticaI to the overall advance of science and technology. Federal responsibility lies in three major categories: 1. The largest fraction of Federal investment serves the government’s direct needs such as defense, space, and air traffic control. . . 2. The Federal government undertakes research and development where there is a national need to accelerate the rate of development of new technologies in the private sector. . . . when the risk is great or the costs are inordinately high. , . . 3. The Federal government supports basic research to meet broad economic and social needs. . . . The majority of Federal support for basic research is in the mission agencies. From the White House Office of Science and Technology PoIicy’s 1982 Annual Science and Tech- nology Report to the Congress: The U.S. science policy is ● to enhance the contribution of science to the two most pressing long-term needs of the United States: national defense and the international competitiveness of U.S. industry. ● To maximize the return on national R&D investments; and ● To ensure the long-term vitality of the U.S. science and technology base.

The strategy to implement U.S. science pOLiCY Emphasizes excellence–in research results and in people: Stresses the importance of scientific relevance to national needs, and more clearly defines the appropriate roles of the Government and the private sector in supporting R&D; Facilitates cooperation in scientific research among Government, industry, and academia; Seeks to support sufficient basic and long-term applied research to ensure that the United States maintains the world’s strongest science and technology enterprise; Emphasizes the importance of having the leading research universities in the world and of trainin g the highest quality scientists and engineers to ensure continued U.S. qualitative leader- . . . . Ship; and Allocates Federal R&D resources to support this strategy.. - The U.S. technology policy is to ensure that U.S. scientific leadership results in economic and . defense leadership. . ~+ . ~ .* The strategy to implement U.S. technology policy’ ‘ r - Provides tax and other incentives to the private sector for commercial R&D; Continues to emphasize the different private sector and government roles in developing new technologies, products, and processes so as not to discourage private sector initiative with the threat of Government intervention and competition; - Improves the climate of cooperation so that maximum cross-stimuli occur among Government, industry, and academia; . .= Improves the ability of Federal laboratories to contribute to U.S, industry, and also takes advantage of foreign research results; Encourages the change in industry’s outlook to emphasize long-term viability rather than only short-term gain; Recognizes that the service sector in the U.S. economy is gaining in importance, and focuses emphasis on R&D accordingly; and Recognizes the effect of economic and regulatory policies on U.S. science and technology and, ultimately, on U.S. economic competitiveness. 284 ● Information Technology R&D: Critical Trends and Issues

2. that the investment of public resources The clearest landmark event in the post-war in research should be moderated and di- era was the Soviet launch of Sputnik in 1957, rected at specified high-priority national which had two major kinds of effects. The re- needs, to which the private sector is un- sponse to Sputnik-the technological venture willing or unable to respond. to put a man on the Moon, and bring him back, by the end of the decade of the 1960s–was The first principle was embodied in the cre- unique for a non-war effort in having a singu- ation of the National Science Foundation and lar clarity of mission and unequivocal criteria the National Institutes of Health to manage for success. This mission galvanized a large the distribution of public funds to support portion of the scientific and technological basic research. The primary mechanism for enterprise to a single clear goal. The second, this support is research grants, which are more diffuse consequence was to redirect the made in response to requests by recognized Nation’s attention, albeit for only a brief scientists and validated by the judgment of period, to science, science education and new their peers. scientific opportunities. The roots of the second principle, which The premises of science policy, as described underlies all “mission-oriented” Government- above, have gradually evolved into a set of rel- funded research, go far back into our national atively consistent basic tenets or assumptions history. The second principle is evident in the that guide Government’s actions. OTA de- science policy statements of box A, particu- rived the science policy statements in table 46 larly those from the Reagan administration, primarily from the practices and behavior of which emphasize the payoffs of science and U.S. policymakers and institutions over the technology for the economy and defense. past 25 years, as well as from published state- The accountability and focus indicated by ments and policies and interviews with science the second premise is often at odds with basic policy analysts. Although these principles research, which sets its own directions and have been relatively stable, they may contra- often leads investigators down blind alleys or dict one another and come into conflict in spe- toward ends that may have no immediate or cial cases, and exceptions could certainly be foreseeable practical applications. found for each item in the list. Furthermore, they have rarely been stated explicitly; in- Until the 1940s, most federally supported stead, they are embodied in a diverse collec- research was closely related to well-established tion of decisions, practices, and legislation. Government responsibilities such as defense While table 46 is in no way a complete set of or exploration and development in the West, the principles which guide U.S. science policy, or to areas basic to the national economy (agri- the essential tenets relevant to information culture, water, and public health). After World technology are included. War II, leaders such as Vannevar Bush–realizing that we had entered an era of rapid ad- Each of the science policy statements dis- vancement in scientific knowledge that could played in table 46 has varied in importance create new technologies and industries-force- and salience in driving programs, projects, and fully led the Nation to accept increased, sys- organizational relationships. Often, the proc- tematic, and continuing support for science esses by which a policy issue is resolved re- through funding of basic research and science sult in an overcorrection of some situation education. 4 which, in turn, later leads to the recognition that the pendulum has swung too far. For ex- 4 See Vannevar Bush, “Science, the Endless Frontier: A Re- ample, the advent of Sputnik was perceived port to the President on a program for Postwar Scientific Re- search, ” originally pubIished in 1945, reprinted by the National to indicate that support for basic science had Science Foundation, Washington, DC, 1960. been too weak. On the other hand, the Mans- Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy . 285

Table 46.–Tenets of Science and Technology Policy, 1960-84

Basic Research 15. Public and stakeholder participation in science and tech- 1. Basic research is a Federal mission. nology decisionmaking is appropriate, desirable, and en- 2. The best model for conduct of the basic scientific enter- couraged. prise is physical science, and in particular, physics. 3. Peer review will be the primary means for selecting topics Special Federal Roles for basic research. Management concerns will play a role 16. The Federal Government will help assure the strength of in more mission oriented research. the research system by collecting, analyzing, and dis- 4. Manpower for the scientific enterprise will be produced seminating information on subjects such as science, primarily as derivative of, and as an intimate part of, basic scientific and engineering manpower, and technological research at universities. innovation. 5. Social sciences will flourish under the traditional (physi- 17. National laboratories are general assets to the nation, well cal science) model of scientific research. Social and in- beyond the particular missions for which they were estab- terdisciplinary research are keys to the more effective ap- lished. plication of knowledge to many classes of societal problems. R&D Funding 18. The scientific community may operate on the assump- Mission Agency R&D tion that there is a firm long-term implicit commitment 6. There is a useful and significant distinction between basic to incremental funding increases. and applied research and between research, engineering, 19. To avoid disturbances in the established pattern of sup- and technological applications. These distinctions are of port for science, the identification of new problems, primary value in defining the role of Government in rela- issues, and options will be handled primarily by budget tion to the general economy and the role of Government augmentation, rather than by reprogramming of existing agencies in relation to their missions and to each other. programs. The primary instrument for effective infusion 7. Mission agencies will define their knowledge needs which of money in large quantities for new scientific enterprises may be satisfied through R&D and present their case will be the establishment of an office, a bureau, an agency, through the budget process. or a division. 8. Federal agencies are expected to undertake research in 20. In most fields, the most appropriate method of support support of the commercial, business, and private sector will be funding individual projects by individual in- insofar as support of that research will yield substantial vestigators. public benefit, especially to the clients and constituents 21. On large expensive basic science projects, the Federal of that agency. Support is encouraged only in those cases role is to provide large block funding and long-term sup- where research to satisfy nongovernmental needs is un- port. It will stand clear of the programmatic side of those likely to be adequately sustained by private initiative. activities. Defense R&D Nonfederal R&D 9. Defense research, although a major part of U.S. R&D ex- 22. Both basic and applied research in the commercial sec- penditures, will be treated as an isolated, separate case tor is best and most effectively handled by individual cor- with the expectation that side benefits will accrue to the porations and will best prosper under competition. To fa- larger scientific and industrial community. cilitate that development certain public strategies, such 10. DOD will have a restricted and limited role in support of as patents, copyrights, tax write-offs, and a variety of other basic and social research at universities. This policy, measures are appropriate for Government. American com- manifested in the Mansfield amendment under the re- mercial research requires no particular Government in- newed pressures of the Cold War, has been relaxed. tervention, attention, or assistance, since it can cope with any foreign competition. Organization of Government 23. Applied research ”applicable to the private and nonfederal 11. Voluntary coordination, rather than legislative or cen- public sector requires little attention. It will take care of tralized control and coercion, will be the primary instru- itself. ment by which programs in and among agencies will be 24. Good relationships between universities and industry are integrated, Coordination will be a primary mechanism for beneficial to both institutions, and Government will act assuring completeness of coverage of essential fields and to support such relationships, but not directly intervene. the primary instrument for reducing overlap and redun- dant budgets and programs. 12. At the Executive level, the Office of Management and Bud- Utilization of Research get will exercise its statutory role in assuring that mis- 25 The free and open dissemination of research results, ex- sion needs are met and that research and development cept those of a commercial proprietary sort or affecting programs are reasonable and realistic. There is also a role national security, is the best guarantee of the effective for a White House science policy advice mechanism. use of new knowledge in the service of the nation. 13. In the Congress, oversight, both general and budgetary, 26 With regard to basic research, technology transfer, that will be the primary technique by which quality, complete- is, the practical use of research results, is best handled ness, and fullness will be assured. by the delivery of scientific information through journals 14. Planning for science and setting the agenda for science and monographs. Commercial use best occurs through and technology are best handled by the mission agencies scientific channels, through the employment of univer- or the specific disciplines. sity scientists as consultants, and through private sec- 286 ● Information Technology R&D: Critical Trends and Issues

Table 46.—Tenents of Science and Technology Policy, 1960-84—continued

tor organizations assuming responsibility for remaining be placed upon trade and exchange where appropriate, alert to developments in their own interests. primarily at the discretion and behest of the national secu- 27. Mission agencies have the primary responsibility for get- rity establishment. ting the scentific and technical results of research to potential users within their mission areas. Technology 28. It is in the national interest to deprive Iron Curtain coun- 29. Policy toward technology is unnecessary or will be treated tries of the benefits of Western, that is American, science as an adjunct to science policy. This policy has, of course, and technology. Consequently, systematic restraints will come under some challenge in the last few years.

SOURCE: OTA analysis, interviews with science policy experts, and synthesis of published materials. See footnote 3. Note that these are statements of underlying policies, and there are exceptions and contradictions to each statement field amendment6 was a rebuke to DOD for international competitiveness. While these are obscuring the distinction between basic and largely issues of trade policy, there are strong mission-oriented research, but it also swung connections between trade and science pol- Government away from the Sputnik-induced icies, especially in information technology. changes and back toward accountability and Other aspects of policy toward information strictly defined applied research. technology R&D that stem directly from these Nevertheless, most of the principles high- general science policies include the separation lighted in table 46 have remained effective and of military and civilian research in information functional over the past quarter-century. As technology R&D, and the implicit belief that general science policies they influence research the market process will take care of the down- and development in the area of information stream social effects of information technol- technology. While many of these influences are ogy. The next section of this chapter will set subtle or indirect, the principles shown in table forth policies toward information technology 46 can be seen in the current situation. For ex- R&D in more detail. ample, the U.S. Government’s position toward Since 1960, several important trends and the global market, reflecting the propositions forces have affected both general science pol- above, is that industrial competition will deal icy and policies toward information technol- effectively with issues of international com- ogy R&D in particular. Table 47 highlights petition and no special Government policy is some of these forces. As is evident from the required. This is in sharp contrast, of course, second column of table 47, many of the forces to Japanese and European strategies, as noted affecting science policy generally have been in chapter 7, and this contrast has intensified particularly prominent in information technol- the debate about appropriate Federal roles in —. —.. .— ogy R&D. In some respects, policy toward in- ‘As described in W. C. Boesman, “U.S. Civilian and Defense formation technology R&D is the leading edge Research and Development Funding,” Congressional Research of issues in science and technology policy. This Service, Science Policy Research Division, Aug. 29, 1983, p. 23: is in large part due to the rapid pace of change Even after the establishment of the National Science Foundation, whose mission is the support of basic and in information technology, the emphasis placed applied research and education in the sciences, DOD con- on information technology for economic growth tinued to fund a significant amount of basic research un- and for national security, and the pervasive til, in 1969, the Congress passed the “Mansfield amendment” to the fiscal year 1970 military procurement or “core” nature of the technology. authorization which prohibited funds authorized by that law from being used to conduct R&D not having “a direct and apparent relationship to a specific military function or operation. ” Information Technology R&D Policies The following year, the Congress passed the “modified Mansfield amendment” to the fiscal year 1971 military Like general science and technology policies, procurement authorization which prohibited funds au- policies toward information technology R&D thorized by that “or any other Act” from being used to are not often explicit or coordinated. Instead conduct R&D unless the Secretary of Defense determines the existence “of a potential relationship to a military there have been many decisions, actions, state function or operation. ” ments, and organizations, which taken togeth- Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy . 287

Table 47.— Forces Affecting Science and Technology Policy Since 1960

Trend or force Implications for R&D in information technology 1. Growing pervasiveness of science and technology in Information technology is one of the most vivid examples society. This is accompanied by rapid blurring of of this growing pervasiveness. The intertwined nature traditional distinctions between basic and applied of basic and applied work, and of information science research, and between science and technology. and technology, raises questions about appropriate Federal roles, which have traditionally been based on those distinctions. 2. /integration of the global economy. This is Information technology is an area in which these accompanied by an increase in international challenges have become quite intense: while we still competition, a challenge to U.S. supremacy in certain lead in most areas of R&D, our lead is narrowing, and research areas, and a growing consensus that U.S. our ability to use our technological leadership in industries are not invincible and may need help. applications for economic gain is in question. The margin of error for actions in information technology R&D has been dramatically reduced because of international competition. 3. The shifting role of the Department of Defense. DOD DOD was an early and strong supporter of many areas of sponsorship of R&D was dominant in the post war information technology. 70-80 percent of Federal era, then was shifted away from basic research and funding for R&D in information technology now comes other agencies played stronger roles. Now DOD is from DOD. In certain areas (e.g., artificial intelligence, once again dominant in most areas of R&D funding, software engineering), DOD continues to be a very although the funding is much more directed than it strong influence on the directions for R&D. was after World War Il. 4. The side effects of technology. The public seems to Though there are concerns about privacy and equity have grown increasingly wary of technology, issues, use of information technology seems to be particularly in the ’60s and ‘70s. At the same time, viewed as inevitable, and, in many cases, desirable. science and technology are viewed as a way out of R&D in information technology, as the basis for our economic malaise. innovations, is viewed as essential to support an economy heavily oriented toward high technology. 5. Big budgets for R&D. Demand for accountability has The demand for accountability has just begun in grown as R&D budgets have swelled and agencies information technology R&D, particularly in major have undertaken major projects (e.g., accelerators, software projects, or use of supercomputers. weapons systems). Universities in particular are squeezed by rapidly rising costs for this research. 6. Internal upheaval in the science enterprise. The Information technology R&D is acutely affected by the decade-long search for a more effective science multiplicity of agencies and roles in setting policy. policy apparatus has bounced around government, Because of the technology’s pervasiveness, more than focusing at various times on agencies such as NSF a dozen agencies set policy for R&D in information and OSTP. None have been conspicuously effective in technology. None of them have devoted high-level a broad-scale science policy role. policy attention to information technology;—. it tends to be viewed as a tool. SOURCE OTA analysls, !ntervlews with science and technology policy experts, and synthesis of published material. See footnote 3 er comprise the de facto U.S. information tech- judgment that they are necessarily appropri- nology R&D policy. Using techniques similar ate at present— or in the future. to those used to develop table 46—i.e., analy- 1. The Federal Government has operated sis of published material and the actions of under the assumption that the United policymakers, and discussions with policy ex- States should not be dependent on foreign perts—OTA produced a list of policies specif- information technologies to ensure its na- ically related to information technology R&D, tional security. which are displayed in table 48. Each of these, elaborated below, can be seen to stem rela- A primary mission area for information tech- tively directly from one or more of the tenets nology R&D is national security. Defense outlined in table 46. Note that the following spending dominates Federal support in this are statements of the effective principles that area of R&D. The U.S. position as leader of appear to underlie Government’s actions over the Western military alliance has led to a com- the past quarter-century. As in the statements mitment to keep the United States at the fore- of general science policies, OTA has made no front of information technology developments. 288 ● /formation Technology R&D: Critical Trends and Issues

