SFI Bulletin Bulletin SANTA FE INSTITUTE • FALL 1999 • VOLUME 14 • NUMBER 2

SFISFI Bulletin Bulletin SANTA FE INSTITUTE • FALL 1999 • VOLUME 14 • NUMBER 2

H E T ERA RCH Y IN THE ALLEY SFI Bulletin In this issue FALL 1999 • VOLUME 14 • NUMBER 2

The Bulletin of the Santa Fe Institute is published by SFI to keep its friends and supporters informed about 2 its work. The Bulletin is free of charge and may be obtained by writing to the managing editor at the address below. PHOTO: HAWLEY JOHNSON

The Santa Fe Institute is a private, independent, multidisciplinary research and education center founded in 1984. Since its founding, SFI has devoted itself to creating a new kind of scientific research community, pursuing emerging synthesis in science. 6 Operating as a visiting institution, SFI seeks to catalyze new collaborative, multidisciplinary research; to break down the barriers between the traditional disciplines; to spread its ideas and methodologies to other institutions; and to encourage the practical application of its results.

Published by the Santa Fe Institute 1399 Hyde Park Road Santa Fe, New Mexico 87501, USA Phone (505) 984-8800 fax (505) 982-0565 home page: http://www.santafe.edu 8

Note: The SFI Bulletin may be read at the website: www.santafe.edu/sfi/publications/Bulletin/. If you would prefer to read the Bulletin on your computer rather than receive a printed version, contact Patrisia Brunello at 984-8800, Ext. 269 or [email protected]

EDITORIAL STAFF: Ginger Richardson Lesley S. King Andi Sutherland

CONTRIBUTORS: Cosma Rohilla Shalizi Janet Stites Hollis Walker

DESIGN & PRODUCTION: Patrick McFarlin

COVER ILLUSTRATION: PATRICK MCKELVEY

BASED ON PHOTO BY JULIE GRABER PHOTO: JULIE GRABER Position scan strip in 16 seps ART: HOLLY ROBERTS, COURTESY LEWALLEN CONTEMPORARY, SANTA FE SANTA CONTEMPORARY, ROBERTS, COURTESY LEWALLEN ART: HOLLY

9 FEATURES

The Organization of Diversity—Travels with Sociologist David Stark ...... 2 PHOTO: AMY SNYDER, COURTESY OF EXPLORATORIUM, SAN FRANCISCO PHOTO: AMY SNYDER, COURTESY OF EXPLORATORIUM, Discovering Patterns—The Interface Between Art and Science ...... 9 Homo Reciprocans: Political Economy and Cultural Evolution ...... 16 Pursuing Complexity ...... 21

WORK IN PROGRESS

Kevin Bacon, the Small-World, and Why It All Matters ...... center section

NEWS

SFI Artificial Stock Market Soon Online ...... 6 SFI Trustee John Powers, 1916-1999 ...... 7 DIGITAL COLLAGE: PATRICK MCKELVEY COLLAGE: PATRICK DIGITAL New Books From Oxford ...... 8 center section Oprea Earns Ph.D...... 8 New Members Named to SFI Boards, External Faculty ...... 24 Motorola’s Galvin Takes SFI Post ...... 27

INSIDE SFI

Language Issues ...... 28

21 PHOTO: JULIE GRABER

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 1 HETER ARCHY

The Organization of Diversity— Travels with Sociologist David Stark through the Uncertain World of Heterarchy PHOTO: JULIE GRABER by Janet Stites

profile

2 SANTA FE INSTITUTE BULLETIN FALL 1999 ECONOMIC SOCIOLOGIST DAVID STARK has spent the greater part of emerging markets find themselves mired in dissonance. his academic career traveling to Eastern Europe, particu- Players with varied backgrounds, motives, and even lan- larly Hungary, to gather data for his research on how post- guages, have to learn to work together. There is no socialist enterprises reorganize and recombine resources established corporate ladder per se, often no company to shape their own distinctive path to capitalism. But it’s handbook, metaphoric or otherwise. In the case of in his own backyard where the Columbia University pro- Silicon Alley, the CEO might want to go public and cash fessor and Santa Fe Institute external faculty member is out as soon as possible only to start over with a new idea, now studying the emergence of new organizational forms the director of business development wants to spend in the raw lofts and hidden cubbies of downtown money to acquire competitors, the marketing director Manhattan, where since the advent of the graphical user wants to produce award-winning design, the program- interface for the World Wide Web, an industry dubbed mer wants to write immortal code. Silicon Alley has emerged to become one of the most “Dissonance can be positive,” Stark says (which is influential centers of the digital economy. probably easier to swallow in theory than in reality). Stark, Arnold A. Saltzman Professor of Sociology and “These organizations can benefit from the active rivalry International Affairs and chair of the Department of of competing belief systems. Rivalry fosters cross-fertil- Sociology at Columbia University, has found many sim- ization.” The example he uses is the infantry officer ilarities between the organizations of Silicon Alley and who instructs drummers to disrupt the cadence of those in Hungary. “I ask the question,” says the soft- marching soldiers while they are crossing bridges, “lest spoken Stark, “are there some forms of organization that the resonance of uniformly marching feet bring calami- are more likely to be able to rede- ty,” and cause the bridge to col- fine, redeploy, recombine assets? lapse. The goal, Stark contends, is The answer: Yes, organizations to coordinate diverse identities with a capacity for reflexivity. What without suppressing differences. is the basis of that? Active rivalry of Stark sees both post-socialist coexisting principles—the organi- Hungary and Silicon Alley as social zation of diversity.” laboratories that researchers can Central to his work is a phe- use to test competing theories, nomenon he calls “heterarchy.” He because in both, people are active- defines organizations operating ly experimenting with new organi- under the principles of heterarchy zational forms. Also common to as those which operate with mini- both is the idea that their experi- mal hierarchy and which have orga- mentation is like bricolage—mak- nizational heterogeneity. There is ing do with what is available, rede- uncertainty and self-organization. ploying assets for new uses, recom- To prosper in such a situation, bining resources within and across Stark believes management organizational boundaries. becomes the art of facilitating orga- In Hungary, that might mean nizations that can perpetually reor- working suddenly without govern- ganize themselves. “The solution is to minimize hierar- ment support, with workers who need retraining, with chy,” he says. “Authority is no longer delegated verti- old equipment, or not enough supplies. In Silicon Alley, cally, but emerges laterally.” it means working without the resources of a corporate To be sure, in Silicon Alley, firms are working at structure, in offices where workspace and hierarchy are lightning speed, and traditional management models not well defined, or having to use speculative stock with academic flow charts, regular quarterly reviews and options and new media panache to compete for an oth- the like, are often disregarded if ever implemented at erwise pampered workforce. all. In some cases, it’s not even clear if successful man- But there are differences between post-socialist agers from the traditional corporate world can operate in Hungary and Silicon Alley. “The Hungarian firms knew the start-up world of the digital economy. what they were working toward; there was a model,” “I’m interested in finding out to what extent hierarchy Stark says. “They knew what capitalism was and how it versus heterarchy is there in the new kind of firm which is worked. In Silicon Alley no one knows what the market emerging,” Stark says. The benefit of heterarchy, accord- will be. They are still working in the dark. It’s like ing to Stark, is that firms become capable of learning. building a ship while out at sea.” In the case of Hungary and Silicon Alley, both Indeed, Silicon Alley hosts an eclectic group of peo-

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 3 ple scrambling to build businesses, some of which will computer programs that would be capable of adapting to develop technologies which will be obsolete before they new problems in the environment. I read it around 1990 are launched (remember “Push?”), others only to be at the height of the craze of foreign advisors making beaten to market by a competitor or monolith like recipes, blueprints, and formulas in Eastern Europe— Microsoft. Some will be acquired and then quietly and it meshed with my criticism of ‘designer capital- absorbed or disbanded altogether; others will go public ism.’” Since, Stark has been influenced by other SFI in a flurry of publicity, some successfully, others unsuc- scientists, most importantly, economist David Lane for cessfully to the point of embarrassment. his work on “complex strategy horizons,” and theoreti- As an external faculty member of SFI, Stark has cal chemist Walter Fontana. Lane, Fontana, Padgett, found himself most recently discussing the economies Powell, and Grabher will be among the participants in of Hungary and Silicon Alley under the sun of Santa Fe. an ongoing faculty seminar on “Heterarchy” that Stark Stark was first invited to SFI in 1997 by SFI external is organizing at Columbia in 1999-2000, supported by a faculty member John Padgett. Having read Stark’s work, major grant from the Andrew W. Mellon Foundation in Padgett, a political scientist at the University of cooperation with SFI. Chicago, recognized Stark’s cross-disciplinary approach In regard to his work on-site in Santa Fe, Stark to studying social transformations. Out of a cast of 20 or believes the benefits of being exposed to creative ideas so social scientists invited to the Institute in the fall of from scientists from different fields will be long-term. 1997, Stark now meets and corresponds with a working “In a certain way, my involvement at SFI mirrors my group of four, including Padgett, Walter “Woody” own view of how organizational innovation comes Powell from the University of Arizona, and Gernot about,” Stark says. Grabher, an economic geographer from the University While hesitant to shift the focus of his work in of Bonn. (In 1997, Stark and Grabher published Eastern Europe, Stark believes that what is going on in Restructuring Networks in Post-Socialism: Legacies, Linkages, New York City’s Silicon Alley is worth trading in his and Localities, Oxford University passport for a Metrocard. So signifi- Press.) cant is the new industry, he and sever- Each brings to the group similar al colleagues, including his wife data sets from vastly dissimilar anthropologist Monique Girard, are in economies. Padgett’s data is gathered the process of setting up a center to from 15th-century Florence (see SFI chart the emergence of collaborative Bulletin, Summer 1998); Powell’s organizational forms in an era of inter- from the bio-tech industry; and active media. Grabher from the advertising industry Already the group is collecting an in London’s Soho. archive of the websites of 200 of Silicon In each, the scientists studied the Alley’s most prominent firms to explore network of firms and the people oper- connections among the firms through

ating the firms. They looked to find where the networks hyperlinks. They are collecting data on the activities of GIORGIO BASARIM (1511-73) “VIEW OF FLORENCE” FRESCO crossed, where people crossed, and if the firms were firms in the industry from trade journals and business pub- lasting. What is common and of great value among lications, and developing a database of transactions across Stark’s small group of peers, is that they all have theory firms by using a search engine to look for phrases like supported by a solid data set, and each studied an econ- “merged,” “invested in,” “inked a deal.” They have spent omy where there was a fast-paced change. “All the time closely observing the operations of a number of Alley economies had a one- to two-year period of vast companies and interviewed numerous Alley entrepreneurs. upheaval, and sometimes external shock,” Stark said. But to focus on Silicon Alley is to begin to let go of “There was a high degree of uncertainty in the organi- Eastern Europe, where he had been doing research zation environment.” since 1977. “Staying in familiar territory would have Stark says he was thinking in SFI terms long before been the easier route,” Stark says. “Beginning a com- he got there. “I was looking at changes in economies, pletely new project is like launching a start-up company. not as planned or designed, but as evolving and self- It’s difficult, but exhilarating. The learning curve is organizing,” he says. While browsing the library of the steep.” Wissenschaft Zentrum in Berlin, Stark came across an Stark is currently leading a year-long Sawyer article in Daedalus on advances in computer program- Seminar entitled “Distributed Intelligence and the ming written by SFI external faculty member John Organization of Diversity” sponsored by the Holland. “It was a paper on using ‘cross fertilization’ for Department of Sociology at Columbia, with support

4 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 from the Andrew W. Mellon Foundation. The aim of this faculty seminar is to explore the emergence and ABOUT dynamics of new organizational forms in response to SILICON ALLEY the extraordinary uncertainties of rapid technological,

economic, and social change. New York's Silicon Alley started with a whimper, not a bang, on On the subject of learning and adapting, Stark the eve of the Internet revolution, when a few individuals—long likes to tell the story about a souvenir he keeps on his on foresight—saw infinite possibilities available in consumer and desk: “I have a tin can that I bought in Budapest in business applications via the World Wide Web. This, thanks to the autumn of 1989. It’s considerably smaller than the introduction of a graphical user interface that would become your standard tuna can and extremely light in weight. known as a browser. If you tap your fingernail on it, it gives a hollow ring. Initially, the core of the industry was numerous small Web But the label, complete with a universal bar code, design shops which, because of the agility small businesses announces in bold letters that, in fact, it’s not empty: afforded them, were able to scoop corporate design shops and tra- Kommunizmus Utolso Lehelete—The Last Breath of ditional advertising agencies to lock in Fortune 1000 design jobs. It became evident very quickly that there was much to do on Communism.” the Web beyond design. New York, being the media capital of the The trinket came not from a clever entrepreneur world, suddenly found itself rife with companies hoping to devel- in post-socialist Eastern Europe, but from a state- op "content" for the Web. They included iVillage.com, now a Web owned work team which took advantage of legislation site for women, initially for parenting; theglobe.com, started by that allowed employees of socialist firms to form two young Cornell University graduates, who worked tirelessly to “intra-enterprise partnerships.” Such practices were corner the market on Generation X; and FEED Magazine, the the beginning of organiza- Internet's answer to Harper's or The Atlantic Monthly. tional hedging, according Beyond design and content, companies like Internet to Stark—one toe in the Advertising Network (now DoubleClick) helped to define water. the way advertising on the Internet would work. Others “The challenge of the specialized in extending the effectiveness of the banner modern firm, whether it ad or measuring "clicks." With interest in the Internet high, more traditional software companies, previously be a post-socialist firm working in near anonymity, began to surface. Suddenly, if coping with the uncer- you weren't working in new media, you weren't anybody at tainties of system change all. Otherwise well-paid people traded in their $50-per- or a digital technologies square-foot office space, regular salaries and great bene- firm coping with unpre- fits, to work in dingy makeshift lofts, or in second-rate dictable strategy hori- buildings in the financial district which had stood empty zons,” Stark writes in his since the late ‘80s bust of the market. From the Flatiron paper, “Heterarchy: Asset District to Wall Street, a new type of company emerged on Ambiguity and the the landscape—the “start-up.” A journalist dubbed the Organization of Diversity area “Silicon Alley” and the moniker, as well as the indus- in Post-socialist Firms” try, stuck.