Table 48.-Federal Government Policies Toward R&D priorities—the National Weather Serv- information Technology R&D ice, U.S. Postal Service, the Department of the Treasury, the Federal Aviation Administra- 1. The Federal Government has operated under the assumption that the United States should not be dependent on tion, the Federal Reserve Board, and so on. foreign technologies to ensure its national security. This area of R&D has not received high level 2. Information technology R&D has been funded separately or coordinated policy attention. for civilian and military applications. 3. R&D priorities have been set in Government by mission 4. The Federal Government has assumed agencies, and in commercial application areas by the private sector. that it should promote continuous growth 4. The Federal Government has assumed that it should pro- and change in information technology. mote continuous innovation in information technology. 5. The market has been assumed to be the best mechanism Innovation in information technology is viewed to bring the civilian benefits of R&D in information technology to society. as overwhelmingly beneficial to society. On 6. The market has been the primary means to attend to the the military side, Government seeks continual consequences and effects of information technologies. innovation in order to keep ahead or abreast 7. Where necessary, the Government has used traditional of potential adversaries. On the civil side, the means for regulating the behavior of firms in information technology industries. contributions of information technology to 8. The short- and long-term manpower needs of information productivity argue for continued advances to technology R&D have been addressed through traditional keep the U.S. economy prosperous. On the means. 9. Government has followed industry’s lead in setting stand- other hand, the concentration on innovation ards except where Government is a dominant purchaser. tends to shift attention, especially within Fed- 10. The Federal Government has assumed that free trade eral R&D, to new and glamorous technologies policies benefit the United States in the long term. 11. U.S. Government has restricted the export of sensitive and away from improving or reducing the technical information, as well as advanced information costs of existing technologies.6 technology itself, to Eastern Bloc nations. 12. The primary international role for the U.S. Government in 5. The market has been assumed to be the information technology has been to promote equitable best mechanism to bring the civilian bene- use of common global resources. SOURCE: See text. fits of R&Din information technology to society. For the most part Government policy has 2. Information technology R&D has been assumed that the market will identify and funded separately for civilian and military meet the needs of society for information tech- applications. nology and that the market will make the ap- This policy is not unique to information tech- propriate investments in R&D to meet those nology R&D. It assumes that a useful distinc- needs. Similarly, Government policy has as- tion can be drawn between civilian and mili- sumed that industry will develop the support- tary uses of information technology. It also ing technologies and infrastructure such as assumes that there is little overlap between software quality control processes as part of the civilian and military uses in this area, and meeting the market’s needs. that where such overlap exists, as in weather Government frequently encourages innova- forecasting, the results of military R&D will tion in the private sector through indirect find their way into commercial uses. In a few measures such as procurement and tax allow- cases, there are small transfers of funds from ances. In cases where developments are impor- military to civilian agencies performing R&D. tant to the national interest, such as national 3. R&D priorities have been set in Govern- electronic mail, Government has used R&D ment by mission agencies, and in commercial application areas by the private sector. ‘See L. Thurow, “The relationship between defense-related and civilian-oriented research and development priorities, ” in Government sees information technology as Prion”ties and Effia.encyin Fderal Research and Development, a compendium of papers submitted to the Subcommittee on Pri- a tool. Therefore, information technology R&D orities and Economy in Government of the Joint Economic is decentralized. Each agency sets its own Committee of the Congress, Oct. 29, 1976. Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy ● 289 contracts to promote innovation while limiting necessary investments in equipment to pro- the risks to industry. vide the appropriate training for these future information scientists and engineers. 6. The market has been the primary means to attend to the consequences and effects 9. Government has followed industry’s lead of information technologies. in setting standards except where Govern- ment is a dominant purchaser. This policy assumes positive impacts will result in new markets, while producers will re- Government treats information technology duce adverse impacts in order to remove im- like any other industrial product in terms of pediments to present and future applications. standards, relying mostly on voluntary indus- Protection of individual rights is the major ex- try standards. When Government does get inception to the reliance on the market. The Con- volved in standards-setting, it is usually at the gress, the executive branch, and the courts request of industry. In the computer field, the have all attempted to cope with the issues of Institute for Computer Sciences and Technol- individual privacy, intellectual property, and ogy (ICST) at the National Bureau of Stand- freedom of access. For the most part, the as- ards is responsible for developing standards sumption underlying their deliberations and for the Federal Government, and it also par- actions is that traditional legal, regulatory, or ticipates in and coordinates a variety of indus- organizational mechanisms can handle these try standards efforts. In certain cases such as issues. computer networking standards, ICST has taken a firm leadership role, both domestically 7. Where necessary, the Government has 7 and internationally. used traditional means for regulating the behavior of firms in information technol- 10. The Federal Government has assumed ogy industries. that free trade policies benefit the United States in the long term. Historically, the Government seems to have assumed that there is nothing special about From transistor radios to microchips, the U.S. information technology industries. Govern- Government has maintained a position of free ment has not seen antitrust, patent, tax and trade in the area of information technology. other regulatory policies as major impedi- The assumptions underlying this policy have ments to innovation. For the past half-century, been: 1) free trade will open up foreign mar- Government has assumed that: 1) regulated kets for U.S. products, 2) the U.S. lead in in- monopolies such as AT&T are effective per- formation technology is largely unassailable, formers of R&D; and 2) developments in reg- 3) the marginal benefits of lower costs to the ulated and unregulated areas of telecommu- consumer outweigh the threat to U.S. indus- nications and computers are not in conflict. try, and 4) competition promotes innovation. 8. The short- and long-term manpower needs This openness has included access to R&D of information technology R&D have been through the published literature, licensing, addressed through traditional means. joint ventures, and other commercial routes. As noted in chapter 5, the Government has 11. The U.S. Government has restricted the relied on the universities to meet the needs of export of sensitive technical information, the market for the trained scientists and engi- as well as advanced information technol- neers necessary for innovation in information ogy itself, to Eastern bloc nations. technology. Support of research in information The importance of information technologies to technology at universities is the primary meth- U.S. national security has led to Government od by which the Federal Government supports actions to preserve the superiority of U.S. in- manpower development in the field. Govern- ment has also assumed that the universities, 7J. H. Young, “Effects of Standards on Information Tech- assisted by various subsidies, will make the nology R& D,” paper prepared for OTA, Nov. 25, 1983. 290 ● Information Technology R&D: Critical Trends and Issues formation technology. One way is to prevent 12. The primary international role for the U.S. its adversaries from getting access to, for ex- Government in information technology ample, certain research publications, advanced has been to promote equitable use of com- chip designs or cryptography software. This mon global resources. policy assumes that: 1) the Government can U.S. actions regarding international use of in- effectively control the international operations formation technologies have focused on issues of U.S. researchers and firms, 2) these controls such as spectrum allocation, the use of the geo- will not unduly harm the viability of the U.S. synchronous orbit, the international use of information technology industry, and 3) the communications satellites, and access to data advantage of such controls outweighs harm from weather and other Earth applications done to the U.S. R&D enterprise through re- satellites. striction of information flow among researchers.

Part II: Key Issue Areas

The previous section provided background ment to deal with information technology on the nature of policies related to information R&D. The second issue area is the balance of technology R&D, and on the connections be- civilian and military funding in this area of tween those policies and broader science and R&D. The final area is international com- technology policies. A central conclusion is petitiveness. that information technology R&D has been influenced to a substantial degree by policies ISSUE A: Organization of Government applicable across many areas of science and Demands for coherent Federal policy to- technology. However, factors such as the rewards information technology R&D conflict liance on information technology for economic with the traditional system of pluralist deci- growth and national security, and the perva- sionmaking by various agencies and the pri- sive, core nature of the technology, are increas- vate sector. ingly stressing policy toward information technology R&D, focusing attention upon it, and Introduction setting it apart from policy for other areas of science and technology. The search for coherence and effectiveness in science policy of all kinds has been the ob- This section attempts to build on that foun- ject of many commissions, proposals, legisla- dation by examining three particular areas of tive initiatives, and reorganization plans. A policy toward information technology R&D hundred years ago the first plan for develop- that may be ripe for change or improvement. ing a Department of Science and Technology OTA selected the three issue areas because was introduced in the Congress.8 Several times they are key problems for the future develop- in the 1970s and again within the last year, ment of information technology R&D. Through- the idea of such a central department has re- out this report, scores of issues have been iden- emerged.9 The fundamental issue coming out tified which, in themselves, merit attention. of all such proposals is whether science is bet- The three issues discussed in this chapter are ter managed through a central organization overarching, in that they subsume many of the —— earlier issues. Addressing these three areas ‘For a discussion see A. H. Dupree, Science in the Federal could help set the direction for many of the Government, Harvard University Press, 1953, p. xi. ‘See, for example, H.R. 481, The National Technology Foun- more detailed issues. The first, and most fun- dation Act, introduced Jan. 6, 1983, by Rep. George Brown, damental, topic is the organization of Govern- et al. Ch. 8—Information Technology R&D in the Context of U.S. Science and Technology Policy . 291 or as an adjunct to the missions of the Fed- and technology areas, ranging from complete eral Government as reflected by the mission decentralization and pluralism to a central agencies. agency which handles the bulk of R&D funding and science policymaking. Despite the ap- While the proposals for a central Depart- ment of Science and Technology have con- peal of coordination in principle, it has costs that include decreased flexibility of mission sistently failed, the Nation has reorganized its agencies, creation of cumbersome bureaucra- science and technology apparatus several cies, and potential loss of multiple funding times to meet new and emerging needs. The avenues—and hence multiple approaches—for Department of Energy, the Environmental Protection Agency (EPA), and the National researchers. One report notes: Oceanic and Atmospheric Administration Coordination is like motherhood; everyone (NOAA) all resulted from the coalescence or agrees it must be done but it lacks an opera- rearrangement of diverse scientific and tech- tional definition. nical functions. Coordination is not a homogeneous activity; but rather an umbrella which encompasses Nevertheless, the traditional Government many different activities performed by dif- organization for science in general is decen- ferent people, for similar effect. tralized, pluralistic, and only loosely coordi- Coordination requires significant effort at nated. The strong centralizing force on science all levels of management and, therefore, both comes from the annual budget review by the horizontal and vertical structures need to be Office of Management and Budget. Even considered. there, concern for scientific issues is for the A certain amount of coordination is good for the health of Government, but like exer- most part split up along agency lines; for ex- 10 cise, too much will cripple or kill. ample, NASA is in one area, NOAA in another, and Defense science in still another. Consequently, most science policy experts ar- This pluralistic system of Federal policymak- gue that some combination of centralized deci- ing has the advantages of allowing mission sionmaking, ad hoc coordination, and mission agencies to tailor R&D to their own needs, autonomy is appropriate.ll which is a basic tenet of science policy as discussed earlier. However, several trends are Dimensions of the Issue putting stress upon the ability of the current decentralized system to cope with new and Current responsibilities for information emerging problems: technology are dispersed all over Govern- ment—from the Department of Defense to the ● Increasing international competition, and General Services Administration, from the De- the presence of coordinated technology partment of Justice to the Federal Communi- policies among our trading partners, are cations Commission. The ad hoc nature of pol- highlighting our absence of coordination icy in this area is even more evident than most and causing many to call for reexamina- other types of science policy because agencies tion and change; and tend to see information technology as a tool, ● The costs of R&D of all kinds have risen. not as something warranting significant pol- At the same time, there is increased pres- icy attention in itself. In addition, there are sure on the Federal budget from entitle- simply more agencies involved because of the ments, defense, and the deficit. Some ar- pervasive nature of information technology. gue that a more coordinated and coherent Federal science and technology apparatus could be more cost-effective and ac- 10W. A. H~n, D. S. Alberts, and J. Lovelace, “Interagency Coordination: Workshop Report,” in The Management of Fed- countable. erai Research and Development: An Analysis of Major Issues and Processes (McLean, VA: The Mitre Corp., 1977), pp. 93-97. There is a broad spectrum of possibilities for *’See, for example, Harvey Brooks, The Government of coordination of Federal activities in science Science (Cambridge, MA: MIT Press, 1968). 292 ● Information Technology R&D: Critical Trends and Issues

More than a dozen agencies fund or affect rele- certain very specific areas. The lack of vant R&D.12 reliable assessments of manpower and the associated uncertainties hinder policy- There is, however, some coordination in Fed- making in all areas of the Government eral policy. To the extent that DOD dominates that work with information technology. R&D funding, it is the de facto lead agency ● The Federal Government has an exten- and informal or formal coordination point. sive network of national laboratories, al- And agencies often coordinate their work on though the quality and relevance of some an ad hoc basis. For example, the Defense Ad- of these facilities has periodically been in vanced Research Projects Agency (DARPA) 15 question. Researchers at various na- and NSF brief each other on computer science tional laboratories constitute the largest research programs. 13 And ICST at the Nation- concentration of expertise in use of super- al Bureau of Standards has a group of senior computers. The question of how best to officials from major agencies who advise the use the national labs in this and other Institute on its programs. fields of information technology R&D Several specific factors focus attention on cuts across a variety of agencies, in par- the degree of coordination and coherence in ticular the Departments of Defense and policy toward information technology R&D.14 Energy. Among them: ● Related to use of the national labs, chapter 3 pointed out that researchers are in- In general, there is a lack of high-level pol- creasingly requiring advanced computers icy commitment in this area, which has or “supercomputers” to perform a wide several kinds of effects. One is that the variety of research. The question of where role of mission agencies in information to house such machines and how to pro- technology is shifting, uncertain and as vide access is of concern to a wide vari- divergent as the roles of those agencies ety of agencies involved with information themselves. Coordination would be useful technology R&D. Committees of the Fed- in such subjects as database collection, eral Coordinating Council for Science, R&D research topics, and compatibility Engineering and Technology (FCCSET) of technology and information. at OSTP have attempted to address this Many have argued that there are substan- issue. tial shortages of manpower. However, as ● As discussed in chapters 2 and 3, the discussed in chapter 5, the evidence for emerging shared wisdom in the industry such shortages is inconclusive, except in is that software is a key problem in cost ‘These include the Department of Defense (itself divided into and effectiveness of computing systems. the Defense Advanced Research Projects Agency, the Office Many believe that American industry has of Naval Research, the Air Force Office of Scientific Research and various other units in the Pentagon and the three services), failed to give adequate research attention the National Science Foundation, the Department of Commerce to software problems, preferring for a va- (largely through the National Bureau of Standards, the National riety of reasons to emphasize hardware. Oceanic and Atmospheric Administration, and the National Telecommunications and Information Administration), the Na- The reliability and maintainability of soft- tional Institutes of Health, the Department of Energy, the Na- ware will become an increasingly large tional Aeronautics and Space Administration, the Federal Avia- issue, raising questions of quality control, tion Administration, the Patent and Trademark Office, and the Department of Justice (both in their jurisdiction over antitrust, standards, manpower, and education pol- and in R&D for law enforcement systems). icy for many agencies in the Federal Gov- ‘SElias Schutzman, National Science Foundation, personal ernment with large information systems. communication, Mar. 2, 1984. 14For fither elaboration of some of these factors ~d discussion of institutional options, see “Institutional Options for Ad- dressing Information Policy Issues: A preliminary Framework “See the Report of the Wlu”te House Su”ence Council Federal for Analyzing the Choices, ” a staff memorandum prepared by Laboratory Review Panel, May 1983, sponsored by the Office the Communication and Information Technologies Program of of Science and Technology Policy. (Also called “The Packard OTA, NOV. 29, 1983. Report, ” after its chairman, David Packard.) Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy . 293

● In the area of regulation, forcing new Option 1: Maintain the Status Quo. The technology into old categories is nearly strongest argument for no major change in universal because the new is not always Federal activity is that the technologies and seen as new, or the new is seen as a prob- their effects are highly fluid, and it maybe too lem, not an opportunity. For example, early to devise appropriate policy or Govern- cable television throughout the country ment organizations. In addition, some may is being treated as a local utility occupy- view coherent, coordinated Federal policy ing some status resembling, perhaps, elec- toward information technology R&D as unnec- tric power. Each local government treats essary or infeasible. The ad hoc coordination what could be an integrated national in- currently used has been relatively effective, formation utility on a short-term and and attempting more formal coordination or somewhat parochial basis. In sharp con- more elaborate national policy could be cumber- trast, Canada and France have adopted some. In addition, pluralistic research fund- policies toward cable that aim to develop ing has the advantage of funding more than a national utility. one approach to a research topic or problem. ● Uncertainties about funding levels for in- The disadvantage of maintaining the status formation technology R&D contrast with quo is that we may reduce opportunities to en- the needs for stability in budgets or sup- hance our competitiveness and to use our port as a base for long-range research in R&D resources in a more socially productive universities and industry. Examples of manner. such uncertainty include some recent va- cillations in funding of certain informa- Option 2: Improve Monitoring and Coordina- tion technology areas by DOD agencies, tion. A first step toward coordination of Fed- particularly DARPA; and the current eral roles would be to provide new mechanisms uncertainties concerning supercomputer for the various agencies involved in this area research support between DARPA and to communicate in a systematic way. Such co- NSF. (See ch. 3). ordination mechanisms would at least raise ● Information technology industries are the level of attention to information technol- now combining technologies developed ogy R&D issues and provide a forum which under regulation (as in radio, telephone, could facilitate a common understanding of and television) with computer technolo- areas of strength and weakness in Federal sup- gies basically developed in the market port. Congress could designate a formal coor- system. The convergence of these two dination group with representatives from ap- types of technologies creates new regula- propriate agencies involved in information tory issues dealing with ownership, pub- technology R&D. In fact, the first priority of lic versus private control, privacy, and a coordination group could be a report to the access. Congress, and subsequent hearings, on those areas of strength and weakness. Though DOD Options for Addressing the Issue would be a major player in such a coordination effort, it is important that it not domi- As a core technology, information technol- nate; the status, needs, and objectives of the ogy is used by everyone but is not clearly the civil sector should have an adequate platform. responsibility of anyone. Yet, given its value to the balance of trade and productivity of Other coordination and monitoring steps U.S. industry, the demands for new, more may also be desirable. To the extent that coherent action have become increasingly States play a stronger role in promoting in- strong.l6 formation technology R&D centers, and in using information technology for delivery of services, it may be useful to establish mecha- “%, for example, Science Policy Research Division, Congres- sional Research Service, The Information Science and Technol- nisms whereby States and the Federal Gov- ogy Act of 1981, June 1982. ernment can cooperate in setting priorities for —

294 ● Information Technology R&D: Critical Trends and Issues information technology R&D. Such mecha- Option 4: Establish a New Federal Organiza- nisms could include a national conference on tion. Congress could create a new organization, the intergovernmental research needs in infor- transferring to it much of the current dis- mation technology R&D, hearings on State persed responsibility for information technol- and local information technology R&D needs, ogy and adding new functions. These new and commissioning studies of the needs of functions could include compiling and inter- State and local governments for improved in- preting information on Federal procurement formation technology. of information technology, civilian vs. military priorities in R&D, regulatory actions with di- Option 3: Set New National Policy. A more rect or indirect effects on the technology, the comprehensive alternative is to make a high- U.S. position in domestic and international level policy commitment to information tech- markets, social impacts of information technology R&D. This could be accomplished by nology, high priority issues to be resolved, and reestablishing an office such as the Office for recommendations for congressional action. Telecommunications Policy in the Executive The advantage of such a new organization Office of the President, or elevating the re- would be that it would assure that the tech- sponsibilities, status and visibility of the Na- nology would be visible and explicitly ad- tional Telecommunications and Information dressed; on the other hand, it could diminish Administration and the Institute for Com- the effectiveness of other organizations that puter Science and Technology (ICST) in the pursue information technology R&D as part Department of Commerce. To complement of their mission. A new organization could be this action, Congress could establish a lead part of a new “National Technology Founda- agency for information technology R&D poli- tion, ” or it could be a freestanding “Institute cy that could devote a substantial amount of of Information and Communication. ” high-level attention to the issue. However, the establishment of a lead agency has the disadvantage that that agency’s mission may be ISSUE B: Military/Civilian Balance 17 pursued at the expense of others. Note that Relying primarily on DOD for funding of in the last several years Congress has been information technology R&D may conflict considering various proposals for centralized with the pressing demands of international oversight bodies for information technology competitiveness and productivity. policy-la Con=ess may also wish to consider restructuring the basic oversight mechanisms for information technology R&D in the Con- Introduction gress and/or the executive branch. The Department of Defense (DOD) has been Finally, it maybe an opportune time to take by far the largest supporter of information action on several more detailed issues. The one technology R&D among Federal agencies. that is most prominent is software, as dis- With increasing budgets for R&D in the re- cussed in the chapter 3 case study of software cent past, DOD is sponsoring a higher propor- engineering. Federal standards for supporting, tion of many fields of R&D activity. The dom- using, testing, updating and documenting inance of DOD in information technology is software could add much reliability to Gov- perhaps the most striking of all, however; esti- ernment information systems and consistency mates of the proportion of DOD funding range from 70 to 80 percent or more of all Federal to relations between the Government and in- 19 dustry. One mechanism for dealing with these funding. In some parts of the field, DOD has issues is to work through ICST. sponsored pioneering work which established

‘These estimates are based on W. C. Boesman, “U.S. Civil- “See Brooks, op. cit. ian and Defense Research and Development Funding, ” Science ‘Wee “Institutional Options for Addressing Information Pol- Policy Research Division, Congressional Research Service, Aug. icy Issues, ” op. cit. 29, 1983. Also see ch. 2 for further discussion. Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy . 295 foundations for both commercial and military Figure 50.—Federal R&D Budget Authority for applications. However, with tightening budg- Defense and Nondefense Activities ets and growing concern over international competitiveness, the wisdom of DOD’s continued dominance of information technology R&D funding is coming into question. Specif- ars ically, three questions have surfaced in the course of OTA’s study: 1. Is the military work siphoning off too much talent from civilian applications? 2. Is the military work changing the direction of research in information technology in ways that are disadvantageous for the commercial sector or for the public? 3. Are existing efforts to transfer technology from military to commercial applications adequate?

Dimensions of the Issue 1975 1977 1979 1981 1983 1985 DOD and civilian agency funding of R&D Fiscal year (est.) SOURCE: Division of Science Resources Studies, National Science Foundation have varied in relative emphasis and roles over “Federal R&D Funding: The 1975-85 Decade, ” March 1984 the past decades. The issue of DOD vs. civil funding of R&D has received little emphasis since the late 1960s, when DOD R&D was Figure 51 .—Federal R&D Budget Authority for drastically reduced because of a perception Defense Activities by Character of Work that the agency had overstepped its mandate, and because of social concerns about the DOD budget. As shown in figure 50, in the past decade (and particularly during the Reagan administration) DOD funding for R&D of all kinds has risen dramatically faster than civilian agency funding, which has actually dropped in real terms. It can be misleading to use the combined term, R&D, in this discussion; as figure 51 shows, for all fields combined, the 6 dramatic increase has been almost exclusively in development, rather than in basic or applied c research. 3 Applied research More specifically, DOD support for work in

L information technology, particularly through 0 the Defense Advanced Research Projects Basic research Agency (DARPA), has remained strong and 1975 ’77 ’79 ’81 ’83 ’85 1975 ’77 ’79 ’81 ’83 has grown dramatically. As shown in table 49, Fiscal year (est.) Fiscal year (est.) SOURCE: Division of Science Resources Studies, National Science Foundation, support for basic research in mathematics and “Federal R&D Funding: The 1975-85 Decade,” March 1984

38-802 0 - 85 - 20 —.”

—

296 ● Information Technology R&D: Critical Trends and Issues

Table 49.—Department of Defense Funding for Basic Research by Discipline, Fiscal Years 1982, 1983, and 1984 (budget authority In millions)

Fiscal year Fiscal year Fiscal year Fiscal year 1982 1983 1984 1985 Physics, radiation science, astronomy and astrophysics ...... 76.0 80.9 87.2 96.6 Mechanics, aeronautics, and energy conversion ...... 73.1 79.5 86.3 92.2 Materials ...... 71.5 81.0 82.8 87.5 Electronics ...... 90.0 90.5 97.9 93.7 Oceanography...... 51.1 50.2 53.4 57.5 Biology and medical sciences ...... 64.9 66.3 79.8 86.7 Chemistry ...... 53.1 58.9 62.0 66.3 Mathematics and computer sciences. . . . . 83.3 98.8 111.7 124.9 Terrestrial sciences, geophysical research ...... 24.3 29.0 30.8 33.9 Atmospheric sciences ...... 20.8 21.8 25.0 28.2 Behavioral sciences, human resources . . . 33.9 33.6 35.2 36.3 Special studies ...... — 2.0 — University instrumentation ...... — 30.2 30.0 300- In-house laboratory independent research , ...... 54.1 57.5 58.4 67.0 Total ...... 696.1 780.0 839.3 899.9 SOURCE: Leo Young, Department of Defense, presentation to AAAS Colloquium on R&D Policy, Mar. 29, 1984. computer science grew from $83.3 million in ation desperately calls for a civilian balance fiscal year 1982 to a planned $124.9 million to DOD’s funding. Despite these strong warn- in fiscal year 1985. In applied research and de- ings, however, there is inconclusive evidence velopment, several major projects at DOD that these negative results of DOD’s funding have pumped many hundreds of millions of are occurring. dollars into information technology. These projects include Very High Speed Integrated For example, in artificial intelligence (AI) Circuits (VHSIC); Command, Communica- the pool of researchers is very small and tions, Control, and Intelligence (C31); and more almost all receive DOD funding. As noted in recently the Strategic Defense Initiative (SDI) chapter 3, DARPA and ONR have been almost or “Star Wars” program, and the Strategic the exclusive funders of artificial intelligence Computing program. from the start. Yet, some AI researchers noted during OTA’s case study that relatively basic Science policy experts interviewed by OTA research which could lead to nonmilitary ap- were almost universally concerned about this plications-such as intelligent libraries-is be- resurgence of DOD funding for R&D, and for ing neglected. The assumption that AI R&D information technology R&D in particular. funded by DOD is equally applicable to both Comparing the current situation to the post- military and civilian applications is, therefore, War era when DOD research funding was also under question, although more than anecdotal dominant, they point out that current research evidence is needed to assess the problem. is generally much more mission-oriented and, consequently, less productive for nonmilitary Other controversial topics for the science uses. Some argue that we are endangering our community in general are export and publica- international competitiveness in the long term tion restrictions on scientific and technical in- by monopolizing the information technology formation. The dominance of DOD funding of R&D community with defense-related proj- information technology R&D raises the dan- ects. Others point out that it is unwise to have ger that the research will be classified too early a monolithic source of funding for any area— to allow nonmilitary users to benefit. This dan- e.g., certain technical approaches may tend to ger is particularly prominent in large scale de- be ignored —and argue that the current situ- fense initiatives such as the “Strategic Com- Ch. 8—Information Technology R&D in the Context of U.S. Science and Technology Policy . 297 puting” program. For academic scientists in Options for Addressing the Issue particular, free information flow is viewed as essential to productivity and to the ethos of Option 1: Maintain the Status Quo. While it science as an international enterprise. Hence, may be desirable to have a stronger civilian the tension has produced some strong rhetor- government presence in information technol- ic. A university association president recent- ogy R&D, some would argue that our national ly told a gathering of the American Associa- security requirements mandate the current tion for the Advancement of Science: “These level of DOD involvement, and that it is im- people feel that any delay or inconvenience for practical or unnecessary for civilian agencies the Russians is worth whatever it costs to to have a balancing involvement. us . . . What we are seeing now is not disagree Potential negative consequences of main- ments among reasonable people; it is ideolo- taining the status quo are that we may be com- gy without restraint, and it is dangerous to promising international competitiveness, and 2 us all. ” ° hence national security, in the long term. The Again, however, the extent to which restric- concerns surfacing about DOD’s funding of tions on information flow have actually been R&D may be early signals of a serious prob- onerous or counter-productive has not been lem, or they could be insubstantial worries. carefully examined. More broadly, except for Currently we do not have reliable information science policy analysts, most people—at uni- to tell the difference. versities, in Congress, or in associations or re- Option 2: Increase Monitoring and Analysis. search groups-have not raised DOD funding The clearest need in addressing the impact of as an issue. They maybe comfortable with the DOD priorities is that it be explicitly addressed current situation, or they may be uncomfort- and more monitoring and analysis be done. able alienating a powerful source of funding. Specific topics in need of monitoring and anal- Part of the reason for infrequent question- ysis include: ing of DOD’s dominance in this area is that ● Effectiveness and effects of national secu- defense applications for computer-related de- rity restrictions on access to information vices are fascinating problems. One computer technology research and devices—espe- columnist noted that state-of-the-art equip- cially the exchange of ideas among lead- ment, challenging problems, and the mystique ing researchers and the ability to use for- of “secrecy” are powerful lures for computer eign graduate assistants on DOD-related scientists. 21 —. — projects. ‘“R. Rose~weig, president, Association of American Univer- ● Effects of DOD support on the research sities, address to American Association for the Advancement priorities of leading researchers in the of Science colloquium on R&D policy, Washington, DC, Mar. 30, 1984. This comment was made when the Department of De field. fense was considering placing restrictions on the publication ● Transferability of information technology of “unclassified but sensitive’ research. At the time, the developed for DOD–the ease of transfer, presidents of Stanford, California Institute of Technology, and Massachusetts Institute of Technology “warned the Reagan the time lag—compared with primarily Administration that their institutions may be forced to stop commercial development. conducting unclassified research for the Pentagon if they are ● Use of limited manpower in certain fields required to give military reviewers the right to restrict publication of some findings. ” (Kim McDonald, “3 Universities such as artificial intelligence and software Warn Pentagon on Censorship,” The Chronicle of Higher Edu- engineering. cation, Apr. 4, 1984, p. 1.) In part in response to this outcry, ● Relation of DOD’s requirements for infor- the Pentagon rescinded its plan for restrictions. (“White House Decides to Cool Campus Secrecy Issue, ” Science and Govern- mation technology R&D to commercial ment Report, June 15, 1984, p. 1), See also Albert H, Teich and requirements, and more broadly, the Jill P. Weinberg, American Association for the Advancement tradeoffs between national security and of Science, “Issues in Scientific andTechnicaJ Information Pol- icy, prepared for Office of Special Projects, National Science international competitiveness in this area. Foundation, Dec. 28, 1982. 2*D. Clapp, “While Japan Builds Computers, We’re Making One factor working against explicit consid- Missiles, ” Infoworld, June 27, 1983. eration of military vs. civilian priorities in 298 • Information Technology R&D: Critical Trends and Issues

R&D is the fact that Congress is ill-equipped way similar to that of the Apollo program— to balance the funding of research among dif- it could be a 5- or 10-year effort toward spe- ferent agencies because the agencies’ budgets cific objectives such as uses of computers for are independent and handled by different con- education, to aid the handicapped or poor, or gressional committees. The executive branch other social goals. is theoretically capable of such considerations, but in practice, as noted above, it also handles ISSUE C: International Competitiveness agencies’ budgets as discrete units. Hence, there may be a need for a new mechanism to U.S. policies and practices are based on an assumption of unassailable U.S. dominance weigh R&D goals in information technology in information technology R&D, which is in- from a multiagency perspective. Such mech- creasingly inaccurate. anisms could include joint congressional hearings, activities of the interagency coordination Introduction body discussed in Issue A, and/or a joint study by DOD and a prominent civilian group such Information technology is an important ele- as the National Science Board on the relation- ment of global high-technology trade. As dis- ship of DOD spending to R&D priorities. To cussed in chapter 7, the efforts of the French, be most effective, such a study should prob- Japanese, and other governments to target in- ably be tied to subsequent congressional hear- formation technologies as tickets to future in- ings on the issue. ternational prosperity attest to that fact. Since Option 2 does not preclude either of the the advent of information technologies, the other options, and hence may be a wise course United States has had the lead in development of action in any case. and in global market share. That situation is changing as other advanced nations are in- Option 3: Bolster Civilian R&D funding. Con- creasing their patents in international com- gress could act to provide a stronger civilian merce, as the U.S. balance of trade in infor- balance to DOD’s information technology mation technology begins to weaken, and as R&D funding, on the basis of the suggestive the industry becomes more global in charac- evidence of problems, or on the assumption ter and thus less amenable to traditional meth- that domination of information technology ods of governmental control. R&D by one mission agency is unwise. Though such a move would require budgetary in- Dimensions of the Issue creases of several million dollars, many industry and policy experts suggest that the ul- Stresses of global integration and foreign timate payoffs in innovation and productivity competition on U.S. information technology would be substantial. R&D policies emerge in several areas. One is the effect on policies promoting development Such funding may go beyond some policy- of the R&D base of U.S. industry-manpower, makers’ notion of appropriate roles for Gov- facilities, R&D information and R&D behav- ernment. There does seem to be room, how- ior. The issues in this area include: ever, for more civilian agency funding of “fundamental” (in the sense of being widely ● Foreign versus domestic high-tech man- applicable and long-term) if not “basic” (in the power. A large percentage of the graduate sense of being disinterested in applications) re- students in science and engineering are search in information technology. The fund- foreign nationals. It has been a matter of ing agency involved would have to be careful significant national pride that the world that the research community had sufficient comes to the United States for training manpower to absorb such funds. Some experts in science; on the other hand, some have have called for a civilian research effort that argued that we are investing resources in would mobilize the research community in a these foreign students which those that Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy ● 299

leave then take away to their home coun- subsidies and generic research input from try. The value we have placed on science their governments. as an international enterprise is under ● The openness of U.S. markets to foreign stress as international competition inten- competition is not met by symmetrical sifies. U.S. access to foreign markets. U.S. man- ● A related issue is that in those areas that ufacturers still primarily focus on the U.S. involve national security or commercial market-the world’s largest for informa- secrecy, there are increasing pressures to tion technology. These firms may not be restrict access of foreign scientists and giving adequate attention to developing graduate students. Such restrictions could nations’ markets, leaving them largely to be a major dislocation in the ethos of the other nations. For example, the Japanese university. are now vigorously pursuing countertrade ● Restricting the access of foreign scientists with the Chinese to exchange mineral re- to U.S. research may isolate the U.S. resources for Japanese high technology. search community. Such isolation could ● One element of the dominance of DOD in reduce the infusion of new approaches and information technology R&D is the poten- ideas into the R&D process, and thus hin- tial diversion of talent and resources away der R&D in the United States. from nonmilitary science and technology. ● Our traditional linguistic chauvinism con- The rigid specifications and limited appli- flicts with a recognition of the need to cability of many DOD-sponsored technol- translate literature from other countries ogies could skew the development of U.S. in this area. Little of the Japanese tech- information technologies away from those nical literature, and only somewhat more products that are of most use in foreign European literature, is routinely available markets-especially markets in develop- in English translations. ing nations. ● Internationalization of the information ● There is a conflict between the need for technology industry is leading to the glo- free trade and the need to protect sensi- balization of R&D. Countries such as tive science and technology. The issue is Great Britain, the United States, France, whether U.S. export controls on informa- and Italy are competing for the location tion technology are unnecessarily ex- of research centers of the major multina- cluding U.S. companies from effectively tional firms in information technology. competing for large foreign markets. Scotland’s Silicon Glen is an example. In ● With the development of foreign markets addition to the competition for multina- for information technology, concerns arise tional R&D facilities, there is growing in- over the access of small and medium-sized terest in joint ventures among firms in ad- firms to foreign markets. Some feel that vanced nations such as Japan, Germany, where U.S. firms have penetrated foreign the United States and France. markets, the larger firms have dominated trade, to the exclusion of the smaller Issues also emerge concerning U.S. policies firms. This is not unusual, since a large to maintain or improve the current U.S. leader- majority of manufacturing exports come ship in information technology R&D and mar- from large firms. However, with the glo- keting. These include issues related to the bal integration of the industry, the United structure of the industry, the role of DOD, and States may wish to encourage smaller, in- the effects of regulation on the industry: novative firms to seek out foreign trade. ● The individual American corporation in ● Promotion of a rational world system for the information technology market may managing the use of information technol- confront foreign government-coordinated, ogies is also important to the long-term sustained, and supported consortia or leadership of the United States. Though consortia of private companies enjoying this has long been an area of recognized 300 • Information Technology R&D: Critical Trends and Issues

importance for U.S. policy, a number of debate have left the Nation long on conjecture important global standardization issues and short on facts. remain. For example, an increasing issue Option 2: Monitor and Support International will be global compatibility of communi- Trade in Information Technology, and Related cations systems. The fact that foreign Efforts in R&D. Various measures have been systems are generally run by national proposed for the support of international governments, while the U.S. system is a trade, and it is beyond the scope of this report market-based system, tends to put the to discuss them in detail.22 A key aspect of United States in a different, often disad- trade support is ensuring that foreign markets vantageous, position from all the other are open to U.S. industry, and helping U.S. contenders in international negotiations. companies to actively seek developing markets for the technology. This could involve Options for Addressing the Issue increasing the commitment and attention of The central issue facing information tech- the U.S. Special Trade Representative, the In- nology industries is how best to enhance their ternational Trade Administration, and the Foreign Commercial Service to the needs of ability to compete on equal footing with companies from other nations, especially where information technology firms. those companies are strongly supported by Options more specifically related to R&D in- government. This problem affects many other clude promotion of generic information tech- industries-from steel to shoes. Continuing de- nology R&D centers in the United States, and bate over the need for an industrial policy close monitoring and evaluation of alternative flows directly from this issue. As noted in institutional models-both domestic and for- chapter 7, the essential question is not how to eign-for cooperative research in information imitate the policy strategies of Japan or other technology. countries which appear successful, but to come Further, support for international competi- up with a response that could build on unique U.S. strengths. tiveness in R&D could include establishing mechanisms to monitor foreign technical lit- Option 1: Maintain the Status Quo. Some erature and disseminate translations to Amer- would argue that the Federal Government is ican scientists and technologists. Such sup- ill-equipped to become more involved in a fast- port could also provide funding for our research paced area such as information technology. personnel to travel overseas for conferences The current scheme of activities to promote and consultations, and for American students international competitiveness works well in or professors to study overseas. Congress may some respects; the prime role of the Govern- wish to beef up scientific bilateral agreements ment should be to provide a healthy macro- and exchange programs. economic business climate. In addition, it would be appropriate to The disadvantage of the status quo is the analyze: increasing evidence that the traditional pattern of policies related to international com- ● The amount of foreign purchasing by the Bell Operating Companies that were petitiveness does not allow our industries to formerly part of AT&T. As a vast mar- compete on “a level playing field” with com- ket for information technologies, the be- panies from other countries. At present, there havior of these companies will be critical is little basis for deciding among options for to the future of the U.S. industry. dealing with foreign competition and its effects on information technology R&D. The rel- —.—— ative newness of the threat and the rapid ideo- 22See the recent OTA report, International Competitiveness logical polarization of the industrial policy in Ek+ctroru”cs. Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy ● 301

● The extent and type of foreign ownership information technology in U.S. trade, and to of U.S. information technology firms and continue the debate on how the United States the effects of such ownership on the trans- might restructure its trade policies to respond fer of information science and technology. to those of Japan and other nations. In addi- ● The career paths of foreign computer and tion, the United States could assist the inter- electrical engineering graduate students. national competitiveness of U.S. firms by de- ● The extent to which major U.S. informa- veloping a national position in international tion technology firms undertake R&D in standardization which would take into account foreign countries versus in the United the needs of Government, the private sector, States. and the consumers as well as balance short- ● The effects of joint ventures, countertrade term needs with the long-term development agreements, licensing, and other arrange- of foreign markets. ments on the transfer of U.S. information The United States could establish a more science and technology. coherent policy on the flow of scientific infor- These information gathering activities would mation, and could establish a review and ap- be helpful regardless of the path Congress peal mechanism for DOD’s restrictions on the chooses to take in addressing this issue area. flow of information and technology. Option 3: Set National Policy. While there is Options 2 and 3 are not mutually exclusive, much we need to know, one alternative is to and probably make best sense in concert with begin setting a long-term policy on the role of each other.