PHOTOS: HAWLEY JOHNSON PHOTOS: HAWLEY Now the term Silicon Alley is more metaphor than (July 1996), “is the chal- actual place, representing the growth of technology or lenge of building organizations that are capable of Internet-based companies throughout the New York City metro- learning. Flexibility requires an ability to redefine and politan area-and beyond. Companies like DoubleClick, iVillage, recombine assets: in short, a pragmatic reflexivity.” and theglobe.com have made international headlines with their Stark believes these attributes are imperative in a successful public offerings. The venture capital community and volatile, fast-moving market. “It’s the difference,” he investment banks, once skeptical, are swarming the market. Big says, “between a firm that is adaptable and one that is city that it is, one can hear people discussing their business just ‘adapted.’” plans in parks, on the subways, in coffee shops. Once almost a foreign concept to the New York area (particularly the city), small Janet Stites is a free lance writer and publisher of AlleyCat business and entrepreneurism have surfaced from the shadow of News (www.alleycatnews.com), a New York-based monthly the alleys and skyscrapers to become the darlings of economic magazine which covers the business of Silicon Alley. She has development and Wall Street—the core of the big apple. URLs: written for OMNI Magazine, Newsweek, and The www.iVillage.com New York Times. www.theglobe.com www.feedmag.com www.doubleclick.net Janet Stites

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 5 Agents, initially endowed with a certain sum sure that indicates how well the rule has SFI ARTIFICIAL of money, must decide in each time period performed in the past. Once the agent has STOCK MARKET of the simulation how to allocate their capi- chosen a specific rule to use, the rule is tal between the two assets. They do this by employed to make an investment decision. SOON ONLINE forecasting the price of the stock, and Agents determine how much stock to assessing its riskiness (measured by the buy, sell or hold, using a standard risk-aver- Recent work on the Santa Fe Institute variance of the prices). Forecasting rules sion calculation. They submit their decisions Artificial Stock Market (ASM) model by SFI are IF-THEN statements: IF (a certain mar- to the market specialist, an extra agent in graduate student Shareen Joshi, in collabo- ket state occurs) THEN (a certain forecast the market who functions as a market- ration with Reed College advisors Jeffrey is made). maker. The specialist may use a number of Parker and Mark Bedau, suggests that this Agents may recognize two different different techniques to declare a market- model may be used to understand some kinds of market states (possibly simultane- clearing price (available as parameters). interesting features of financial markets, ously): technical and fundamental. A market A genetic algorithm (GA) provides for such as the destabilizing effects of techni- state detected by an agent is “technical” if the evolution of the population of forecast- cal trading rules, and the profitability of it identifies a pattern in the past price his- ing rules over time. Whenever the GA is such rules. In short, the research suggests tory, and is fundamental if it identifies an invoked, it substitutes new forecasting that the frequent revision of forecasts, immediate over- or under-valuation of the rules for a fraction of the least-fit forecast- accompanied by high technical trading, stock. An example of a technical state ing rules in each agent’s pool of rules. A leads to higher volatility and less overall would be “the price is greater than the 50 rule’s success or “fitness” is determined wealth. period moving average,” and an example of by its accuracy and by how complex it is With public distribution of the ASM code coming soon to the SFI web site, others will be able to probe these results for themselves. The Artificial Stock Market, one of the first agent-based models of a financial market, was developed in the early 1990s by W. Brian Arthur, John Holland, Blake LeBaron, Richard Palmer, Paul Tayler, and Brandon Weber at the Santa Fe Institute. It contains a population of myopic, imperfectly rational, heterogenous agents BRANDON WEBER

who make investment deci- — sions by forecasting the future states of the market, and who also learn from their experience over time. The model illustrates how simple interactions among WEB SITE IN PROGRESS such agents may lead to the appearance of realistic market behavior. a fundamental state would be “the price is (the GA has a bias against complex rules). Results suggest that many features of over-valued by 10 percent.” New rules are created by first applying the real-world markets (such as bubbles and If the market state in a given period genetic operators of mutation and crashes in prices) that have in the past matches the descriptor of a forecasting rule, crossover to the bit strings of the more suc- been attributed to external influences may the rule is said to be activated. A number of cessful rules in the agent’s rule pool. The actually be internally generated within the an agent’s forecasting rules may be activat- GA may be compared to a real-world con- market structure itself. ed at a given time, thus giving the agent sultant. It replaces current poorly perform- many possible forecasts from which to ing rules with rules that are likely to per- THE MODEL choose. An agent decides which of the form better, much the same way as a con- The simulated market contains two active forecasts to use by choosing at ran- sultant urges her client to replace poorly assets: a risky stock, in finite supply, and a dom among the active forecasts with a prob- performing trading strategies with those risk-free bond, available in infinite supply. ability proportional to its accuracy, a mea- that are likely to be more profitable. news 6 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 Market, they goals Environment attempted to find the population of models optimal rate at which [fittest in bold] machinery traders should to decide past experience action Agent revise their reper- toire of market-fore- feedback actions casting rules using the GA. They showed that the market has only one symmetric Nash equilibrium, and that this equilibrium lay in the “Complex Regime.” Most important of all, they concluded A Typical Agent-Based Model that this symmetric Nash equilibrium It is important to note that agents in was “sub-optimal” because in this regime this model learn in two ways: First, as each the wealth accumulated by agents was rule’s accuracy varies from time period to lower than in the Rational Expectations time period, each agent preferentially uses Regime, the asset was riskier (prices were the more accurate of the rules available to less stable), and the market was noisier. her; and, second, on an evolutionary time These recent results suggest that finan- scale, the pool of rules as a whole improves cial markets can end up in situations anal- through the action of the genetic algorithm. ogous to a multi-person Prisoner’s Dilemma game in which frequent revision of fore- RESULTS casting rules can lead to increased price

The most significant early finding was variability and thus reduced overall earn- ESHBAUGH HAYES PHOTO: MARY that this market exhibits two quite different ings. When traders do not know a priori SFI TRUSTEE kinds of behavior, corresponding to differ- what other traders are doing, their optimal ent rates at which market-forecasting rules strategy is to revise forecasting rules fre- JOHN POWERS are being revised by the genetic algorithm. quently. But when this dominant strategy 1916-1999 When the GA-invocation interval is large prevails and market beliefs co-evolve rapid- (between 1,000 and 10,000) resulting in ly, the market falls into a symmetric Nash SFI Trustee John Powers, a long-time sup- forecasting rules evolving relatively slowly, equilibrium with relatively low average earn- porter of the Institute’s activities and an prices are more stable; evolved forecasting ings for traders. In other words, this work influential figure in the modern art commu- rules are simple; levels of technical trading indicates that if every trader used technical nity, died in September at the age of 83. are low; trading volumes are low; and there analysis the result would be a general loss Powers and his wife Kimiko moved from is little evidence of nonlinearity. Since this of profit. New York to live full-time in Aspen in the kind of behavior resembles the predictions Much research remains to be done in mid-1970s when Powers retired as presi- of the theory of efficient markets, this establishing the robustness of these results dent of Prentice Hall Publishing Company. regime has been termed the “Rational to variations both in the model’s parameters The couple lived first in one of the Aspen Expectations Regime.” and in the structural design of the model Institute trustee houses, and then later On the other hand, when the GA-invoca- itself. As part of this effort SFI plans to moved to Carbondale, Colorado. tion interval is small (between 10 and 100) establish a Web site for public distribution of Powers donated to the Institute many it results in forecasting rules evolving rela- the ASM code. The Web site will be designed works of contemporary art—including prints tively quickly, and the variance of the price to encourage general testing of the code, val- by Cristo, Roy Lichtenstein, and Jasper time series is relatively high; the evolved idation against empirical data, and collection Johns-—which today grace the campus. rules are complex; levels of technical trad- of detailed simulation studies. SFI Founding President George Cowan ing are high; trading volumes are higher; notes, “John’s support of SFI activities and there is strong evidence of nonlinearity. never wavered through the years. His own This regime is called the “Complex intellectual curiosity, coupled with his cre- Regime.” ative and artistic sensibilities, found a Since 1998 Joshi, Parker, and Bedau home here at SFI and at the Aspen Institute have been further studying the dynamics of where he was also active for many years. the ASM. Using a simple game theoretical We will all miss him.” model together with the Santa Fe Stock news