Concluding Thoughts

As part of the preparation for this chapter, Indeed, U.S. policy toward information tech- interviews and workshops were conducted nology R&D, as in many other areas, is par- with several dozen experts in science policy tial and incremental. This lack of long-term and information technology R&D. Box B is a perspective may in part be inherent in the sample of their responses to the question, policymaking machinery; in other cases, pol- “What single message would you like to get icymakers have explicitly assumed that the across to the Congress concerning information Government will be most effective when it technology R&D?” The diversity of these re- responds to a mature issue—an issue that has sponses indicates the multifaceted nature of reached a level of public concern where action issues related to information technology R&D. is clearly called for, and the background of the Their responses are reprinted in box B in or- issue is well understood. der to illustrate the wide-ranging priorities of a group of well-informed specialists, and to However, the nature of a “core” technology, provide a different perspective on some of the facilitating major and pervasive social changes, issues discussed earlier in the chapter. raises questions about the utility of partial and incremental policies. Many of the issues evolv- A common theme in these comments, and ing from a core technology are likely to evolve in many other discussions of policy in this late rather than early, and are likely to be area, is a drive for perspective: for a long-range structural-that is, deeply built into the so- view of technological and social changes, and ciety, and hence very disruptive and traumatic for policies that work together effectively in to correct. a wide range of areas. Ch. 8—information Technology R&D in the Context of U.S. Science and Technology Policy ● 303

ons Of this country re- en given the latitude to do so. t legislate on matters of R&D except— where you

- Take “ rous action now to enhance science education for peoT e at aU levels and all education. Provide %~~”:“ .- strong support for it. WXIRC13: UI’A wdshaw ad intervkws.

In response to some of these concerns, vari- rather than long-term research on social im- ous interested parties have called for some pacts of science and technology .25 kind of prestigious national body which could Given that such little effort is now being help sort out issues and lay the groundwork 23 undertaken to understand the long-term ef- for a long-term perspective. Congress may fects of information technology, it is difficult wish to consider such an option. Although one to say whether development of such a perspec- could be skeptical about creating another com- tive would be possible in the United States. mission, other countries have tried variations However, such efforts probably entail little on this theme with some apparent success in 24 risk, in that any insights derived could help developing long-term perspectives. In the inform policymakers on information technol- United States, this area of research is rather ogy R&D, and on use of the technology itself. anemic. For example, the National Science An examination and anticipation of the social Foundation recently reorganized its Policy Re- and cultural impacts of information technol- search and Analysis Division to address the ogy could at a minimum suggest avenues to short-term needs of the executive branch explore and monitor, possible options to con- “see, for example, J. L. Kirkley, “Backing into the Future, ” sider promoting, and identification of poten- Ikkrmtion, February 1982, p. 31; M. R. Wessel and J. L. Kirkley, “For a National Information Committee,” DAuna- tial developments that one might wish to tion, 1982, p. 234; David Burnharn, The Rise of the Computer thwart or prevent. State (New York: Random House, 1983). For a discussion also see OTA, “Institutional Options for Addressing Information Policy Issues, ” op. cit. 24% Telecom Australia, Telecom 2000: An Exploration of the Long-Term Development of Telecommunications in Aus- tralia (Melbourne, Australia: Australian Government Printing Z6SW Nation~ Science Foundation, Division of policy Unit, 1975); and Nora, S. and A. Mine, The Computerization Research and Analysis, “Program Announcement and Solicita- of Society (Cambridge, MA: MIT Press, 1980). tion, ” January 1984. - -

Chapter 9

Technology and Industry ——

Contents Page Importance of Information Technology ...... 307 Findings ...... 307 Composition of Information Technology ...... 308 Functions ...... 308 Examples of System Applications ...... 309 Characteristics of theU.S. Information Technology Industry ...... 316 Employment ...... 317 International Trade ...... 317 Industry Structure ...... , ...... , ...... 320 Small Entrepreneurial Firms ...... 321 Where the Technology is Heading ...... 323 Microelectronics...... 324 Silicon Integrated Circuits...... 324 Gallium Arsenide Integrated Circuits ...... 328 The Convergence of Bio-technology and Information Technology ...... 330 Software ...... 331 The Changing Role of Software in Information Systems ...... 332 The Software Bottleneck ...... 333 Software and Complexity ...... 334 Conclusion ...... 335

Tables Table No. Page 50. Functions, Applications, and Technologies ...... 309 51. Characteristics of Human Workers, Robots, and Nonprogrammable Automation Devices...... , ...... , ...... , . 312 52. Comparison of the U.S. Information Technology Industry With Composite Industry Performance, 1978-82 ...... 317 53. Employment Levels in the U.S. Information Technology Industries ...... 318 54.U.S. Home Computer Software Sales ...... 320 55.U.S. Office Computer Software Sales...... 321 56. Percentage of Total industry Sales and Employment by Size of Small Businesses ...... 322 57. Venture Capital Funding and Information Technology Firms, 1983 ...... 323 58. ICs Made With Lithographic Techniques ...... 326 59. Software Sales by Use , ...... 332 Figures Figure No. Page 52. Direct Broadcasting Satellite System ...... 311 53. The Changing Structure and Growth of the U.S. Information Technology Industry ...... 318 54. Balance of Trade, Information Technology Manufacturing Industry ...... 319 55. U.S. Trade Balance with Selected Nations for R&D Intensive Manufactured Products ...... 319 56. Historical Development of the Complexity of Integrated Circuits ...... 324 57. Read-Only Semiconductor Memory Component Prices by Type ...... 325 58. Median Microprocessor Price v. Time ...... 325 59. Minimum Linewidth for Semiconductor Microlithography ...... 326 60. A Schematic of an Electron Beam Lithography System, and Its Use to Make LSI Logic Circuits ...... 327 61. Trends in Device Speed–Stage Delay per Gate Circuit for Different Technologies ...... 328 62.The Relative Costs of Software and Hardware...... 332 Chapter 9 Technology and Industry

Importance of Information Technology

Information technology and the industries nology industry. ” Because of the varying that advance and use its products are becom- definitions and the differing statistical anal- ing increasingly important to America’s eco- yses arising from them, it is impossible to pin nomic strength. Information technology is a down the size of the industry. A relatively con- core technology, contributing broadly to the servative estimate, discussed later in this Nation’s trade balance, employment, and na- chapter, places the annual sales of the U.S. in- tional security. The information sector already dustry at over two hundred billion annually accounts for between 18 and 25 percent of the and growing. From another perspective: over gross national products of seven of the mem- one-half million jobs depend on computer in- ber nations of the Organization for Economic dustry exports.4 Cooperation and Development (OECD) and between 27 and 41 percent of employment in these countries.12 2 Findings

An important characteristic of this technol- In examining the composition of informa- ogy is the rapidity with which research results tion technology, the characteristics of the U.S. are translated into “advanced” products and industry, and where the technology is heading, from there to the mainstream consumer market- several trends become apparent: place. This rapid transfer has been particularly 1. The U. S. information technology industry evident in semiconductor research leading to is large and growing rapidly; there is a world cheaper and more powerful microprocessors. market for its products and services as the Given the accelerated movement from the developed countries increasingly move into research laboratories to a worldwide market- the information age. place, it is natural that the attention of the 2. The technology is pervasive, making pos- trading nations has focused on information sible major productivity improvements in technology research and development (R&D). other fields, creating new industries, and Two basic building blocks of information tech- enlarging the range of services available to nology are: the public. The technology diffuses through ● the microelectronic chip-the large scale most facets of life, including business, engi- integrated circuit that permits the stor- neering and science, and government func- age, rapid retrieval and manipulation of tions. vast amounts of information, and 3. With each level of technological advance in ● software-the sets of instruction that di- basic information technology, the level of rect the computer in its tasks.3 complexity and cost increases for R&D, as does the demand for additional technical As noted in chapter 2, there is no consen- training for R&D personnel. sus on what composes the “information tech-

‘Information Activities, Electrom”cs and Tekwomtnum”cations Technologies: Impacts on Employment, Growth and Trade ‘Robert G, Atkins, The Computer Industry and International (OECD, Paris, 1981), pp. 22 and 24. Trade; A Summary of the U.S. Role, Information Processes ‘M. R. Rubin, Information Econom”cs and Policy in the Group, Institute for Computer Sciences and Technology, Na- United States, Libraries Unlimited, Littleton, CO, 1983, tional Bureau of Standards, December 1983, p. IX-1. Draft of pp. 32-44. a report under joint development by ICST and the International 3Software may also refer to the stored information. Trade Administration, as of February 1985.

307 308 • information Technology R&D: Critical Trends and Issues

4. Domestic and foreign competition for world 5. There will be continued rapid advances in markets is both intensive and escalating. the capabilities of the industry’s underlying The stakes for the United States are great basic building blocks, microelectronics and in terms of corporate sales and competitive software. ness in world markets, and employment.

Composition of Information Technology

Individual information technologies have 3. Information Storage. The storage media grown explosively in technical sophistication associated with the information industry and, at the same time, diminished in cost. That are the electronic-based devices which rare combination has propelled the new tech- store data in a form which can be read by nologies into the mass marketplace. As the a computer. They include film, magnetic two basic building blocks continue their ad- tape, floppy and hard disks, semiconduc- vances, the parade of new information tech- tor memories, and so on. The ability to nology capabilities and applications will go on. store increasingly vast amounts of data There is a recurrent pattern in which one has been essential to the information technology revolution. technological advance makes possible still 4. Information Processing. Information proc- other advances, often in a “bootstrap” fashion. For example, sophisticated computer- essing is the primary function of a com- aided design (CAD) equipment is a tool used puter. The information stored by a computer can be numeric (used for com- in development of state-of-the-art random ac- putations), symbolic (rules of logic used cess memories and microprocessors; develop- for applications such as “expert” sys- ment of the next generation of super-comput- tems), or image (pictorial representations ers depends on use of today’s most advanced used in applications such as remote map- computers; “expert” systems— still in their ping). The stored information–in what- technological infancy-are already employed ever form—is manipulated, or processed, in configuring complex computer systems and in response to specific instructions (usu- custom-designing integrated circuits. ally encoded in the software). The increasing speed of information processing has Functions been another essential factor in the infor- For purposes of this report, the term “information technology revolution. mation technology” has been used to refer to 5. Communications. Electronic communica- the cluster of technologies that provide the fol- tions utilizes a variety of media-the air lowing automated capabilities: waves (for broadcast radio and television), coaxial cable, paired copper wire (used, 1. Data Collection. Examples of automated among other things, for traditional tele- data collection systems range from large- phony), digital radio, optical fibers, and scale satellite remote-sensing systems communications satellites. Communica- such as weather satellites to medical ap- tions systems play a major role in broaden- plications such as CAT-scans and elec- ing the use of other facets of information trocardiograms. technology and make possible distributed 2. Data Input. Input devices include the fa- computing, remote delivery of services, miliar keyboard, optical character read- and electronic navigation systems, among ers, video cameras, and so on. They are many other applications. the means by which data are inserted and 6. Information Presentation. Once the infor- stored, communicated, or processed. mation has been sent, it must be “pre- ‘~

Ch, 9–Technology and Industry ● 309

sented” if it is to be useful. This can be Table 50 shows some of the technologies and accomplished through a variety of output application areas that depend on these functions. devices. The most common display technology is the cathode ray tube or video Examples of System Applications display terminal. Hard-copy output devices include the most commonly used im- A few examples of the many information pact printers as well as those using non- technology applications are provided below. impact technologies such as ink-jet and They have been chosen to illustrate the diver- xerography. There are also audio systems sity of applications, and in most cases reflect that permit the computer to “speak”- capabilities that have only become feasible on exemplified by the automobiles that ad- a significant scale in recent years. The ex- monish you to fasten your seat belts. amples include: cellular mobile radio commu-

table 50.—Functions, Applications, and Technologies”

Representative Function Typical application areaa information technology Data collection Weather prediction Radar, infra-red object detection equipment, radiometers Medical diagnosis CAT-scanners, ultrasonic cameras Data input Word processing Keyboards, touch-screens Factory automation Voice recognizes (particularly for quality control) Mail sorting Optical character readers Storage Archives Magnetic bubble devices, magnetic tape Accounting systems Floppy disks Scientific computation Wafer-scale semiconductors (still in research phase), very-high-speed magnetic cores Ecological mapping Charge-coupled semiconductor devices, video disks Libraries Hard disks Information processing Social Security payments General purpose “mainframe” computers, COBOL programs Traffic control Minicomputers Distributed inventory control Multi-user super-micros, application software packages Medical diagnoses “Expert” systems Engineering design Spreadsheet application packages, microcomputers Scientific computation Supercomputers: multiple instruction-multiple data (MI MD) processors, vector processors, data driven processors, FORTRAN programs Ecological mapping Array processors, associative processors Factory automation Robotics, artificial intelligence Communications Office systems Local area networks, private branch exchanges (PBX), editor applications packages Teleconferencing Communications satellites, fiber optics Rescue vehicle dispatch Cellular mobile radios International financial Transport protocols, data encryption, Integrated transactions Services Digital Networks (ISDN) Data output and presentation Word processing Personal computers, printers (impact, ink jet, xerographic) Management information Cathode ray tubes, computer graphics Pedestrian traffic control Voice synthesizers aThi~ list is not ~xhau~tive; any given technology may also be used for some of the other applications mentioned.

SOURCE Office of Technology Assessment. 310 . Information Technology R&D: Critical Trends and Issues

nications, direct broadcast satellites, robotics, mobile telephone services are now available in computer simulation, a biomedical application Tokyo. In Spain, some 20,000 subscribers are in heart pacemakers, and financial services. anticipated by 1990. The Netherlands reached approximately 50,000 inhabitants by 1984. Cellular Mobile Radio Communications Saudi Arabia introduced cellular mobile radio communications in 1981 in three cities; by For decades, use of mobile telephone serv- 1982, there were 19,000 subscribers, and the ice has been limited by over-crowding of the Saudis have now extended coverage to 32 electromagnetic spectrum; major cities have cities. had only some few hundred subscribers because of spectrum constraints. Demand for ad- Direct Broadcast Satellites ditional mobile voice service has been growing at a 12 percent rate annually for the past The use of direct broadcast satellites (DBS) 20 years. A new technology-cellular mobile has recently become feasible. With DBS, over- radio communications—holds promise for the-air signals (e.g., TV, radio) can be received vastly increasing the number of potential directly by small rooftop antennas (see fig. 52). subscribers (to more than 100,000 in particu- These new, commercial, geostationary sys- lar geographic areas),’ while providing serv- tems may have widespread use in the United ice of improved quality compared to that availa- States and in other countries, primarily pro- ble previously. What is most significant viding entertainment, but also with potential about the cellular concept is that it permits for education, advertising and other business conservation of the electromagnetic spectrum uses. As of early 1985, nine applications had —a limited natural resource. The older technology provided service from a single antenna to the area served. The cellular concept is based on dividing the service area into a number of geometric shapes, or cells, each served with its own antenna. The cell antennas are interconnected with leased telephone lines. When a vehicle leaves a cell, that cell passes control to the next cell, and so on. The number of subscribers is much greater with the cellular concept because the transmission frequencies can be reused re- peatedly in nonadjacent cells. Thus the impact of increasing demands on the spectrum is min- imized. According to some estimates, cellular radio will be a $4 billion U.S. industry by 1990 and could reach $6 billion by the mid-1990s. Cel- lular mobile telephones recently cost about $3,000 but prices are likely to fall rapidly because of competition among vendors. Other countries are also finding considerable demand for mobile radio services. In Japan,

60peration of cellular mobile radio systems in dozens of U.S. cities has been approved by the Federal Communications Com- Photo credit: Motorola mission. Mobile cellular radiotelephone Ch. 9—Technology and Industry ● 311

Figure 52.—Direct Broadcasting Satellite System in components and subsystems available today. DBS systems had their genesis when NASA’s ATS-6 satellite was launched in 1974—a system having a multiplicity of payloads, two of which included TV broadcast capabilities. A number of countries have DBS systems in , Indoor unit place or in planning stages. Included among these are: Canada’s medium-powered ANIK C-2, which is serving Canadian audiences as well as providing five channels of television entertainment to the Northeastern United States; Japan’s modified BS-2 satellite, launched in early 1984; and France’s TDF-1 and West Germany’s TV-SAT, which are expected to be operational in 1986.

Robotics Robots are mechanical manipulators which been approved by the Federal Communica- can be programmed to move workplaces or tions Commission (FCC) for subscription serv- tools along various paths. They are one of the ices to the public, some of which may become four tools employed in computer-aided manu- operational. facturing (i.e., robots, numerically controlled machine tools, flexible manufacturing sys- DBS services will compete with other media tems, and automated materials handling such as multiple distribution systems, STV systems).6 (subscription TV), and cable TV (CATV) for serving television viewers. DBS services may Robotics emerged as a distinct discipline prove especially valuable for providing serv- when the century-old industrial engineering ice to sparsely populated geographical areas automation technologies converged with the that are not economical to reach by other more recent disciplines of computer science means, as well as to densely populated cities and artificial intelligence. The convergence where new cable installations are prohibitively produced the growing field of robotics, which expensive. has had wide factory applications in areas that include (but are not limited to) materials han- Among the technical improvements that dling, machine loading/unloading, spray paint- make DBS systems feasible are higher power ing, welding, machining, and assembling. The and more efficient on-board power amplifiers; robots are most often used in performing par- solar power generation equipment; more ac- ticularly hazardous and monotonous jobs curate satellite station-keeping control; im- while offering enough flexibility to be easily proved ground receiver sensitivity; the use of adapted to changes in product models. higher frequencies; and better transmitter downlink beam control. Higher power and Neither today’s robots, nor those likely to higher frequencies also make practical the use be available in the next decade, look like of small (about 1 meter or less) rooftop an- humans nor do they have more than a fraction tennas. Some of the early U.S. experimental of the dexterity, flexibility, or intelligence of satellite communications systems, along with humans. A simple “pick and place” machine parallel advances in solid-state electronics and ‘See Computerized Manufacturing Automati”on: Employment, new materials processing techniques, have Education, and the Workplace (Washington, DC: U.S. Congress, contributed significantly to the improvements Office of Technology Assessment, OTA-CIT-235, April 1984.