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 7 NEW BOOKS FROM OXFORD Dynamics of Human and Primate OPREA EARNS PH.D. SFI is pleased to announce two new books Societies: Agent-Based Modeling In April 1999, Mihaela Oprea was awarded from Oxford University Press, publisher of of Social and Spatial Processes a doctorate in computer science from the the Santa Fe Institute Studies in the Edited by Timothy A. Kohler University of New Mexico. The main thrust Sciences of Complexity (SISOC) book and George J. Gumerman of Oprea’s work focuses on the evolutionary series. Orders can be placed by calling aspects of the immune system: how anti- 1-800-451-7556. This volume presents a series of stud- body gene libraries encode information ies from archaeologists, enthnographers, about the pathogen environment of the Swarm Intelligence: primatologists, computer scientists, sociol- species; how the immune system improves From Natural to Artificial Systems ogists, and philosophers who use agent- its efficiency of generating high affinity anti- By Eric Bonabeau, Marco Dorigo, based models to examine social and spatial bodies during on-going immune responses; and Guy Theraulaz dynamics. The contributors, convened at an how mutation rates are estimated; and international conference at the Santa Fe what mechanism is responsible for somatic Social insects—ants, bees, termites, Institute in December 1997, consider a hypermutation of antibody genes. and wasps—provide us with a powerful variety of societies, from ones as apparent- Before coming to the United States, metaphor to create decentralized problem- ly simple as those of primates, to ones as Oprea earned an M.D. from the University solving systems composed of simple inter- complex as those of contemporary industri- of Medicine and Pharmacy at Timisoara, acting, and often mobile, agents. The emer- al nations. These papers ask and find pro- Romania. gent collective intelligence of social insects, visional answers to fundamental questions Her thesis is entitled “Antibody swarm intelligence, lies not in complex indi- such as: How do levels of selection and Repertoires and Pathogen Recognition: vidual capabilities but rather in networks of spatial configuration of resources interact The Role of Germline Diversity and Somatic interactions that exist among individuals and in the evolution of cooperation? How can we Mutation.” Much of her research has been between individuals and their environment. explain the evolution of inference? And done at the Santa Fe Institute in collabora- Swarm intelligence offers another way what is the role of warfare in the emer- tion with external faculty members Tom of designing “intelligent” systems, where gence of state-level societies? Other papers Kepler and Alan Perelson under the aus- autonomy, emergence, and distributed func- are concerned with understanding how set- pices of the SFI Theoretical Immunology tioning replace control, preprogramming, tlement patterns are generated among program supported in part by the Joseph P. and centralization. This book surveys sev- Mesolithic foragers in the Southern and Jeanne M. Sullivan Foundation. eral examples of swarm intelligence in Hebrides or small-scale Neolithic societies Currently Oprea is continuing her research social insects and describes how to design in the North American Southwest. with Perelson as a postdoctoral fellow at distributed algorithms, multi-agent sys- Los Alamos National Laboratory. tems, and groups of robots according to the social insect metaphor. PHOTO OF MIHAELA OPREA: JULIE GRABER news 8 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 DESIGNING THE PHOTO: DAN BARSOTTI

DISCOVERING PATTERNS—THE INTERFACE BETWEEN ART AND SCIENCE

BY HOLLIS WALKER PHOTO: PATRICK MCFARLIN PHOTO: PATRICK

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 9 These days, Jim Crutchfield is thinking about During the years Crutchfield was at UC Berkeley, vocabularies—about how humans continuously extend Kahn would call him for help on the scientific end of his our vocabularies to describe new realms of experience. constructions: “Hey, Jim, I've been squirting water into His concerns run counter to those espoused by philoso- a satellite dish, and the vortex detaches from the drain. pher Ludwig Wittgenstein early in this century, who What do you think is happening?” Often Kahn’s queries said that if you don’t know about something, you can't would send Crutchfield into his Berkeley physics lab, talk about it. Crutchfield is more worried about how our his own curiosity piqued. The artist, meanwhile, was perception of the world is limited by our vocabularies looking for a way to put a frame around an active, natur- and how we transcend those limitations. “How do you al system, then create a way for observers to alter it—in extend your vocabulary in a dynamic way? How do you effect, manufacturing curiosity to engage ordinary peo- teach yourself to see new patterns you ple with nature, science, and art. haven’t seen before?” he asks. While Kahn formalizes opportuni- This fascination with what he calls ties to perceive and effect natural “pattern discovery” is one of the rea- phenomena, Sara Roberts borrows sons Crutchfield spends more time the mathematics used to describe hanging out with artists than other sci- those phenomena and employs it in entists. Like mathematics, art can be the design of her anthropomorphic an iconic substitute for language— computer programs. Roberts teaches and at the same time a vocabulary at the California Institute of the Arts unto itself. “In a way, art is a theory in Valencia, where she also founded about the way the world looks to and now directs the Integrated Media human beings,” Mitchell Program. Feigenbaum told James Gleick, “Dynamical systems theory has author of Chaos: Making A New Science. lots of useful material for me,” “What artists have accomplished is Roberts explains. “It’s not that I’m realizing that there’s only a small interested in looking at the world amount of stuff that’s important, and through a dynamical systems filter. then seeing what it was. So they can It’s useful to me as technique.” Much do some of my research for me.” of Roberts’ work, including her 1994 Crutchfield says his long-standing Elective Affinities (based on the friendships with many Goethe novella), used artists, including Ned dynamical systems Kahn, Sara Roberts, models to drive the and Gail Wight, have “emotional engines” of inspired some of his multimedia installa- more rigorous scientific tions. In Elective inquiries. Kahn, with Affinities, the installa- whom Crutchfield tion became a worked on the 1996-97 metaphor for the San Francisco dynamics of complex Exploratorium exhibit, human relationships. Turbulent Landscapes: A married couple The Natural Forces and two close friends That Shape Our World are riding in a car (http://www.explorato- together and, as a result rium.edu/complexity), of sexual innuendoes creates art installations bandied about at a pic- inspired by atmospher- nic from which they’re PHOTOS COURTESY OF THE EXPLORATORIUM, SAN FRANCISCO PHOTOS COURTESY OF THE EXPLORATORIUM, ic physics, geology, returning, they are astronomy, and fluid thinking about betray- motion: flapping flags, ing each other. The dust devils, swirling top: “Circling Wave” by Ned Kahn, Exploratorium, San Francisco. characters are projected streams. bottom: “Encircled Stream” by Ned Kahn, Seattle, Washington video images in front