38-8o2 0 - 85 - 21 312 ● Information Technology R&D: Critical Trends and Issues

At the end of 1983 Japan had about 43,000 operating robot installations-by far the largest number in any country. The United States, where the original patents for robots were obtained, had about 9,400 installations (v. about 13,000 in early 1985 and 6,300 in 1982), followed by West Germany (4,800), and France (3,600). Some projections indicate that by 1990 there will bean installed base of nearly 90,000 robots in the United States. * As reported by OTA6a, significant robotics research is being conducted in over a dozen universities, about three dozen industrial firms and independent laboratories, and in Federal Government labs at the National Bureau of Standards (NBS), NASA, and DOD. The report further noted that the research was in- with two or three degrees of freedom may cost tensively addressing several problem areas: roughly between $5,000 and $30,000, while improved positioning and accuracy for the ro- more complex programmable models, often bot’s arms; increased grace, dexterity, and equipped with microcomputers, begin at about speed; sensors, including vision, touch, and $25,000 and may exceed $90,000. force; model-based control systems; software; mobility; voice recognition; artificial intelli- Human workers, robots, and nonprogram- gence; and interface standards. The interface mable automation devices each have certain area is critical to widespread use of automated characteristics which offer advantages for the manufacturing. NBS has created an Auto- manufacturing process. Table 51 compares mated Manufacturing Research facility in part some of the salient characteristics. to perform research on the interfaces between different computerized devices in a factory.

Table 51 .—Characteristics of Human Workers, Computer Simulation a Robots, and Nonprogrammable Automation Devices “Computer simulation” is a process that Non programmable employs a computerized model of certain sig- Human Automation nificant features of some physical or logical Characteristic Worker Robot Device system which is undergoing dynamic change. Flexibility ...... 1 2 3 We therefore find computers used in applica- Consistency...... 3 2 1 Endurance ...... 3 1-2 1-2 tions such as economic modeling and war Ability to tolerate games where “what if” scenarios can be played hostile out to test suggested social policies or military environments...... 3 1-2 1-2 cost ...... n/a 2 1 strategies. As relatively inexpensive computer Speed ...... 2-3 2-3 1 memories grow in size and integrated circuits Intelligence/ programmability . . 1 2 3 grow in speed, increasingly complex opera- Sensing ...... 1 2 3 tions are being modeled. Software advances Judgment...... 1 2 3 have also played a large part in the growing Ability to adapt use of computer simulations. to change...... 1 2 3 Dexterity...... 1 2 3 *worldwide Robotics Survey ~d D&toW, 1984, to ~ pub- aNumerical ranking indicates relative advantage, With “l” Indicatingthe greatest advantage. lished, Robotics Industries Association, Dearborn, MI. This data is based on the RIA’s definition for robots which excludes SOURCE: Computerized Manufacturing Automation: Employment, Education, and the Workp/ace, (Washington, DC: U.S. Congress, Office of the simpler, nonreprogr amrnable machines. Technology Assessment, 1964). This table was derived from the fore- @Compute~~ M~ufactufl-ng Automation: Employment, going report Education and the Workplace, op. cit.

Ch. 9–Technology and Industry ● 313

When combined with other information sible by the rapid calculation and implemen- technologies, some startling simulations are tation of flight characteristics related to the possible. In flight simulators, for example, known aerodynamic and engine properties of computer-generated imagery techniques are the craft, air speed, altitude, attitude, and lift. used to create background, images, sounds, Some systems use 10,000 variables associated and sensations, and to position discrete ele- with a single-engine jet fighter. In order to ments in a scene—elements which have been achieve visual accuracy, simulator systems obtained from video images of real trees, calculate the aircraft’s position between 25 buildings, clouds, aircraft, and other charac- and 60 times per second. teristics of the environment to be simulated. The flight simulator is placed in the dynamically changing recreated environment and the The Federal Aviation Administration per- pilot then “flies through” in the simulator, mits 100 percent simulation training for ex- which duplicates the performance of the actual perienced pilots who are upgrading their flight airplane. Realistic simulations are made pos- certifications to more advanced aircraft. Simu- 314 . /formation Technology R&D: Critical Trends and Issues lated dogfight training at Williams Air Force Pacemakers are medical devices that can be Base is so realistic that some F-4 pilots have surgically implanted under a patient’s skin become airsick.7 with electrode leads fed through the veins into the heart. The two main components of a pace- Computer simulation is proving effective for maker are the pulse generator and the leads. improving the quality of training in a variety Pulse generators contain a power source and of jobs—e.g., operating a locomotive, control- the electrical circuitry for sensing, pulsing, and ling a supertanker, operating a space shut- progr amming. Some new pacemakers are also tle—while reducing the cost, time, and risk capable of telemetry functions, which transmit associated with conventional training. certain critical measurements of performance of the pacemaker and the patient’s condition. Heart Pacemakers Programming changes and telemetry func- Microelectronics and software systems are tions require external microprocessors that finding a variety of applications in medical communicate with the implanted devices in or- diagnoses and treatment. The heart pace- der to monitor performance and to modify the makers now being used by more than 500,000 pacemaker’s operation. patients in the United States are illustrative The first implantable pacemaker was devel- of the widespread use of these technologies for oped in 1958, and the first implantation in a medical purposes. human was performed in 1959. Early pace- The heart muscle contracts in response to makers sent electrical impulses to the heart electrical impulses generated by apart of the at fixed intervals. By 1970, the technology for heart itself. The contractions force the blood “demand” pacemakers was developed. These through the arteries to all parts of the body; pacemakers sense when the heart is not work- but if there is something wrong with the elec- ing properly and, when necessary, send out trical impulses, blood flow is impaired and the electrical impulses to trigger the contraction heart muscle becomes injured. of the cardiac muscle. Programmable pacemakers-which are continually being updated ——-——. - “’The Technology of Illusion,” Forbes, Feb. 27, 1984, pp. with advanced technology— can be repro- 158-162. grammed without additional surgery, in response to changes in the patient’s physiology. This capability reduces the risks of additional surgery for cardiac patients and decreases the costs of their care.8 The latest generation of pacemakers is designed to be flexible, enabling updating of the software and external keyboard to accommodate newer technology. The flexibility ex- tends to enabling pacemaker parameter settings to be customized for each patient’s special needs, and to the use of the same technology for other medical applications, such as drug dispensers and pain controllers.9

%pecisl Committee on Aging, U. S. Senate, Fraud, Waste, and Abuse in the Medz”care Pacemaker Industry, (Washington, DC: U.S. Govemxnent printing Office, 98-116. September 1982). ‘Richard M. Powell, “A New Programm er for Implanted Pulse Photo credit: Medtronlc, Inc. Generators,” pp. 678-687, technical paper presented at the 15th Hawaii International Conference on Systems Sciences, Janu- Heart Pacemaker ary 1982. —

Ch. 9—Technology and Industry . 315

Financial Services There are also several card technologies. The The financial service industry would not pro- embossed plastic card with its strip of mag- vide the level of service it does without infor- netic tape provides the primary means for mation technologies. 10 The numbers of checks accessing credit and debit services that are delivered through both paper-based and elec- (over 37 billion annually), credit card drafts tronic systems. Laser cards-not yet used by (over 3.5 billion annually), and securities trades the financial service industry–can store dig- (over 30 billion shares traded annually) are ital data signifying the bearer’s fingerprints. simply too much for any manual system to Electron cards combine three encoding tech- handle. In the 1960s, for example, before the nologies-the banking industry’s magnetic financial service industry was substantially automated, there were days when the New tape, the retail industry’s optical character York Stock Exchange suspended operations recognition, and the UPC bar code. because the broker/dealers were unable to han- Document and currency readers have had dle the workload. some acceptance, particularly for the reading of checks encoded with magnetic ink. Systems Both users and providers of financial serv- that process credit and debit card transactions ices have made use of virtually every category already truncate the paper flow at the earli- of information technology. Banks and other est practical time. The data are recorded on financial service providers have been longtime magnetic media and transferred electronically users of computers. However, an increasing for processing by the card issuer. number of retailers-large and small-are installing the hardware and using the telecom- Together, these technologies have resulted munications networks to verify checks and to in a financial service industry that is offering authorize and complete credit transactions. an increasing number of services such as automated redemption of money market funds. At Applications software packages for financial the same time these services are being pro- services can process market data and gener- vided in an increasing number of ways, in- ate information used for portfolio manage- cluding the use of automated teller machines ment. They can also generate and analyze loan and telephone bill payers. information, and process bank card applications. The spread of remote terminals inter- With all this, there is the possibility of connected with central computers is becom- redistribution of functions among traditional ing ubiquitous, and expanding the number of suppliers as well as potential new entrants. services available to the public. Whereas in the past the payment system has been reserved largely to banks, because they The video-related technologies-videotex (a had access to facilities for clearing and settle- two-way information service) and teletext (a ment, movement of funds electronically makes one-way information service)—are not yet it possible to avoid the traditional payment widely used in the delivery of financial serv- system and to settle directly between trading ices, but experiments are under way and it is partners. Alternative means of distributing in- likely that they ultimately will gain accept- formation could diminish the role of brokers ance. One videotex system, currently being for such products as securities, real estate, and marketed in Florida, uses an AT&T terminal insurance. Cash-oriented businesses, such as attached to a television set to link the finan- gas stations and supermarkets, already use on- cial service provider with the customer. —.— site automated teller machines to relieve the ‘°For detailed discussion, see Effects of Information Tech- requirement to cash checks while minimizing nology on Financial Service Systems, (Washington, DC: U.S. the amount of currency that is held at each Congress, Office of Technology Assessment, OTA-CIT-202, Sep- tember 1984). store location. 316 ● Information Technology R&D: Critical Trends and Issues

Characteristics of the U.S. Information Technology Industry

The information technology industry is a ● profits/sales ratios ranging between 4.2 composite of several industries both in man- to 5.7 percent for the composite vs. 7.9 ufacturing and services. There is no single, to 9.7 percent for the information technol- generally accepted definition of the industry. ogy sector. The definitions in use by the Information In- ● growth in the number of employees: a de- dustry Association, for example, are too broad crease of 7.8 percent for the composite vs. for the purposes of this report, while defini- an increase of 11.8 percent for the infor- tions used by the Computer and Business mation technology industry. Equipment Manufacturer’s Association and ● growth in R&D expenditures: 81 percent by the Association of Data Processing Serv- for the composite vs. 111 percent for the ice Organizations are too restricted. information technology sector. The information technology sector’s R&D expend- Despite the ambiguities, this section at- itures per employee were higher than the tempts to characterize the industry in terms composite by about 20 percent annually of size and structure, growth, employment, infor the period, and both groups increased vestment in R&D, and other areas of special their R&D investments in spite of a reces- concern. The part of the information technol- sion.13 ogy industry portrayed here includes primar- ● investments in R&D as a percentage of ily manufacturers of: electronics; information sales: averaging 2.1 for the composite vs. processing equipment, including computers, 4.2 for the information technology sector; office equipment, and peripherals and services; and, as a percentage of profits, 42 vs. 48, semiconductors; and telecommunications respectively. equipment. The above statistics are incomplete because 52 summary of the Table is a key indicators firms whose primary business is not informa- for the composite 30 industry categories cov- ll tion technology do not appear as part of the ered by Business Week for the period 1978 industry—despite the fact that their informa- to 1982, compared with aggregated data for tion technology activities may be significant. part of the information technology industry.” Among the firms not represented are General A point of contrast worth noting is that infor- Electric, Rockwell International, and West- mation technology firms’ business perform- inghouse. ance during the period outpaced the composite industry average significantly as measured The above data for the information technol- by the following criteria: ogy manufacturing industry, when combined with related sectors such as the software and ● growth in sales revenues: 40 percent for the composite industry groups vs. 66 per- computer services, and telephone and telegraph services sectors of the economy, had cent for the information technology sector. ● total revenues of $229 billion for 1982, up from growth in profits: 6.4 percent for the com- $180 billion in 1980, for a 27 percent increase. posite vs. 36.4 percent for the information Figure 53 illustrates the increasing portion of technology sector.

llBu~jne~~ W=k, R&D Scoreboard, June 20, 1983, PP. 122- 153. See also Business Week, July 9, 1984, pp. 64-77, which ‘Klf the four sectors-industry, Federal Governrnent, colleges shows a continuation of the trends through 1983. and universities, and other nonprofit organizations-only in- ‘zData shown by Business Week for national composite industry increased its funding for R&D in constant 1972 dollars dustry R&D expenditures correlates closely with NSF data in during the 1978-82 period. Probable Levels of R&D Expench”- Science Resources Stud-es “Highlights,” June 11, 1982. tures in 1984: Forecast and Analysis, Battelle, December 1983. Ch. 9—Technology and Industry . 317

Table 52.—Comparison of the U.S. Information Technology Industry with Composite Industry Performance, 1978-82

Percent 1978 1979 1980 1981 1982 change . — — Sales (mill Ions of dollars) . . . . . 1,085,291 1,277,764 1,421,551 1,586,510 1,520,313 40 Composite 131,872 149,783 174,449 193,921 218,862 66 Infotech

Profits (millions of dollars) . . 59,578 72,505 73,493 81,757 63,365 6.4 Composite 12,780 13,821 15,474 16,056 17,436 36.4 Infotech

Profits/sales (percent) ...... 5.5 5.7 5.2 5.1 4.2 Composite 9.7 9.2 8.9 8.3 7.9 Infotech

Employees (thousands) ...... 15,133 15,542 15,498 15,045 13,959 – 7.8 Composite 2,952 3,099 3,226 3,252 3,301 11.8 Infotech

R&D (millions of dollars) ... . 20,610 24,674 28,984 33,285 37,179 81 Composite 4,961 5,885 7,221 8,531 10,473 111 Infotech

R&D $/sales (percent) ...... 1.9 1.9 2.0 2.1 2.5 Composite 3.8 3.9 4.1 4.4 4.8 Infotech

R&D $/profits (percent) ...... 34.6 34.0 39.4 40.7 59.0 Composite 38.8 42.6 46.7 53.1 80.1 Infotech

R&D $/employee ...... 1,362 1,588 1,870 2,212 2,667 Composite 1,680 1,899 2,238 2,623 3,173 Infotech R&D expenditures per employee lnfotech/Composite (percent) . . . 123 120 121 119 120 Business Week “Scoreboard” Numbers Notes: A This IS a sample of R&D spending in information technology by U.S. corporations, It is based on total R&D expenditures for those companies that are publicly held, have annual revenues over $35 million, and R&D expenses of $1 million or 1 percent of revenue. Only that spending by companies whose primary business is Information technology (electronics, computers, office equipment, computer services and peripherals, semiconductors, and telecommunications) is included. B Sales, R&D spending, and R&D spending per employee figures have been adjusted to reflect the numbers from Western Electric and other AT&T subsidiaries that are not included in the “Scoreboard” numbers, This adjustment involves: 1 addition of revenues received from Western Electric to the total operating revenues figures in the AT&T Annual Reports for the years covered, 2 use of the total AT&T spending figures for R&D which include spending by Western Electric and other AT&T subsidiaries as provided in Business Week for the years 1980-82 and as estimated from a chart in the 1983 AT&T Annual Report for the years 1978 and 1979. 3 use of total AT&T employment figures provided in Forbes each May for the years 1978-82. C Employment numbers for all sectors have been calculated from the R&D spending per employee and the R&D spending figures provided in Business Week and may reflect rounding errors, SOURCE. Data obtained, or calculated from Business Week, Scoreboard, June 30, 1983 the U.S. Commerce Department; Forbes, May 1979 throuah 1983; 10K forms filed by AT&T and Western Electric Corp total sales due to services, which are approach- services closely matches that of computer ing 50 percent of total industry revenues. manufacturing, both growing by about 140 percent between 1972 and 1982. Employment International Trade

Employment in information technology Technology-intensive products have impor- manufacturing increased significantly between tant implications for the balance of trade. 1972 and 1982 in most segments of the indus- Studies on U.S. trade and the influence of tech- try (table 53), experiencing employment growth nology have generally concluded that technol- ranging between 9 and 142 percent. Only the ogy serves as an important determinant of consumer products (radio and TV sets) showed comparative advantage in manufactured goods a decline (28 percent). Total employment in in- trade. 14 15 formation technology manufacturing grew by “C. Mi~el Aho and Howard F. Rosen, “Trends in Technol- 51 percent, in spite of economic recessions. ogy-Intensive Trade: With Special Reference to U.S. Competi- tiveness, ” Office of Foreign Economic Research, Bureau of In- Employment in information technology ternational Labor Affairs, U.S. Department of Labor, p. 11. services is about equal to that of the manu- “T. C, Lowinger, “Human Capital and Technological Deter- minants of U.S. Industries Revealed Comparative Advantage, ” facturing segment, with some 1.6 million em- Quarterly Review of Econozzu”cs and Business (1974), winter ployees. The employment level in computing 1977, pp. 91-102, as reported in Aho and Rosen, pp. 11, 12. 318 ● Information Technology R&D: Critical Trends and Issues

Figure 53.—The Changing Structure and Growth of the U.S. Information Technology Industry Information industry revenues; 1980-83 (dollars in billions) 300

258

229

200 180 Software 34.7% and computer . servicesl’2 - Telephone and telegraph 34.3% service2 100 n Information technology manufacturing

0 1980 1981 1982 1983 Years SOURCES: IADAfJSO; 2U.S. Industrial Outlook, 1983, 1984.

Table 53.—Employment Levels in the U.S. Information Technology Industries Employees (in thousands)

Percent change 1972 1982 1972-1982 Manufacturing a Computers ...... 145 351 + 142 Office equipment...... 34 51 +50 Radio and television receiving sets ...... 87 63 –28 Telephone and telegraph equipment ...... 134 146 +9 Radio and television communications equipment ...... 319 454 +42 Electronic components...... 336 528 +57 Totals, manufacturing ...... 1,055 1,593 Services Telephone and telegraph ...... 949 1,131 +11 Computing b...... 149 360 + 141 Radio and television broadcast ...... 68 81 +19 Cable television ...... 40 52 +30 Totals, services ...... 1,206 1,624 aEstimates provided by the U..S. Department of Commerce, Bureau of Industrial Economics. bF\gures are for 1974 and 1983. Source: U.S. Industrial Outlook, 1984.

CFigures are for 1979 and 1983, Source: Federal Communication Commission in telephone interview with OTA staff, May 1984, dF\gures are for 1981 and 1982. Ibid. (FCC). —

Ch. 9—Technology and Industry ● 319

The U.S. merchandise trade balance has Figure 55.— U.S. Trade Balancea With Selected been negative in recent years: deficits of $42.7 Nations for R&D Intensive Manufactured Products billion in 1982; $69.4 billion in 1983; and over $30 $100 billion in 1984. In 1980, advanced tech- 28 nology products showed a positive trade bal- 26 ance of $31 billion, compared with a deficit of 24 more than $50 billion for all manufactured 22 goods.” ●✍ 20 Developing nations Information technology manufactured prod- 18 ucts17 to usually made positive contributions 16 the U.S. balance of trade between 1972 and 14 1982 (fig. 54), led by sales of computer equipment. However, 1983 and 1984 saw trade 12 deficits of $0.8 and 2.3 billion according to De- partment of Commerce estimates.18 The deficits would have been much greater without about $6 billion in exports of computer equip- 4 ment in each of those years (see fig. 55). —.—— 2 “International Competitiveness in Advanced Technology: Decisions for Aznen”ca, National Research Council (Washing- 0 ton, DC: National Academy Press, 1983, pp. 23-24. -2 “The products include: computer equipment, SIC 3573; office equipment, SIC 3579; radio and TV sets, SIC 3651; -4 telephone and telegraph equipment, SIC 3661; radio and TV -6 communications equipment, SIC 3662; and electronic comp~ 1966 67 68 69 70 71 72 73 74 75 76 77 78 79 80 nents, SIC 367. Years ‘aU.S. Industn”d Outlook, 1978 through 1984. aExports less imports. blncludes West Germany, SOURCE: Science Indicators— 1982, National Science Board, Na- Figure 54. —Balance of Trade, Information tional Science Foundation, 1983. Technology Manufacturing Industry (dollars in billions) Although the United States holds the major market share of the industrialized coun- 6 - tries’ exports of high-technology products, that share has declined from 30 percent in equipment (SIC 3573) 4 1962 to 22 percent in 1978, and has increased only marginally since. In absolute terms, the

2 U.S. positive trade balance in high-technology products increased over eightfold from 1962

o to 1980. During that same period West Ger- many, and especially Japan, starting from smaller bases, have had impressive gains in their surpluses of exports over imports—a ninefold increase for West Germany and two- , hundredfold increase for Japan.19

-6 Radio and TV Communications equipment, 19U+S. Dep~ment of Commerce, International made Admin-

1972 1974 1976 1978 1980 1982 1984 est istration, as reported in “An Assessment of U.S. Competitive ness in High-Technology Industries, ” prepared for the Work- Years ing Group on High Technology Industries of the Cabinet SOURCE: U.S. Industrial Outlook for years 1978 through 1984. Council on Commerce and Trade, final draft, May 19,1982. —— —

320 . Information Technology R&D: Critical Trends and Issues

Industry Structure The telecommunications equipment providers are exceptionally diverse and competitive. Information technology companies in the There are hundreds of firms in this field, rang- United States are heterogeneous in dollar ing from a few U.S. and foreign multinational value of sales, breadth of product lines, num- companies manufacturing a full range of ber of employees, and degree of vertical in- equipment to dozens of medium size, and tegration. Competing for those markets are many hundreds of small companies concen- thousands of firms ranging in size from the trating on a more limited range of products— vertically integrated multinational companies e.g., speech compression products, multiplex- such as IBM, ITT, and the still-large, post- ers, modems, data terminals and local area divestiture AT&T to the more typical, smaller networks. companies that produce a narrow cluster of products with a large technology content. The electronics industry is even more di- While the majority of sales dollars are concen- verse. The thousands of electronic systems trated in the large companies, the vast ma- and equipment providers, as well as consulting jority of companies are relatively small. firms, range in size from those with billions of dollars in annual sales to small, start-up The diversity in the sizes of the businesses ventures such as the Apple Computer’s garage that populate the information technology field operation of a few years ago. For those seg- is important for reasons that include the bene- ments of the industry undergoing rapid tech- fits of competition; the impetus to innovation nological change, diversity is often accelerated due to firms’ willingness to take risks and to by spin-offs from rapidly growing companies, try new directions; and the tendency to fill and other new entries into the field. niche market demand for specialized products. The U.S. telecommunications, electronics, and The U.S. microcomputer software industry software industries provide three illustrations is an example of rapid change, growth, and of the contrasts in the breadth of that di- diversity. Future Computing, Inc. estimates versity. that the U.S. market for home computer software grew by 168 percent from 1982 to 1983, The telecommunications services industry and projects an 85 percent growth from 1983 is simultaneously both a mature industry to 1984, with the growth rate declining to 26 dating back to the 19th century and a grow- percent by 1988 in a projected $5 billion an- ing industry marked by limited competition nual market by then (table 54). Software unit in some markets and diversity in others. While sales for office use are projected to enjoy com- local public telecommunications services are parable growth. Table 55 indicates growth largely dominated by regulated monopolies, rates of 87 percent between 1982 and 1983, long-distance services are now offered in a and 58 percent from 1983 to 1984, gradually competitive market. New services, such as tapering off to 24 percent in 1988 in a $6.7 bil- paging, cellular mobile radio services, bypass lion annual market.20 Like the electronics in- services and direct broadcast satellite services, dustry, software firms number in the thou- are not dominated by the established carriers,

20 Future computing, ~r n~ but are being offered by a variety of firms. go communication, Mmch 1984.

Table 54.—U.S. Home Computer Software Salesa

1981 1982 1983 1984 1985 1986 1987 1988 Software units (millions) ...... 1.1 6.9 19.0 37.0 58.0 82.0 112.0 144.0 Percent growth (units) ...... 486 168 85 52 42 32 26 Revenue (millions of dollars)...... 48 282 757 1,400 2,100 3,000 4,000 5,000 aData for 19u.M are estimates by Future cOrnPUllW. SOURCE: Future Computing, Richardson, TX, by OTA staff telephone interview, March1964, Ch. 9—Technology and Industry . 321

Table 55.—U.S. Office Computer Software Salesa

1981 1982 1983 1984 1985 1986 1987 1988 Software units (millions) ...... 3.2 6.1 11.0 17.0 25.0 35.0 46.0 59.0 Percent growth (units) ...... 111 87 58 44 36 30 24 Revenue (millions of dollars) ...... 343 724 1,351 2,100 3,100 4,200 5,400 6,700 aData for lgsd.aa are estifnates by Future Computln9 SOURCE Future Computing, Richardson, TX, by OTA staff telephone interview, March 1984 sands. Although no hard data are available, Small Entrepreneurial Firms it is the impression of industry analysts that the rapid pace of startups in the software in- Small businesses play a central role in U.S. dustry seen in recent years now appears to be industry in general, accounting for: slackening. ● thirty-eight percent of the Nation’s gross An important trend in industry structure is national product; toward concentration in the software and data ● two-thirds of all new jobs; ● two-and-a-half times as many innovations services industries, where larger firms have ac- 21 quired hundreds of smaller ones. Among these per employee as large firms; and larger firms are Automatic Data Processing, Ž a tendency to produce more ‘‘leapfrog’ Electronic Data Systems, and General Electric. creations compared to large companies and to introduce new products more Larger firms are also integrating their prod- quickly than large firms.22 uct lines-some moving into subsystems and components, and others, such as semiconduc- The contributions of small businesses to the tor manufacturers, moving into subsystem information technology industry are signifi- and computer manufacturing. AT&T, Texas cant. Small businesses with less than 500 em- Instruments, Intel, and Northern Telecom- ployees account for 33 percent of sales in elec- munications Corp. are examples of companies tronic components and 34 percent of the recently expanding into computers and related Nation’s employment in this sector. In the products. computer services sector, small businesses account for 69 percent of the industry’s sales and There is also an active trend toward affilia- 67 percent of its employees. These firms also tions between information producers and dis- make contributions in sectors where large tributors. Among these are publishers of news- companies dominate, such as in office com- papers, books, and magazines and broadcast, puting equipment, consumer electronics, and cable TV, and interactive computer network communications equipment (table 56). companies. Opportunities for innovation based on ad- Geographic diversification is taking place at vances in information technology have been a rapid pace as foreign-primarily European— numerous. Consequently, hundreds of market companies acquire U.S. firms. Among the niche applications have been created that European firms are: Olivetti, CAP Gemini, Racal Electronics, Schlumberger, Thomson- CSF, and Agfa-Gevaert. “Advocacy: A Voice for Small Businesses, Office of Ad- A number of U.S. firms are expanding their vocacy, U.S. Small Business Administration, 1984. p. 1. markets by establishing joint ventures or by 22Stroemann, 1977. Karl A. Stroetmann “Innovation in Small and Medium Sized Firms.” Working paper presented at Institut acquiring foreign outlets—e.g., IBM, AT&T, Fur Systemchnik und Innovations forschung, Karlsruhe, West and Datapoint. Germany, August, 1977. See Vesper (Entrepreneurship). 322 • Information Technology R&D: Critical Trends and Issues

Table 56.—Percentage of Total Industry Sales and Employment by Size of Small Businesses

Companies with Companies with under 100 employees under 500 employees Sales Employment Sales Employment Office computing machines and computer auxiliary equipment ...... 2.6 2.6 6.0 6.3 Consumer electronics ...... 5.4 5.9 9.3 10.1 Communications equipment. . 4.7 5.2 9.9 10.8 Electronic components ...... 18.3 17.2 32.7 33.7 Computer services ...... 51.2 47.7 68.6 67.1 SOURCE: Small Business Administration, 19S4. small firms are quick to fill-niches too nar- vestments through organized venture capital row or specialized to attract large firms.23 The investment businesses in 1983 amounted to development of the computer in the 1940s and $2.8 billion, up 55 percent from $1.8 billion in the subsequent invention of the transistor led 1982.26 It is improbable that this growth rate to an explosion of new applications and to op- will be sustained for very long.27 portunities for then small companies whose The supply of U.S. venture capital financ- names are now familiar in the information ing has grown at least eightfold since the mid- technology industry: IBM, Fairchild, Intel, 1970s, helping to fund the more than 5,500 Texas Instruments, DEC, Data General, U.S. small startup and expanding firms. These Wang, and Computervision. One market re- firms tend to operate in innovative market search company’s listing of the fastest-grow- niches where the risks are high and the poten- ing niche markets shows that about 90 percent tial payoffs much higher. of the top 50 entries are in information technology,24 and many of these technologies are Information technology firms supported by relatively new. venture capital funds numbered in excess of 1,000 in 1983, and received about $2.1 billion As the successful firms in this industry from organized venture capital investment grow, they tend to generate spin-off companies businesses. These funds were used for both that often specialize initially in a narrow range startups and expansion of existing companies. of innovative products or processes. Fairchild The distribution of funds is shown in table 57. is such a company, having begun with only eight employees and now accounting for over These firms account for a significant num- 80 spin-off enterprises.25 ber of advances in the information technology field. Among the now-well known information Venture capital funding is extremely impor- technology firms started or aided by venture tant to the small entrepreneurial firms. In- capital are: Intel, Apple Computer, Visicalc, .— — ——— DEC, and Data General. R&D limited partner- 231n a July 1981 study, the General Accounting Office found that in concentrated industries, such as the information tech- ships also contribute to this process (see ch. 2). nology industry, small businesses are likely to perform specialized innovative functions and develop products or processes to The existence of so many entrepreneurial be used or marketed by other, usually larger firms in that in- firms, the current abundance of venture capi- dustry, For example, in the semiconductor industry there are tal to fund their growth, and the strengthen- many small companies that manufacture diffusion furnaces, ion implantation machines, epitaxia.1 growth systems, and mask- ing university-industry R&D relationships making systems—all part of a necessary supporb structure for serve as a powerful source of strength for in- the semiconductor industry. novation in the United States. For example, ‘421st Century Research, Supergrowth Technology U. S. A., newsletter as published in EDP News Service, Inc., Computer “Data provided by Venture Economics, Inc., from its data- Age-EDP Weekly, Feb. 28, 1983, p. 9. base covering organized venture capital investment companies. “Carl H. Vesper, Entrepreneursiu”p and National Policy, “’’Scramble for Capital at Almost-Public Companies, ”For- Heller Institute for Small Business Policy Papers, p. 27. tune June 25, 1984, p. 91. Ch. 9–Technology and Industry . 323

Table 57.—Venture Capital Funding and Information Technology Firms, 1983

Distribution of Distribution of available available Percentage of 1,000 venture capital venture capital venture capital firms (percent) (millions of dollars) Computer hardware and systems...... 28 39 819 Software and services ...... 12 7 309 Telecommunications and data communications . 9 11 231 Other electronics...... 10 10 210 Total ...... 59 61 1,569 alncl”des semiconductor fabrication and test equipment, instrumentation, fiber Optics, laser-related devices, etc SOURCE Venture Economics Inc , prowded In telephone interwew with OTA staff, May 22, 1984 over 30 new computer chip firms have started Small business may find both new opportun- operation since the late 1970s. The fastest- ities and impediments. Although small busi- growing niche markets noted earlier are heavi- nesses are currently permitted to undertake ly populated with products that are new, or joint research activities among themselves, radically different from those of only a decade the new joint venture legislation may en- ago. Few of our major competitors now have courage increased opportunities in joint ven- a comparable combination of factors to spawn tures between small and large firms. the next Silicon Valley. In fact, our West Ger- 28 This type of cooperation between large com- man, French, and Japanese trading partners are now actively seeking ways to emulate the panies and small companies has already begun to occur. For example, Control Data Corp. has environment that fosters this type of entre- 29 made its advanced computer design tools prenuership. available to two small companies, Star Tech- nologies, Inc. and ETA Systems, Inc. (the lat- Small Businesses and Joint R&D ter a CDC spin-off). IBM also has a marketing agreement with Floating Point Systems. Congress has passed, and the President If the joint venture legislation can stimulate signed, legislation that would ease the restric- transfers of marketing and technological re- tions of antitrust laws on joint R&D ventures. sources of major U.S. companies to small information technology firms, small companies ———— can realize definite benefits from the en- 28’’ The Technological Challenge: Tasks for Economic, Social, Educational, and European Policy in the Years Ahead, ” speech couragement of these new types of arrange- by Hans-Dieter Genscher, (West German) Federal Minister for ments. A potential impediment could occur if Foreign Affairs, Bonn-Bad Godesberg, Federal Republic of Ger- technology developed through R&D by large many, Dec. 13, 1983. *g”In This New Age of Entrepreneurs, We’re Number One firms is not made available to small firms Again, ” Joel Kotkin, Wdu”ngton Post, Apr. 29, 1984, p. B1. through licensing or other means.

Where the Technology is Heading

This section looks ahead to prospects for fur- tant tools for coping with increasing complex- ther advances in two building blocks of infor- ity through the use of computer-aided design mation technology-microelectronics and soft- (CAD) and computer-aided engineering (CAE) ware. As discussed later, these technologies systems, and are driving a wide range of ap- are being integrated and are providing impor- placations. 324 ● Information Technology R&D: Critical Trends and Issues

Advances in microelectronics and software continued advance of other sciences and are dependent on entirely different under- technologies, particularly physics, chem- pinnings-e.g., physics, chemistry, materials istry and materials science, and manufac- science and electrical engineering for the turing and software engineering. former, and computer and information science, ● The R&D costs for incremental advances psychology, and software engineering, among in microelectronics technology are ex- others, for the latter. Prospects for continua- pected by industry experts to increase as tion of the explosive growth of information the complexity of circuits increases and technology during the past two decades will as the theoretical physical limits of micro- be paced by advances in microelectronics and electronics are approached. software. Silicon Integrated Circuits Microelectronics30 Transistor Size and Density The rate of technological improvement in Between 1972 and 1981, the number of tran- semiconductors has been phenomenal during sistors that could be packed on a chip doubled the past quarter century, and has sparked new each year (11,000 in 1972 and 600,000 in 1981) applications and economic advances world- (see fig. 56). Today, integrated circuit technol- wide. This section reviews the growth of integrated circuit capabilities and seeks to provide insight into future progress, its depend- Figure 56.—Historical Development of the encies and foreseeable limitations, and the Complexity of Integrated Circuits likely timeframe of that progress. Probable trends include: ● Silicon is likely to remain the most important semiconductor material for the next decade and beyond, while gallium arsenide is likely to come into wide use in certain applications where its special prop-

erties provide specific advantages. The 105 impact of other innovative materials is .- likely to be marginal until after the turn > of the century, when biotechnology may begin to converge with information technology. ● Prices for the main output of microelectronics technology, logic and memory, can

be expected to continue to decline on a per 104 function basis, following the general pattern of the past two decades. ● Continued advances during the next two decades in large-scale integration of silicon and gallium arsenide will require the ——— .—.—. 30A ~i~ficmt mount of data in this section is t~en from, or based om 1) the Sixth Mountbatten Lecture,M2”c.melectmm”cs

Progress Prospects, delivered by Ian M. ROSS, President, AT&T 3 10 Bell Laboratories, Nov. 10, 1983, London, England; and 2) J. 1970 1975 1980 1985 D. Meindl, Theoretical, Practical, and Analogical b“zm”ts to Introduction year ULLU, Technical Digest of the International Electron Devices Meeting, December 1983, pp. 8-13. SOURCE: Arthur D. Little,Inc. estimatea. Ch. 9—Technology and Industry ● 325 ogy is approaching the capability of packing Figure 58.—Median Microprocessor Price v. Time a million components (mostly transistors) on (1,000 unit purchase) a single silicon chip.31 100 Along with these improvements in density, costs per function have dropped dramatically until now some memory components are less than 0.01 cent per bit, and many microprocessors are under $10 per chip (see figs. 57 and 58). This is the result of a thousand-fold de- 10 i crease in the cost of manufacturing compared with 20 years ago. The substantial increases in transistor density have been made possible by reductions in 4-Bit integrated circuit feature size, which in turn have depended on advances in photolithog- 1 raphy, the technique by which integrated circuits are manufactured. Minimum line widths have been reduced by a factor of two every 6 or 7 years. Using photolithography with visi- Uncertainty ble light, line widths have been reduced from 25 microns in 1972, or l/1,000th of an inch, to o ,10 1 1 1 1977 1982 1987 1992 1997 $lwhile other technologies (gallium arsenide, iridium ~- timonide, cryogenic superconductors,photonic devices, and PO Years tentially biotechnology) with higher switching speeds, lower SOURCE: Arthur D Little, lnc , estimates heat dissipation, or other qualities, will compete in certain special applications, current expert opinion is that silicon will about 1.5 microns currently, or less than 1/ have the major role for the next decade or more in meeting most needs for digital memory and logic. 10,000th of an inch. Figure 57.—Read-Only Semiconductor Memory Optical, or visible light, lithography, which Component Prices (per bit) by Type currently remains the leading integrated circuit manufacturing method, is expected to reach the practical limits of its capacity for small feature sizes in the range of 1.0 to 0.5 microns (the wavelength range of visible light). At the same historical rate of progress, these limits will be reached in the 1990-94 period (see fig. 59.). Most experts believe that advanced lithographic techniques using light of shorter wave- H PROM, and EPROM . lengths, in the X-ray range, and electron-beam (see fig. 60) and ion implantation machines Fixed that can directly draw lines and features on 32 . ..L— . Mask-ROM semiconductor substrates will be needed in gZ1n Cwent co~erci~ lithographic techniques, the semiconductor substrate is coated with a light-sensitive emulsion, or photoresist, and a mask, much like a stencil, is used to expose a pattern in the substrate that is then etched away with a chemical, creating lines and features. Electron beams can be focused and directed to create patterns in the emulsion without a mask, and ion beams can implant materials on the substrate without SOURCE: Arthur D. Little, Inc., estimates. the need for photoresist or etching. 326 ● Information Technology R&D: Critical Trends and Issues