10 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 and back seats repre- led to the way she now sented by video busts uses science in her on a pedestal; the work—by questioning scenery outside the car it. “I started being very rushes away from them suspicious of my own on a screen mounted infatuation with sci- on the wall behind. ence,” she said. “I Each character runs on began to wonder, how its own computer and did we get to this place, owns its own emotional this thing we now call program and database science?” of thoughts; all four are Her question is networked. apparent in a later Occasionally, one char- installation in which acter glances at anoth- she put 50 mice in indi- er. That glance alters vidual cages; their envi- the state of the emo- ronments illustrated tional engine in the moments in the history other character according to a set of of genetics. One was a portrayal of rules. “The spectator looking at them Mendel’s pea garden. Inside the cage doesn't see the system in action, but was a miniature pea garden, which, when you walk close to each charac- eventually, the mouse ate. Another ter’s pedestal, you can hear their cage illustrated the studies from thoughts,” Roberts says. which scientists concluded twins had Another artist who has worked little in common genetically—only to with Crutchfield is Gail Wight of San later realize that their own biases pre- Francisco, who is developing a new vented them from recognizing the electronic arts program at Mills twins' shared attributes. One mouse College. Wight met Crutchfield while in the twin pair had a tiny baby grand creating a piece for Turbulent piano in his cage; he decided to sleep Landscapes on biological self-organi- inside it. The other had a shabby zation in dictostelium slime mold. upright piano; he ate the instrument. Biologists are fascinated with this Recently, Wight designed a similar species of slime mold because when a installation; this one includes five tiny PHOTOS COURTESY OF GAIL WIGHT/INSTALLATION PHOTO: BEN BLACKWELL. PHOTOS COURTESY OF GAIL WIGHT/INSTALLATION dictostelium cell is in danger of dying tableaux from history representing of starvation or thirst, it sends out “how we came to conceive of our- chemical signals to surrounding selves as electrochemical entities.” Of cells—and they aggregate by moving this exhibit, Wight said, “These tiny in synchronized waves into a slug-like tableaux are sitting inside a square creature that forms a budding stalk, Plexiglas maze, and there’s a rat that which explodes, sending spores flying lives inside of it. The rat will hopeful- to distant, and perhaps more hos- ly eat away the tableaux. The rat is pitable, environments. sort of the artist.” Wight was supposed to grow dic- In the winter of 1998-99, tostelium as part of her Crutchfield organized a public lecture Exploratorium installation, but the series in Santa Fe titled “Arts of the spores she was given were a different Artificial.” Motivated by an interest in species (physarum) that only grows in tree-like struc- how art and science will determine the structure of “vir- tures and does not shift from individual cells to a multi- tual spaces” created by networked computers, the series cellular organism—something she did not discover until included talks by Gail Wight, Roberts, art critic Dave after months of waiting for her slime cells to organize Hickey, and Rodney Brooks, director of MIT's Artificial into traveling waves. The experience caused her to look Intelligence Lab. at science differently, to question her implicit trust, and The idea of the series was that public exposure to the

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 11 thinking of those on the forefront of pattern discovery in art and science might inspire others toward similar inquiries. Putting artists and scientists together often causes each to recognize old things in new ways—that mysterious process of pattern discovery at which humans are so good. Kahn was recently working on a geologi- cal installation that represents a slice through a volcanic landscape. Air is pumped up through two sheets of glass, fluidizing a powdery mixture, erupting to the surface and creating a caldera. “When I got this working I called this geophysi- cist, Raymond Jean- Luz, at (UC) Berkeley,” Kahn recalled. “He was so into this thing he spent three hours just staring at it. He came back the next day with a graduate student and they spent all day staring at it. They were looking at something real. It reminded them of why they got interested in geology in the PHOTOS: DAN BARSOTTI first place.” Kahn says looking “Techne and Eros: Human Space and the Machine” drew participants from around the world. at real phenomena Above; student participant. Below David Dunn. prompts a different kind of thinking. “Your mind is working on a lot of lev- That’s similar to what Crutchfield experienced in els. You’re processing this visual information, and you’re the laboratory during his experiments on video feed- recognizing patterns, some so subtle you probably can't back. These were not focused so much on the rich pat- describe what you're seeing, but on some level aesthet- terns generated by that system, but on the process of his ically . . . there's an indication that there is an order in own perception of those patterns. Eventually, this led there.” him to develop a mathematical framework to describe

12 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 the process of pattern discovery. He would be looking at the video monitor, thinking, “This pattern looks familiar...similar to something else I saw a few days ago,” he recalls. “The empirical facts that I concentrated on were not images that appeared on the screen but the first intuitive impressions of regularity that occurred as I began to see new patterns.” One of the projects Crutchfield began recently involves magnetoencephalogra- phy (MEG), a new imaging technique that measures neural activ- ity via magnetic signals generated by the functioning brain—a potentially more sensitive method than the more familiar electroencephalography (EEG). A clinician typically analyzes such data by visual inspection, that is, studying temporal information recorded on strips of paper or on a screen, and recognizing certain patterns within the data. The problem is that the more sophisticated MEG machines, such as the 122-channel one at the Veterans Administration Hospital in Albuquerque, produce gigabytes of data in just a few minutes of recording from a subject, more data than a clinician could ever analyze by eye. What's needed is the ability to analyze such quantities of data for hundreds of people, over time, to identify norms and anomalies associated with illness. “Can we teach a machine to auto- SPIRAL DESIGN GENERATED BY JIM CRUTCHFIELD. PHOTOS: PATRICK MCFARLIN, GAIL WIGHT MCFARLIN, BY JIM CRUTCHFIELD. PHOTOS: PATRICK SPIRAL DESIGN GENERATED matically discover patterns in such huge quantities of data?” Crutchfield asks. In the past, human beings have been constrained by the limits of our physical world and our evolutionary heritage. Although it may not be possible for the human mind to perceive patterns in more than four or five

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 13 dimensions, once trained, machines may be able to do it for us, Crutchfield says. Such pattern-discovery machines could become our proxies in worlds we cannot visualize. A more intelligent MEG machine would not only produce massive quantities of data, but also be able to recognize patterns in that data—and then point them out to us. Can Crutchfield and his colleagues teach machines to see patterns and regularities in high-dimensional spaces, for example, to analyze those mountains of MEG data? That’s their goal. Some of these ideas will no doubt be on the agenda of a new research facility just formed in Santa Fe. What's tentatively being called The Art and Science Laboratory will involve Crutchfield and pioneer electronic com-

posers and artists including David Dunn and Steina and PHOTO:UURRAE HAYNES Woody Vasulka (founders in the 1960s of The Kitchen, an electronic art performance space in New York City). This core group—plus composer/electronic artist Morton Subotnik and composer/vocalist Joan La Barbara—pre- MURRAY GELL-MANN sented a six-week series of workshops, “Techne and AND THE CREATIVE Eros: Human Space and the Machine,” at the Santa Fe PROCESS Art Institute this summer that drew students from around the world. A permanent exploratory science-arts SFI Distinguished Professor Murray Gell-Mann uses neckties as facility in Santa Fe will offer them and others working in an easy way to talk about simplicity and complexity. “If you’re these loosely defined arenas a way to easily interact with looking at a pattern of a necktie and it’s just regimental stripes, the researchers at SFI and other institutions. it’s simple,” he said. “But you’ve seen neckties with much more New machines that can think better than we can, complex patterns.” His point is that those complex patterns communicate in languages we do not speak, in realms of have regularities that it would take a long time to describe. Gell-Mann’s interest in questions of simplicity and complex- which we cannot perceive—it sounds like science fiction. ity and their intersections with art led last fall to a forum co- In fact, these are the characteristics of cyberspace, only sponsored with SITE Santa Fe called “Simplicity and Complexity the first of the novel non-physical/non-biological realms in the Arts and the Creative Process.” The forum brought togeth- humans are creating. The inventors of these new, very er a number of scientists and artists in discussions at the Santa social spaces should include artists as well as scientists Fe Institute that culminated in a public presentation at SITE and technologists, Crutchfield says. Why? Consider the Santa Fe. Gell-Mann and his wife poet Marcia Southwick (whose innovation in magnetic materials that led to the small, latest book is A Saturday Night at The Flying Dog and Other powerful motors which drive Sony Walkmans. Now peo- Poems, Oberlin College Press, 1999) together organized the ple the world over are running, riding the subway, racing forum. Among the scientists attending were Chuck Stevens and through their ordinary lives while wearing the ubiquitous Jim Crutchfield of SFI. Arts panelists included novelist Cormac headphones attached to miniature music boxes. Science McCarthy, architect Moshe Safdie, poet David St. John, and affects technology which drives culture, and culture indi- visual artist Joseph Kosuth. Gell-Mann and Southwick also have attended recent meet- rectly determines the directions in which society chooses ings (along with others on the faculty at SFI) on connections to invest scientifically. between complexity and the arts in Abisko, Sweden and Catalina “Shift that feedback loop into the new virtual Island, California. Some of the topics explored at the meetings spaces,” Crutchfield suggests. Imagine that artists, as have included: regularities in the visual and musical arts that well as scientists, have primary input into the structure of have counterparts in human brain function; the universal appeal such new realms. He adds, “It will be an entirely differ- of poetry; and measures of effective complexity in art. ent world, one in which physical and biological con- Currently, Gell-Mann, with his assistant Marla Karmesin, is straints are markedly less dominant, and aesthetic choice trying to compile a Digital Video Disk of material from the Santa and design are primary.” Fe forum, including videotaped lectures, photographs of the art- objects shown, recordings and supplementary materials. The DVD will provide a jumping-off point for future discussions. Hollis Walker is arts and entertainment editor at The Santa Fe New Mexican. She was a Pew Charitable Trusts’ Hollis Walker National Arts Journalism Fellow in 1996-97.

14 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 SFI Bulletin

Small-World Networks

DIE-CUT art, do not print this white circle

Kevin Bacon, the Small-World, and Why It All Matters

THE KEVIN BACON GAME is a curious thing to be sure. For somebody unexpected in common. “Well, it’s a small- those who don’t know him, Kevin Bacon is an actor best world!” they exclaim. The small-world phenomenon is known for not being the star of many films. But a few a generalized version of this experience, the claim being years ago, Brett Tjaden—a computer scientist at the that even when two people do not have a friend in com- University of Virginia—catapulted Bacon to true inter- mon, only a short chain of intermediaries separates national recognition with the claim that he was them. Stanley Milgram made the first experimental somehow at the center of the movie uni- assault on the problem (confined to the United verse. This is how the game goes: States) by sending a series of traceable let- • Think of an actor or actress. ters from originating points in Kansas and • If they have ever been in a Nebraska to one of two destinations in film with Kevin Bacon, then they Boston. The letters could be sent only have a “Bacon Number” of one. to someone whom the current holder • If they have never been in a knew by first name and who was pre- film with Kevin Bacon but have sumably more likely than the holder to been in a film with somebody else know the person to whom the letter was who has, then they have a Bacon ultimately addressed. By requiring each Number of two, and so on. intermediary to report their receipt of the The claim is that no one who has letter, Milgram kept track of the letters and been in an American film, ever has a Bacon the demographic characteristics of their han- Number of greater than four. Elvis Presley, for dlers. His results indicated a median chain length of example, has a Bacon Number of two. For real enthusi- about six, thus supporting the notion of “six degrees of asts, Tjaden created a web site that provides the Bacon separation,” after which both a play and its movie adap- Number and shortest path to the great man for the most tation have since been named. obscure of choices. In fact, Tjaden later fireproofed his This result was both striking and surprising and con- claim by conducting an exhaustive survey of the tinues to be so today, because the conscious construc- Internet Movie Database, and determined that the tion of such chains of intermediaries is very difficult to highest finite Bacon Number (for any nationality) is do. Ordinarily, our perception of the social world is con- eight. This may seem nothing more than a quirky fact fined to our group of immediate acquaintances, and about an already bizarre industry, but in fact it is a par- within this group there is a great deal of redundancy; ticularly clear example of a phenomenon that increas- that is, within any one circle of acquaintances, most of ingly pervades our day-to-day existence: something them know each other. Furthermore, our average num- known as the “small-world phenomenon.” ber of acquaintances is very much less than the size of The small-world phenomenon formalizes the anec- the global population (at most thousands, compared dotal notion that “you are only ever six ‘degrees of sep- with billions). So the claim that some very short chain of aration’ away from anybody else on the planet.” Almost acquaintances exists that links us to any other person, everyone is familiar with the sensation of running into a anywhere in the world, does seem unlikely. complete stranger at a party or in some public arena and, From Small-Worlds by Duncan Watts, Princeton after a short conversation, discovering that they know University Press, 1999

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 Do we live in a small world? Under what conditions can any network, social or otherwise, be deemed “small”? And if small-world networks can be shown to exist, what does this imply about the behavior of systems that are connected up that way? These questions, among others, have been the research focus of SFI Postdoctoral Fellow Duncan Watts and his collaborators.

REGULAR SMALL WORLD RANDOM

p=0 p=1 increasing randomness

This illustrates the random rewiring procedure for interpolating between a regular ring lattice and a random network, without altering the vertices in the graph. Three realizations of this process are shown, for different values of p. For p=0 the original ring is unchanged; as p increases the graph becomes increasingly disordered until for p=1, all edges are rewired randomly. For intermediate values of p, the graph is a small-world network—highly clustered like a regular graph, yet with small characteristic path length, like a random graph.