Figure 59.— Minimum Linewidth for Semiconductor Microlithography 10

0.1 1977 1982 1987 1992 1997 Years SOURCE: Arthur D. Little, Inc., estimates.

atoms. The major drawback of such systems is that they are much slower than mask lithography; they are now used only in producing custom chips of low production volume. Elec- tron-beam and ion implantation technology are expected to be in use on commercial integrated circuit production lines by 199435 (see table 58). Research is under way in the United States and Japan that is striving to achieve more Photo credit: AT& T Bell Laboratories densely packed circuit components by stack- Microprocessor WE 32100 ing chips into three dimensional structures. A number of problems need to be solved to real- order to achieve further advances.33 A num- ize 3-D circuits, among them is the difficulty ber of such machines are in use for experi- of making connections with elements buried mental purposes and for production of circuits within layers of semiconductor material. Ion- to be used in-house at major companies such beam equipment is seen as critical to produc- as IBM and AT&T. ing the complex microscopic structures required for such 3-D chips.36 Electron beam lithography and ion implan- 34 —. .—.— tation may become the ultimate semiconduc- 351EEE Swtnn.n, January 1984, p. 63. tor manufacturing techniques, with minimum ‘6’’3-D Clups, ” The Econonu”st, Feb. 12, 1983, p. 84. line widths possible in the range of 0.1 to 0.01 micron. The latter widths are comparable to Table 58.—ICs Made with Lithographic Techniques feature spacing on the order of 20 to 200 (percent) —————— ~~~me en~nWr9 have begun to question the premise that Technique 1983 1985 1987 more densely packed circuits are necessary. See “Solid State, ” optical scanner...... 85 58 38 IEEE Spectrum, January 1984, p. 61. Optical stepper ...... 15 40 52 WI’he Defense Department’s Very High Speed Integrated Cir- Electron beam ...... 0 1 8 cuits (VHSIC) program supports research on electron-beam and X-ray ...... 0 1 2 ion-beam manufacturing techniques. Discussion of both of these Ion beam ...... 0 0 0 can be found in IEEE Spectrum, January 1984, pp. 61-63. SOURCE: IEEE Spectrum, January 1984, p. 80, Ch. 9–Technology and Industry ● 327

Figure 60.—A Schematic of an Electron Beam Lithography System, and its Uses to Make LSI Logic Circuits

Computer

(G)

Beam on-off control I

I I I ICI I a~e (F)...... , I analog circuits

Electrostatic deflection (p) ’ on awxor

SOURCE Mosaic, SS//MS//LVLSMJLSJ,S/, VOI 15, No. 1, p 15.

In devices with such subminiature dimen- determined by manufacturing control–the sions, other limitations become significant, ability to minimize the number of impurities e.g., the dielectric strength, electronic isola- and defects in the semiconductor substrate tion, and heat dissipation of the silicon mate- material and in finished circuits. Today’s rial-some of which may represent more se- typical chip size is about 1 square centimeter. vere limitations than the theoretical limits There are development efforts under way imposed by the most advanced manufactur- toward wafer-scale integration38 which could ing techniques. Transistors with critical di- provide a factor of 100 increase in integrated mensions of 0.1 micron have been demonstrated circuit chip size, to 100 square centimeters or to operate by Bell Telephone Laboratories. larger. This factor, coupled with component These dimensions would theoretically lead to densities of 100 million transistors per square some billion components on a square centi- centimeter made possible with the advanced meter of silicon. A more realistic, practical manufacturing techniques noted above, could limit for silicon may be 100 million compo- result in component counts of 1 billion tran- nents per square centimeter.37 sistors per integrated circuit.39 Increasing the physical dimensions of the .— chip may also contribute to increasing the SsChips me cut from 3 to 6-inch-diameter wafers which are component count per chip. Chip sizes are currently manufactured with some 100 separate integrated cir- — cuits per wafer. “ROSS, op. cit., p. 6. “ROSS, op. cit., p. 7.

38-802 0 - 85 - 22 328 ● Information Technology R&D: Critical Trends and Issues

Integrated Circuit Speed and Reliability Chip reliability has also improved. It is esti- Concurrent with, and largely the result of, mated to have increased by at least a factor of 100, while the scale of integration has in- decreasing circuit feature size and increasing 41 density, circuit switching speeds have increased by a factor of 1,000. creased significantly. A little more than 20 years ago, switching delays were crossing the Gallium Arsenide Integrated Circuits millionth of a second threshold; by 1977 the Another compound, gallium arsenide (GaAs), billionth of a second threshold had been passed; should find an increasing number of applications by 1990, delays of 1 trillionth of a second (a -12 as a semiconductor substrate because of its picosecond, or 10 second) may be possible special characteristics. These include high fre- (see fig. 61). Practical limits are foreseen, how- quency operation, low power consumption, ra- ever, that would reduce actual maximum per- diation resistance and optoelectronic and pho- formance to one-tenth that speed, or 10 pico- tonic properties. seconds .40 As well as reducing circuit delays, the con- Analog Devices tinued shrinking in the size of individual Currently, GaAs is used mainly in discrete components and the concentration of more component (non-integrated) analog devices functions in each succeeding generation of such as microwave and millimeter wave radar, integrated circuits have reduced computer sys- telecommunications and electronic warfare tem chip counts and thus the time used up in transmitters. Electronic signals at these high computers when signals travel between chips. frequencies (10 to 100 Gigahertz or billion — —.— cycles per second) can be generated at much ‘“Ibid. lower power with GaAs than is possible with conventional transmitter technologies. Inte- Figure 61 .—Trends in Device Speed—Stage Delay grated circuit, or monolithic, microwave and per Gate Circuit for Different Technologies millimeter wave GaAs circuits have recently become available. Integration promises improved performance and lower prices in military systems and private sector microwave communications (local loop bypass and other) applications. Cellular radio is expected to provide a large market for GaAs transmitters, receivers and amplifiers. Another application that offers a large potential consumer market for analog GaAs integrated circuits is direct broadcast satellite service (DBS). The principal French laboratory of N.V. Philips Co., based in The Netherlands, recently announced development of a GaAs chip that includes all of the functions necessary for the reception of DBS signals.

Optoelectronic and Laser Devices Gallium arsenide is provoking excitement in another telecommunications technology, fiber 1977 1982 1987 1992 1997 optics. It can be fashioned into tiny lasers to Years SOURCE: Arthur D. Little, Inc., estimates. “ROSS, op. cit., p. 2. Ch. 9—Technology and Industry ● 329 function as light sources for fiber optic trans- Digital Devices mission, into optoelectronic photodetectors to Interest has begun to intensify in the use receive optical signals and translate them into of GaAs as a replacement for silicon in digital electrical signals, and into signal processing logic and memory devices. High switching components for fiber optic system amplifiers speeds and low power consumption make this and repeaters (see Fiber Optics Case Study). semiconductor attractive in certain applica- Recent work at Bell Labs has developed large- tions. For instance, Cray Research has an- scale integrated (LSI) GaAs signal processing 42 nounced its intention to develop and use GaAs circuits that are scheduled for production. chips in the Cray-3 supercomputer which is Heretofore, separate devices have been re- scheduled to be available in 1986 (see Ad- quired for fiber optic light generation, recep- vanced Computer Architecture Case Study). tion and signal processing. Communication But commercial semiconductor firms have among these separate devices introduces de- thus far been reluctant to move GaAs logic lays in fiber optic transmission systems. and memory into production. Only low levels Lasers have been particularly difficult to inte- of integration (fewer devices per chip than grate with the other circuit components on silicon) and low yields (few good chips in a single chips because of their complex structure batch) are currently attainable with GaAs. Al- and the large amount of heat that is generated though 42 U.S. companies conduct laboratory by their operation. work on GaAs, only two firms have announced plans to sell digital GaAs chips.45 Fujitsu, the Japanese electronics giant, has conducted GaAs R&D funded by that govern- The Department of Defense is interested in ment as part of a $74 million, 7-year project digital GaAs because, as well as having speed among a number of Japanese companies. In and power consumption advantages, GaAs 1983, it announced development of a process circuits are highly resistant to radiation ef- that integrates a laser light source and the fects, making them attractive for military and signal processing transistors necessary for fi- space applications. Since 1975, DOD, through ber optic transmission, all on a single GaAs DARPA, has funded a total of $6 million to chip that operates at room temperature. A $7 million in R&D of digital GaAs integrated similar integrated optoelectronic photoreceiv- circuits. DARPA has announced plans to er is in the works. spend an additional $25 million to set up pilot production lines for GaAs 64K RAM and American researchers at Honeywell and 6,000 to 10,000 element gate array (program- Rockwell International have also announced mable logic) chips. The contract has been the development of integrated GaAs optical 43 awarded to a joint venture between Rockwell transmitters. The Rockwell device in particu- International, who will develop the gate array lar is reported to be adaptable to well-known chip, and Honeywell, who will design the RAM LSI circuit manufacturing techniques, promis- chip. Each of the two companies will develop ing low production costs.” production capacity for both the logic and memory chips to provide second sources for — .— .-.— —— each.46 ‘zl??~ectrom”cs, Oct. 6, 1983, p. 154. “Honeywell announced an integrated optical transmitter in 1982, but it originally required special cooling. The Honeywell device has since been improved. See C. Cohen, “Optoelectronic chip integrates laser and pair of FETs, ” Electrom”cs, June 30, “W. R. Iversen, “Pentagon Campaigns for GaAs Chips, ” 1983, pp. 89-90. See also The Econonu”st, Feb. 5, 1983, pp. 82-83. Electronics, July 28, 1983, pp. 97-98. 441, Wa r, ‘lG~s le Optical IC Integrates Laser, Drive Tra- *J. Robertson, “Say Rockwell, Honeywell Get $25M DARPA nsistors, ” Electronics, Oct. 6, 1983, pp. 51-52. IC Pact, ” Ekctrom”c News, Feb. 30, 1984, p. 4. 330 . Information Technology R&D: Critical Trends and Issues

The DARPA program has received mixed to begin to develop optical devices that func- reviews from the industry. On one hand, Rock- tion like the logic elements that make up well believes that the DARPA-funded facil- microprocessors. Hughes Aircraft Research ities will help make digital GaAs profitable for Laboratories, which is performing R&D work the entire industry, since the technology is ex- on optical logic under contract from the Na- pected to be made available to other defense tional Security Agency, has announced the de- contractors, 47 but neither of the two compa- velopment of a prototype device that is ex- nies that have announced plans to develop pected to be capable, in about 3 years, of more digital GaAs for the merchant semiconductor than 10 times the speed of current electronic market. The companies-Gigabit Logic, which logic devices. The device is to be applied in recently entered into a private R&D venture data encryption systems.51 48 to develop GaAs RAM chips, and Harris The major drawback seen with current ap- Microwave Semiconductors-submitted bids proaches to optical computing is that the logic for the DARPA contract. Both cited the elements are much larger than comparable DARPA emphasis on low power for portable electronic ones. Barring fundamental theoreti- military systems at the expense of higher cal and practical advances, especially in physics speed as the major factor in their decision to 49 and in materials sciences, optical logic systems forego bidding on the contract. This trade- will probably not see widespread use before off is seen as forfeiting the major advantage the end of the century. of GaAs in commercial competition with silicon digital memory and logic devices. The Convergence of Bio-technology and Optical Logic Information Technology Another area of applied research in which Advances in the biological sciences and gallium arsenide is playing a critical role is op- breakthroughs in genetic engineeering have tical logic. The replacement of computer proc- encouraged speculation and some preliminary essors based on electronics with ones based on research on the possibility of designing bio- photonics, which would theoretically be capa- computers. Devices constructed from bio- ble of vastly higher speeds, has been consid- chemical molecules potentially offer greater ered a possibility for some time. Such optical logic and memory densities, and the possibil- computers might be capable of trillions of ity of totally novel information processing operations per second.5o With the spread of fi- methods, including the emulation of brain ber optic communications technology, optical functioning. The first interdisciplinary scien- processing is even more attractive. The large tific conference on this subject, sponsored by capacity of light-wave transmission demands the National Science Foundation and the Uni- higher processing speeds to fully exploit the versity of California, Los Angeles, was held available fiber optic bandwidth. Conversions in October 1983.62 The meeting brought to- between optical and electronic signals intro- gether biologists, chemists, physicists, math- duce delays into current fiber optic systems. ematicians, computer scientists and electrical engineers to discuss the prospects for this Only recently, with the use of GaAs and technology and to generate questions to guide another new semiconductor material, iridium research. antimonide, has the technology been available The field is just beginning to adopt theoretical concepts on how such biochemical-based 471 bid., see also Iversen, op. cit., p. 97. machines might operate, and on the more dif- ‘a’’ Gigabit in $6.lM Ga& static RAM R&D Venture,” Elec- trom”c News, Jan. 9, 1984, p. 66. SIL. Wdler, “components for Optical Logic Stat to Click, ” 4eIversen, op. cit., pp. 97-98. Electronics, Dec. 29, 1982, pp. 31-32, 50E. Abraham, C. T. Seaton, and S. D. Smith, “The Optical sZInt,ernation~ Conference on Chemically Based Computing, Computer,” Scientific American, February 1983, pp. 85-88. Santa Monica, CA. Ch. 9—Technology and Industry • 331

ficult questions of how one might construct acteristics of software have been modified by a bio-computer and make it work in concert the uses to which computer systems are put, with existing information systems. Lewis and the meaning of software has broadened. Mayfield, head of the Chemical and Process Now, software is both the instructions that di- Engineering Division at NSF concluded, “The rect the operations of computer-based sys- consensus of the Santa Monica Conference tems, and the information content, or data, was that a great deal more work will be needed that computer systems manipulate. at the fundamental level before development The focus of software production and use is can begin. So although we will accept grant changing from concerns of computer profes- proposals in this area, we have no plans to 53 sionals to the concerns of a broad range of end organize a program to promote it. ” users of information technology -i.e., those with no particular interest in the technology, Software but only its capabilities and results. Because Software and hardware are primary ele- of this and the increasing availability of comments in information technology systems. The puter and telecommunications capabilities, hardware provides the set of generic functions future limits on the utility of information tech- that information systems may perform (input, nology will depend less on hardware capability and more on the difficulties of defining and storage, processing, communications and out- 54 put), and the software combines and organizes accommodating the needs of users. these functions to accomplish specific func- Possible implications of an expanding num- tions. The two elements are responsible for the ber of users of information technology, and a burgeoning range of capabilities of informa- shift in the focus of concern toward the needs tion technology, and, increasingly, for ad- of that expanding set of users include: vances in the technology itself. ● The relative importance of software issues Software is the part of information technol- in the research and development of infor- ogy that is most readily modifiable. A major mation technologies will likely increase. attraction of computer-based systems is that ● Small companies will continue to be im- they are programmable-machine functions portant players in software development are fixed, but their sequence may be modified because of their ability to respond to the to achieve different ends at different times or specialized needs of small segments of the to respond to changing conditions. The con- user population. tent of information systems maybe changed, ● There will be movement toward standard- expanded and contracted, within limits, to ac- ization of software for common applica- commodate the changing needs of users. Soft- tions, such as in accounting, banking, in- ware is the vehicle for providing the flexibility surance and government. of information systems to the users of the ● There will be continuing pressure to in- technology; software aids users in communi- troduce systematic engineering practices cating and managing large bodies of informa- into the production of software to lower tion and in coping with the complexity of large software development costs and to assure or specialized systems. the reliability of software in critical ap- Traditionally, the term “software” narrowly plications. (See Software Engineering referred to programs for large mainframe com- Case Study.) ● puters and was of concern only to computer There will be a demand for higher level, professionals. But as computer power has nonprocedural computer languages for spread to wider segments of society, the char- the development of new and unique appli- ✎✍ —— “An analogy can be drawn between computer users and auto “J. B. Tucker, “Pioneering Meeting Puts Biochips on the mobile drivers: all want the benefits of the technology but most Map, ” High Technology, February 1984, p. 43. are indifferent to how it works. 332 ● Information Technology R&D: Critical Trends and Issues

cations; such languages will allow the user sure to make software more useful, reliable to focus on the characteristics of the tasks and cost effective, and to make software de- to be performed and not on how the hard- velopment more productive. Not unexpectedly, ware performs them. software comprises higher and higher propor- ● The search will intensify for powerful and tions of the cost of new information systems. simplifying concepts for the organization Estimates of the relative cost of software in of information and knowledge in many large systems range to above 80 percent (see fields. fig. 62). ● Computerization will require the users of information systems to more clearly de- The Changing Role of Software in fine and organize their habits and opera- Information Systems tions related to information use. ● There will be more concern among users Computer software is traditionally classified of information technology with the uses as two general types: applications software of information and the content of infor- that is designed to apply computer power to mation systems, and less concern with how the technology works. Figure 62.—The Relative Costs of Software and The production of software has become a Hardware major industry in the United States, with estimated expenditures of $40 billion in 198255 for software products and for in-house development of programs. Sales of software in the United States doubled between 1982 and 1983 from $5 billion to more than $10 billion, and are expected to increase to more than $24 billion in 1987 (see table 59). As information technology applications pro- liferate in the office, factory and home, and in the military, computer software is required to fulfill many new and complex functions. The growing demand for software is causing pres- 1955 1970 1985 5%. Olsen, “Pathways of Choice,” Mosaic, July/August 1983, Years p. 3. SOURCE: Steve Olson, Pathways of Choice, Mosaic, July/August 1963, p. 6.

Table 59.—Software Sales by Use

1982 1983 1984 1985 (millions of dollars)a Software, total , ...... 5,001 10,309 15,017 24,677 Application programs, total ...... 1,997 3,448 5,455 11,100 Computer-aided design, manufacturing and engineering ...... 776 1,126 1,636 3,200 Other applications ...... 1,221 2,320 3,819 7,900 Systems software, total ...... 3,004 6,861 9,562 13,577 Compilers, interpreters, and assemblers . . 541 610 700 932 Data-base management systems ...... 1,100 1,430 1,888 3,500 Diagnostic and performance monitoring . . 493 645 710 Operating system ...... 870 4,176 6,264 8,% aAll figures in current U.S. dollars.

SOURCE: Electronics, Jan. 12, 1964, p. 128, Ch. 9—Technology and Industry • 333 a specific task or tasks, such as computer- divisions or government departments. The aided design of automobiles or inventory man- technology that has heretofore embodied such agement in a retail store; and systems soft- large information bases, for the most part pa- ware that is used to manage the operations of per and print, was not integrated; thus the in- information systems themselves, such as com- formation contained in those files has not been puter operating systems or database manage- accessible to single individuals. Now, rela- ment systems. tively cheap computer terminals and telephone links are making the enormous amounts of in- Applications software has come to include formation being stored in digital electronic a user interface that allows persons other than form (software) accessible by people in their computer professionals to work with the com- homes and offices.56 Through such services as puter system, while systems software is de- videotex, newspapers, periodicals and even en- signed to help computer professionals write tire libraries may eventually be made available and execute programs, make efficient use of “on-line. system components, diagnose and correct faults, and manage and audit system resources, Integration of information management on contents and usage. This distinction, though such a large scale will require sophisticated useful, is beginning to break down with the control software to help people find the infor- advent of personal computers, in which oper- mation that they want; advances in informa- ating systems are important tools to help tion science and contributions from such fields users control the workings of their machines as artificial intelligence and psychology will and perform such mundane tasks as copying be needed for this to occur. The integration of files. However, the ends to which the software information management will also have pro- is applied still remain distinct-applications found effects on the way people view and software solves user problems, systems soft- handle information, and on how organizations ware solves computer problems. conduct their business; the social sciences (sociology, political science, anthropology) will in- In general, systems software is an integral creasingly be looked to for understanding of component of the hardware because its job is the inevitable organizational and human con- to control the hardware and to schedule and flicts that will flow from the changes in infor- accommodate the development and execution mation access and use made possible by infor- of applications programs. The trend has been mation technology. for the manufacturers of hardware or specialized software vendors to write systems programs, and for more of the systems programs The Software Bottleneck to be embedded in hardware, or ROM (Read Demand for computer applications is out- Only Memory), making the end user incapable stripping the ability of the traditional large of altering them. There is also a trend toward data processing organizations (e.g., banks, in- the standardization of operating systems to surance companies, government agencies) to increase the portability of applications soft- supply the programs these organizations need ware, or the ability to use identical applica- in a timely and cost-effective manner. The lag tions programs on different machines.

The expansion of the computer user popula- ~eThe massive integration of information SYSteIXIS rtiSeS a tion is being fueled by the expanding content number of public policy issues including questions of the gov- of information systems, and is leading to an ernment role in protecting intellectual property, pn”vacy and other civil h“berties assuring eqw”ty of access to information, expanding definition of the term “software.” and the effect of information technology on the processes of Information systems are increasingly encom- government itself. These and related issues are the subject of passing and integrating many collections of two OTA assessments now in progress, “Intel.lectusl Property Rights in an Age of Electronics and Information” and “Fed- information formerly considered to be separate eral Government Information Technology: Administrative Proc- and segregated, such as the files of corporate ess and Civil Liberties. ” 334 ● Information Technology R&D: Critical Trends and Issues time between identification of an application These difficulties combine to produce the and completion of the software in the average present software development situation in data processing department is 2½ years and which time is wasted in writing essentially is increasing, probably discouraging the pro- identical parts, testing is laborious and expen- posal of many new applications.57 There area sive, and maintenance programmers spend number of factors that contribute to this “soft- most of their time trying to understand pro- ware bottleneck, ” chief among them that ap- grams rather fixing them. The difficulties have proximately 50 percent of the data processing grown out of the historical context in com- budgets are spent maintainingg old programs— puting in which hardware resources were ex- making changes in existing codes because of pensive, and the creative aspect of program- errors in the design or coding of programs, or ming was devoted mostly to the efficient use because applications requirements were inade of these resources through the clever compres- quately specified.58 sion of code. Now, with the precipitous decline Another factor that slows the development in hardware costs, progr amming productivity of new programs is that software develop- rather than code efficiency is widely recognized as the limiting factor in the use of comment, as it is practiced now, is project-ori- puting power. ented. Each new program is built up from scratch with little systematic reuse of parts The software development resources now in of older programs, even though identical func- use appear to be inadequate and too frag- tions may be implemented. The knowledge and mented to keep up with the demand for com- experience gained in one project are not sys- puting power. These resources include trained tematically preserved and transferred to other manpower and tools (software development projects. Innovations in programming tech- tools include workstations, computer-aided- niques tend to be adopted piecemeal as proj- design, coding and testing programs, and doc- ect budgets allow. As a result of this project umentation) to aid designers, programmers orientation and lack of standardization, large and managers in creating and maintaining programs that must rely on the efforts of a cost-effective, reliable software. number of programmers tend to be patch- Computer manufacturers and software de- work of pieces that reflect different program- velopers, and government, university and in- ming styles, and are often nearly incomprehen- dustry experts are recognizing the problems sible to people other than the original authors. associated with the software bottleneck and These problems are exacerbated by a turnover see the potential for sizable productivity gains rate among computer professionals estimated 59 by changing the nature of software develop- at 15 percent per year. —.. —— ment. For example, some avenues to alleviat- ‘7A recent survey of IBM mainfrarnebased data processing ing the productivity problems of software de- departments found that the average time lag for the initiation velopment and maintenance lie in the use and of applications programming projects is 2 years and this lag is growing by 3 months per year. Once under development, the reuse of standard function software modules, average application requires 8 months to complete. At the same and the development of software to produce, time, software representing 10 months of programming effort test, and maintain other software. Research is discarded yearly by the average data-processing department because of obsolescence. A certain backlog of work is desirable and development efforts in software engineer- to keep data-processing staff busy, but it is clear that many ing are seen as essential to this change (see organizations are falling behind. Application Development in Software Engineering Case Study). Practice, Tech Tran User Survey, Xephon Technology Trans- fer Ltd., 1983., p. 2. ‘“W. Rauch-Hindin, “Some Answers to the Software Prob- lems of the 1980s, ” Data Commum”cations, May 1981, p. 58. Software and Complexity See also the Case Study on Software Engineering. ‘This figure reflects a recent decline in turnover attributable Computer software is an important factor to the last recession and may well be on the rise. Also, it is in making information technologies useful in believed by experts that the greatest turnover is among those professionals writing code, so that this figure understates the a widening array of complex applications— effect of turnover on program development productivity. from space vehicles to telecommunications Ch. 9—Technology and Industry ● 335 networks to weapons systems. Software is Conclusion used to produce simulations of complex systems and events, so that scientists and engi- The microelectronic and software building neers can experiment and analyze without blocks for information technology are likely building expensive prototypes, and to control to advance dramatically in coming decades. the operations of the most complex of man’s The path of progress for microelectronics is creations, such as nuclear reactors and the reasonably clear into the 1990s as practical space shuttle. and theoretical limits are approached for silicon and as other technologies come into The need to manage a variety of increasingly use. For software, the path is less clear, but complex systems is a major source of the pres- most likely will include far more standardiza- sure to improve software capability and reli- tion, reuse of software modules, and develop- ability. For example, a large air traffic control ment of software tools to produce, test, and program can consist of more than half a mil- maintain other software. Most importantly, lion instructions that must mesh perfectly and the integration of the concepts behind ad- be essentially free of errors.60 The efforts of vances in these two building block technol- hundreds of programmers over several years ogies will shape future capabilities. may be required to produce a software system of this size. The complexity of the software Information technology capabilities and makes it virtually impossible to assure that their efficient utilization will depend upon the no errors are present in a completed program.61 integration of ideas from a variety of disciplines. For example, the acceptance and ef- Software written for critical applications ficient use of emerging computer and telecom- such as air traffic control, manned space munications resources in organizations require vehicles, nuclear reactors, weapons systems, understanding of behavioral and social vari- or electronic funds transfer, although impos- ables, as well as understanding of computer sible to guarantee as error-free when com- technology. Similarly, computer-aided-design pleted, must be constructed in such a way that requires both new software and new hardware at least the parts of programs and their inter- for the display and manipulation of complex connections are rigorously designed and ade- designs. The integration of hardware and soft- quately tested as they are built up. The com- ware in new microelectronics and software de- plexity of software development is likely to sign tools leads to further advances in infor- increase dramatically with the advent of mation technology itself. massively parallel computing systems and the construction of large knowledge-based systems. This integrative interaction constitutes a powerful technological engine which is fueling .— ~OolSen, Op. cit., p. 3. the advance of information technology in GIEvery twochoice brnch point in a program (where decision many applications—including computers for is made on what procedure will follow depending on the condi- science, engineering and business, robotics and tions that are present) doubles the number of possible paths that the program can take, so that a program with 10 branch computer-integrated manufacturing, telecom- points will have 1,024 possible paths. A program for a large munications, and artificial intelligence, among air traffic control system may have more than 39,000 branch many others. These applications in turn will points. If a computer could test a trillion paths per second (a capability far beyond toda ‘s most powerful supercomputers) influence opportunities for economic growth it would require over 1011’ 31?’ years (that’s the number one fol- and create alternatives for change in society, lowed by 11,731 zeros) to test all possible paths in such a pro- and thus will have major impacts on some of gram. See P. E. Olsen, and R. W. Adrion, M. A. Dennis Branstad, and J. C. Chemiavsky, ACM Computing Surveys, the policy issues that will become preeminent June 1982, p. 184. for legislators. Index Index

Alvey, John, 106 Department of Justice, 33, 111, 116 American Council on Education, 150 deregulation and divestiture on research, American Electronic Association, 147 effects of, 111-135 AT&T, 14, 16, 78, 111-134 antitrust laws, deregulation and divestiture, 113 Babbitt, Governor Bruce, Arizona, 172 divestiture, 116 Battelle Memorial Institute, 32 factors affecting research, 123 Bell Laboratories, 11, 14, 16, 74, 113, 120, 153 allocation of research and development British Alvey Program, 18 expenditures, 125 Brown, Charles L., president, AT&T, 128 availability of research results, 130 Buchsbaum, Solomon J., 123 basic research, 128 Bureau of Labor Statistics (BLS), 142, 146 Bell Communications Research, Inc., role of, 129 California Institute of Technology, 17 stability of earnings, 123 case studies, 9 policy implications, 131 advanced computer architecture, 9, 55 research at AT&T, management of, 120 changing computer architecture, 56 Bell Labs after divestiture, 122 computer architecture R&D, 57 modified final judgment and Bell Labs, 121 critical areas of research, 63 international efforts, 65 education and human resources for research manpower, 65 and development, 139-166 artificial intelligence, 13, 87 concern about manpower, 139 content and conduct, 96 elementary and secondary education, 154 foreign national efforts, 105 Federal role in manpower development, 157 R&D environments, 90 identifying particular manpower problems fiber optics, 12, 67 and solutions, 141 commercialization trends, 69 problems in higher education, 149 directions of U.S. research, 73 range of manpower predictions, 145 research in Japan, 74 relationship between manpower development United States R&D, 71 and economic growth, 140 software engineering, 11, 75 societal context for determining Federal content and conduct of software manpower policy, 161 engineering R&D, 79 supply of manpower, 149 international efforts in software environment for R&D in information engineering R&D, 85 technology, 25-52 software R&D environments, 76 concepts for R&D, 27 Chynoweth, Alan G., 129 roles of the participants, 28 Computer Technology Corp., 36 conflicts in perspectives, goals, and Congress: policies, 42 House Committee on Energy and Commerce, Federal Government, 28 3 Government funding, 28 House Committee on Science and industrial R&D, 34 Technology, 3 other Federal policies, 31 Joint Economic Committee, 172 antitrust policy, 33 Senate Commerce Committee, 128 industrial policy, 33 Corning Glass Works, 73 patent policy, 31 R&D limited partnership, 33 Defense Advanced Research Projects Agency tax credits, 32 (DARPA), 9, 29, 58, 93, 95, 219 technology transfer, 32 Department of commerce, 32 pattern of Government funding, 30 Department of Defense (DOD), 5, 11, 13, 20, universities’ role in R&D, 37 29, 41, 58, 75, 78, 89, 93, 279, 294, 329 measures of the health of U.S. R&D in Department of Energy, 4, 10, 20, 58 information technology, 43, 51

339 340 ● Information Technology R&D: Critical Trends and Issues

information industry profile, 43 Centre National d’Etudes U.S. patent activity, 45 Telecommunications, 254 European Community (EC), 66 Centre National de la Recherche European Economic Community (EEC), 216 Scientifique, 253 European Program for Research and L’Institut National de Recherche en Development in Information Technology, Informatique et en Automatique, 255 (ESPRIT), 18, 202, 216, 219 Ministere des Postes, Telecommunications, et Telediffusion, Federal Communications Commission, 31, 111, 253 114, 115 Industry, 259 Federal Trade Commission, 33 CII Honeywell Bull, 259 foreign information technology R&D, 201-276 CIT Alcatel, 260 conclusions, 221 Sogitec, S. A., 261 international trends, 201 Thomson CSF, 260 adapting technology, 202 political environment, 248 international technology agreements, 208 size of participation in information multinational corporations, 206 technology markets, 247 patents, 210 social environment, 252 science and engineering students, 212 university, 257 scientific and technical literature, 211 United Kingdom, 261 technology exchange agreement, 207 ESPRIT, 272 trade, 201 government, 263 U.S. computer trade, 203 British Technology Group, 269 Japan, 222 Department of Education and Science, government, 226 268 Defense Agency, 239 Department of Trade and Industry, 265 Information Technology Promotion Ministry of Defense, 268 Agency (IPA), 236 industry, 270 Japan Electronic Computer Co., 236 British Telecom, 271 Japan Development Bank, 239 General Electric Co., 271 Japan Robot Leasing Co., 236 Plessey, 271 Kokusai Denshin Denwa Ltd., 238 size of participation in information Ministry of International Trade and technology markets, 263 Industry, 228 university, 269 Ministry of Posts and U.S. information technology R&D policies, Telecommunications: Nippon Telegraph implications for, 214 and Telephone, 236 government/industry/university national R&D projects, 231 institutional arrangements, 217 Science and Technology Agency, 235 government role, 216 industry, 243 industry participation, 219 Fujitsu Ltd., 245 science and technology policy goals, 215 Hitachi Ltd., 245 France, 66, 70, 87, 246 NEC Corp., 244 French La Filiere Electronique program, 18 Oki Electric Industry Corp., 246 size of Japanese participation in information technology markets, 225 General Accounting Office (GAO), 39, 78, 155 university, 240 General Electric, 181 France, 246 Georgia Institute of Technology, 30 government, 252 goals for Federal R&D policy, 3 Agence de l’Informatique, 256 Great Britain, 66, 71, 86, 261 Centre d’Etudes des Systemes et des Greene, Judge Harold H., U.S. District Court Technologies Advances, 257 for the District of Columbia, 116 Centre Mondial Informatique et Ressource Humaine, 257 Heckler, Representative Margaret, 139 Index ● 341

IBM, 12, 36, 53 National Bureau of Standards, 37, 85 information technology R&D, nature of, 6 National Center for Education Statistics, 34 close boundary between theory and National Oceanic and Atmospheric application, 8 Administration (NOAA), 4 complexity, 8 National Science Foundation (NSF), 8, 9, 27, multidisciplinary nature, 7 29, 30, 40, 41, 58, 94, 142, 147, 159 software as technology, 7 National Science Teachers Association, 155 Inman, Admiral Bobby, 194 National Telecommunications and Information Institute of Electrical and Electronic Administration, 133 Engineers, 148 Nelson, Richard R., 129 issues and strategies, 14 new roles for universities in information changing roles of universities, 17 technology R&D, 169-195 foreign programs, 18 forces driving new relationships, 171 science policy, 19 impacts of new university arrangements, 173 scientific and technological manpower, 16 expected benefits, 174 telecommunications deregulation, impacts of, potential costs, 175 14 new roles, 176 Microelectronics Center of North Carolina, Japan, 65, 70, 74, 85, 222 184, 188 Japanese Fifth Generation Computer Project, Microelectronics & Computer Technology 13, 18, 65, 105 Corp., 193 Japan Ministry of International Trade and Microelectronics and Information Sciences Industry (MITI), 65, 75, 105, 219, 228 Center, University of Minnesota, 178 MIT Microsystems Industrial Group, 177 Lawrence Livermore National Lab, 58 Rensselaer Polytechnic Institute for legislation: Industrial Innovation, 180 Economic Recovery Tax Act, 32 Semiconductor Research Corp., 189 House Authorization Act of 1984, 41 Stanford University Center for Integrated Merrill Act of 1862, 158, 159 Systems, 182, 185 National Cooperative Research Act of 1984, North Carolina State University, 30 33 National Defense Education Act of 1958, 159 Ohio State University, 30 Public Law 96-517, 31 Sherman Antitrust Act, 113 Penzias, Arno, vice president, Bell Labs, 130 Stevenson-Wydler Technology Innovation Perot, H. Ross, 194 Act of 1980, 32, 215 principal findings, 5 Tax Equity and Fiscal Responsibility Act of 1982, 32 Rensselaer Polytechnic Institute, 30, 73, 180 Lewis, David, 148 Research Triangle Park, NC, 173, 231 Library of Congress, 133 Los Alamos National Lab, 58 Semiconductor Research Corp., 36, 120, 189 Lotus Development Corp., 202 Stanford University, 182, 185

Massachusetts Institute of Technology (MIT), technology and industry, 307 17, 177 characteristics of the U.S. information McCarthy, John, 90 industry, 316 Microelectronics Center of North Carolina, 36, employment, 317 184, 188 industry structure, 320 Microelectronics & Computer Technology international trade, 317 Corp., 36, 120, 172, 193 smd entrepreneurkd firms, 321 Microsoft, Inc., 204 composition of information technology, 308 examples of system applications, 309 National Aeronautics and Space cellular mobile radio communications, 310 Administration (NASA), 4, 10, 30, 31 computer simulation, 312 342 ● Information Technology R&D: Critical Trends and Issues

direct broadcast satellites, 310 University of Rochester, 73 financial services, 315 University of Texas, 194 heart pacemakers, 314 U.S. patent activity, 45 robotics, 311 U.S. Patent and Trademark Office, 47 importance of information technology, 307 U.S. science and technology policy, 279 where the technology is heading, 323 background, 280 convergence of biotechnology and general policies, 280 information technology, 330 science policy statements, 282 gallium arsenide integrated in circuits, 328 tenets of science and technology policy, microelectronics, 324 285 silicon integrated circuits, 324 information technology R&D policies, 286 software, 331-334 key issue areas, 290 Texas A&M University, 194 international competitiveness, 298 top 15 U.S. companies in R&D spending, 44 military/civilian balance of R&D funding, Trilogy Ltd., 33 294 Tsukuba Science City, Japan, 231 organization of government, 290 Turing, Alan, 90 Vetter, Betty M., 147 University of Arizona, 74 von Neumann, John, 90 University of Minnesota, 178 University of Rhode Island, 30 West Germany, 66, 70