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 CLOSE CONNECTIONS To understand the resulting networks, Watts and Networks are ubiquitous. The brain is a network of Strogatz computed two statistics. The first—the charac- neurons. Organizations are networks of people. The teristic path length—is defined as the length of the short- global economy is a network of national economies, est path (i.e., smallest number of edges) required to con- which are networks of markets, which in turn are net- nect one node to another, averaged over all pairs of nodes. works of producers and consumers. Diseases and rumors The second parameter—the clustering coefficient—mea- both transmit themselves through social networks, and sures the average probability that two nodes with a mutu- computer viruses propagate via the Internet. al “friend” will be connected, or in other words, the aver- Any kind of network can be represented abstractly age cliquishness of local neighborhoods. According to by a graph, composed of nodes (or vertices) and a set of these measures, lattice-like networks have long charac- lines, edges, joining the nodes. The nodes represent, teristic path lengths and large clustering coefficients, say, members of a population, and the edges, their inter- whereas random networks are “small” and exhibit very personal ties, business ties, friendships, etc. . . . little clustering at all. Traditionally, however, networks have been modeled as Studying intermediate kinds of networks led to the either completely ordered or completely random. In an discovery that when just a few long-range, random con- ordered network, like a crystal lattice, each node has the nections replace the local edges of a lattice-like network, same number of edges that join a small number of the characteristic path length decreases dramatically; a neighboring nodes in a tightly clustered pattern. In a “shortcut” occurs. And while the first random rewiring has random network, each node is arbitrarily connected to a great impact on path length, the clustering changes very nodes that can lie anywhere. Although ordered and ran- little. Even when the separation of elements in a network dom networks are in one sense extreme opposites, they is very small the clustering can remain almost as high as share the common feature of uniformity; that is, locally possible. This result is what Watts and Strogatz call a each network “looks” the same everywhere, and this “small-world network.” The name derives from the fact simplifies their analysis. that it exhibits the short global separations that are typi- However, most real-world networks appear to fall fied (anecdotally) by social interactions while maintaining somewhere in between the ordered and random the high degree of clustering exhibited in most social net- extremes. Friendship networks are a good example of works. this in-between state. Since people meet most new friends through existing friends, the networks are locally REAL-WORLD NETWORKS, ordered. (Here order means that if A knows B and B SMALL-WORLD NETWORKS knows C, then A is more likely to know C than some Idealized models like the one just described suggest other random element.) The outcome of local ordering that the small-world phenomenon might be common in in such a network is that one individual’s friends are sparse networks with many vertices, as even a tiny frac- more likely than not to know one another: a characteris- tion of long-range shortcuts would suffice to make the tic that is called “clustering.” Many real-world networks, world “small.” But does it arise in the real world? Watts including friendship networks, tend to be highly clus- and Strogatz set out to check this, selecting three differ- tered, but they are not entirely so. If a person joins a club ent real-world networks, for which all the data necessary and meets new people or moves to a different city to to compute characteristic path length and clustering take a job, new connections can form that are not coefficient was available. The first system was a data- ordered by the existing network. base of feature-film actors ordered by their appearance In order to simulate this kind of intermediate system in different films; the second was the electric power grid one might take an ordered network and deliberately of the Western United States; and the third was the introduce increasing amounts of randomness into it. neural network of the nematode worm C. elegans. As Watts and Steven Strogatz, Watts’ thesis advisor at Watts and Strogatz wrote in a letter in Nature, “All three Cornell University, took this approach, called “random graphs are of scientific interest. The graph of film actors rewiring” in order to explore more deeply what would is a surrogate for a social network, with the advantage of happen to the properties of initially-ordered networks. being much more easily specified. It is also akin to the For example, beginning with the graph of a 1D lattice graph of mathematical collaborations centered, tradi- (simply a ring of nodes, each connected only to its near- tionally, on P. Erdos. The graph of the power grid is rel- est neighbors within a specified radius) they began evant to the efficiency and robustness of power net- replacing near-neighbor edges with edges to randomly works, and C. elegans is the sole example of a complete- selected nodes chosen uniformly throughout the net- ly mapped neural network.” work. Each of the three graphs turned out to be a small-

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 6º DIGITAL COLLAGE: PATRICK MCKELVEY COLLAGE: PATRICK DIGITAL

Empirical examples of small-world networks

L actual L random Cactual Crandom Film actors 3.65 2.99 0.79 0.00027 Power grid 18.7 12.4 0.080 0.005 C. elegans 2.65 2.25 0.28 0.05

Characteristic path length L and clustering coefficient C for three real networks, compared to random graphs with the same number of vertices (n) and average number of edges per vertex (k). All three networks show

the small-world phenomenon L > L random but C >>Crandom.

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 world network. The scientists were careful to note that After one unit of time, the infective is “removed” these examples were not handpicked. They were cho- (either because it dies or becomes immune) but in that sen for their inherent scientific interest and complete- interval it can infect (with some probability) each of its ness of available data. What this research suggests is that healthy neighbors. The process is then repeated until it the small-world phenomenon may be common for many reaches a steady state. large networks found in nature; it is not merely an arti- Their findings show three distinct regimes of behav- fact of an idealized model. ior. In the first (for diseases with low infectiousness), the disease infects little of the population before dying out. NETWORK STRUCTURE AND THE In the second, a highly infectious disease infects the BEHAVIOR OF DYNAMICAL SYSTEMS entire population regardless of its connective topology, An obvious question that arises from these conclu- but the time taken to reach this steady state varies dra- sions is: What impact might this phenomenon have on matically as a function of characteristic path length of the the dynamical behavior of a distributed system? Say network. (Shorter path length implies faster spreading of we’re looking at social structure: What is its role in gen- the disease.) For intermediate levels of infectiousness, erating globally observable, dynamical features in a there is some complicated relationship between struc- social system? In general, if a set of relatively small ture and dynamics, which has not yet been completely changes to the edge set of a graph can have a dramatic characterized. Nevertheless, there is a clear correlation impact on its global structural properties, might the between critical infectiousness—the point at which the same changes affect the behavior of dynamical systems disease infects a macroscopic fraction of the popula- that are coupled according to such a graph? tion—and the amount of randomness in the network. The question is far from straight forward. Even in Beyond those conclusions, not much more can be said. idealized systems whose behavior offers a relatively However, it is clear that for this dynamical system the clear interpretation, the relationship between structure attractor for the global dynamics does depend on the and dynamics builds on more than one factor. One must coupling topology. consider network structure, as well as a whole literature In epidemiological terms, small-world networks of distributed dynamical systems, which again tradition- imply that the level of infectiousness required for a dis- ally assumes that the relevant coupling topology is ease to grow to epidemic proportions can be highly sen- either completely ordered or completely random. sitive to the connective topology of the population. This However, Watts and Strogatz’s work suggests that real- may change our way of looking at social diseases, which world networks may combine significant elements of are often perceived as confined to isolated subgroups of order and randomness with resultant properties (like a population. The highly clustered nature of small- small-world connectivity), that cannot be captured by world graphs can lead one to believe that a given disease either traditional approach. Thus a realization about is “far away” when in fact it is very close. In other words, network structure suggests a new question for distrib- when looked at on a local level, the change in structure uted dynamical systems: Do significant new dynamical that causes the disease to spread much further and faster phenomena emerge when the corresponding network is may not be observable by an individual who has access a small world? only to local information.

SPREAD OF INFECTIOUS DISEASE GAMES ON GRAPHS A very simple kind of distributed dynamical system While the spread of disease presents a relatively sim- is that of a disease spreading from a small seed of initia- ple dynamical system, the question of cooperative tors into a much larger population, whose structure is behavior emerging among competitive agents playing a described by some underlying graph. Typically, work on many-player game is significantly more complicated. the spread of diseases focuses on populations in which Since disease is involuntary and mechanical, it is only on complete mixing is assumed between elements. With the edge of truly social behavior. However, human this assumption, subsequent analysis of population behavior as measured through games can bring up more structure can be ignored and only the relevant sizes of complex behavior. healthy, infected, and immune populations along with One such game is the Prisoner’s Dilemma, which the rate of infectiousness need to be known. models a situation of two partners in crime who are cap- Watts and Strogatz took a different approach, simu- tured and locked in separate cells between which they lating the spread of an infectious disease on a simple are unable to communicate. The dilemma concerns small-world network model. At time t=0 a single infec- whether each prisoner ought to “cooperate” by remain- tive is introduced into an otherwise healthy population. ing silent or “defect” by selling out their partner in

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 order to reap a reward. This game acts CDnumber of shortcuts increases, coopera- as a model for many kinds of interaction is increasingly difficult to sustain. tions (from common pool resource C (3,3) (0,5) This is due to the fact that cooperation problems to international arms negotia- D (5,0) (1,1) in the Prisoner’s Dilemma context tions). The main focus is that essential- requires that potential cooperators ly competitive agents need to over- PRISONER’S DILEMMA interact with each other preferentially. come the temptation to exploit each The conundrum of the Prisoner’s In this case, in a highly clustered graph, other if they are to reap the rewards of Dilemma is captured in the chart cooperators located in the same cluster reciprocal cooperation. above. Defection (D) yields either 5 or can survive, even thrive, even within a Some early and influential research 1, and cooperation yields either 3 or noncooperative majority. However, the 0. The rational decision is to defect. If on the subject was done by Robert introduction of even a few shortcuts both prisoners are rational, since they Axelrod at the University of Michigan, both have the same information, both can weaken the cooperators’ position. who devised a computer tournament in defect, yielding a payoff of 1 each. If Those who believe in altruism will be which players (computer programs) both prisoners cooperate—if both happy to know that if the greater popu- using different strategies competed remain silent-—both will earn a 3, a lation is sufficiently predisposed to against each other. Axelrod found that considerably better outcome than that cooperate in the first place, then coop- the strategies that performed best did generated by the rational action. eration will spread much faster in a so not by exploiting the weaknesses of small world than in a large one. others, but by eliciting cooperation, thus allowing both For those seeking to optimize both the spread and sides to do well. A strategy known as “Tit-for-Tat” sustenance of cooperation, the primary task is striking a devised by Anatol Rapoport (on the basis of his obser- balance between high clustering and speedy transmis- vations of human behavior two decades earlier) emerged sion. Since small-world topology exhibits both these as the most effective. Its rules are simple: “Cooperate at features it may be truly useful to such an endeavor. first and then, on succeeding turns, do whatever the However, the matter is not straightforward, even for this other player did on the previous turn.” The trademarks model, because for any given constant population struc- of TFT—being “nice,” “forgiving,” “retaliatory” and ture, any change in the disposition toward cooperation “transparent”—emerged as generic markers for any (hardness) can tip the balance from rapid growth to strategy to be successful in a sufficiently heterogeneous rapid decline. Nevertheless, in the design of organiza- environment. tional structures, the idea of optimizing between clus- In much of the early work on the Prisoner’s tering and length may be a useful concept. Dilemma, the population is treated as essentially struc- tureless, a reasonable starting assumption, but not very NEXT STEPS realistic for large social systems. Lately, interest has The work by Watts and Strogatz suggests that dis- emerged into the evolution of cooperation in the pres- tributed systems can exhibit dramatically different ence of population structure, but generally this work has behavior within the structural context of small-world focused on either one- or two-dimensional space. Watts networks. In fact, the research has set off a small and Strogatz, interested in this question, adopted a gen- avalanche among researchers in both the natural and eralized version of Tit-for-Tat that could be used by social sciences to explore the implications of the small- players located on an arbitrary graph. In their version world phenomenon. At the Institute, Watts and SFI the formulation was entirely local; players could only Research Professor Mark Newman have been examin- react to the conditions within their immediate social and ing the scaling properties, phase transitions, and site temporal neighborhoods. percolation properties of small-world graphs. Another Similar to the case with the spread of disease, Watts aspect of the phenomenon will be studied in a working and Strogatz found three regimes of behavior in their group organized by Charles Sabel on using the intu- simulation. For large h, where now the variable para- itions from small-world dynamics to increase the effec- meter is the average hardness (h) of players—the “hard- tiveness of organizations solving complex problems in er” a player is, the more reluctant it is to cooperate. rapidly changing environments. As Watts notes in one of Regardless of population structure, cooperation always his early papers on the subject, the notion of small- dies out; for sufficiently low h, cooperation always dom- world connectivity “may have implications in fields as inates, but the timescale depends quite sensitively on diverse as public health, organizational behavior, and the fraction of shortcuts; and for intermediate h, how design.” The work has just begun. much cooperation succeeds depends on the fraction of shortcuts. What became clear throughout is that as the

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 PHOTO: BENN MITCHELL/IMAGE BANK Homo Reciprocans: Political Economy and Cultural Evolution By Cosma Rohilla Shalizi

Position scan strip in seps ART: HOLLY ROBERTS, “OLD DOG” OIL ON SILVER PRINT ON PANEL, COURTESY LEWALLEN CONTEMPORARY, SANTA FE SANTA CONTEMPORARY, COURTESY LEWALLEN PRINT ON PANEL, ROBERTS, “OLD DOG” OIL ON SILVER ART: HOLLY

16 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 English-speaking social science, espe- developments in the social sciences But consider the results of the “ulti- cially economics, is dominated by a in recent years is the emergence of matum game,” with two players, an tradition going back to Adam Smith someone to take his place—called experiment that has been carried out and the other late 18th- and early Homo reciprocans by Samuel in over one hundred studies in twenty 19th-century British political econo- Bowles, a professor of economics at countries with highly consistent mists and historians of civil society. the University of Massachusetts at results. The experimenter picks a It focuses on individuals, and sees Amherst and a member of SFI’s coor- player at random, hands him a wad of their acts and choices as primary. dinating committee for its Keck cash, and tells him to divide it Larger entities—markets, states, Foundation evolutionary dynamics between himself and the other player. institutions, cultures, and classes— program. Bowles described “recipro- The second player can either accept are shorthand ways of speaking cans” during his talk “Social the offer, in which case they split the about patterns in the acts of many Organization and the Evolution of pot as agreed, or reject, in which individuals. Norms” at SFI’s May 1999 Science case both get nothing. But Homo Partly because they lend them- Board Symposium which focused on economicus, playing against another selves to precise, mathematical “Humans and Other Social Animals.” economicus, offers only one cent, expression, individualist theories Perhaps the most striking way to which is accepted. Most people offer have proven theoretically insightful, introduce this character is with some between forty and fifty percent, and practically useful, and surprisingly results from experimental economics. routinely reject offers of less than a powerful. They are also basically Take public goods games: the experi- third, even in one-shot games (where unrealistic. The standard individual menter gives his subjects some there’s no chance for retaliation), economic agent, Homo economicus, money and explains that they can even when the pot amounts to sever- has been called a “hedonistic choose, separately, how much to al months’ earnings. That people sociopath.” He also has no culture at keep and how much to contribute to a make large offers is striking enough, all, and is far too smart. Nobody, not common pool, which will be, say, dou- but what really rules out Homo eco- even exponents of “rational choice” bled, to pay for a benefit in which all nomicus is that people reject quite theories, is much like Homo econom- will share equally. The payoffs are substantial offers in order to punish icus, which is good for humanity, but such that contributing nothing maxi- others for not cooperating, even when bad for those theories. mizes one’s individual gains. In such it costs a lot to do so. There is another social science “collective action” situations, Homo This suggests a very different tradition, going back to Herder, economicus contributes nothing, and view of what economic agents are Hegel, and other German contempo- hopes to exploit everyone else. In actually like, and thus emerges raries of Smith, which evades these real-life experiments, however, few, Homo reciprocans. As Bowles puts it problems by focusing on collective often less than a half of the subjects, in an essay with his long-time collab- entities like cultures and classes: start out doing this. When given the orator and fellow U-Mass economist these entities, proponents say, are opportunity, experimental subjects Herbert Gintis: “Homop=1 reciprocans real, and they do things to people; can be surprisingly determined to comes to new social situations with indeed, they shape the people who punish those who cheated them, a propensity to cooperate and share, belong to them in fundamental ways. even at considerable cost to them- responds to cooperative behavior by This isn’t much of an alternative, selves. More surprisingly, this is true maintaining or increasing his level of however, because it amounts to say- even on the last round of the game, cooperation, and responds to self- ing that cultural effects are produced when they couldn’t hope that punish- ish, free-riding behavior on the part by—culture. This is like saying that ing the cheaters now would change of others by retaliating against the opium puts people to sleep because their behavior in the future. Homo offenders, even at a cost to himself, it possesses a “dormative virtue.” economicus, by contrast, realizes and even when he could not reason- This is a dilemma for social scien- that punishing cheaters under these ably expect future personal gains tists: do they invoke incredible crea- conditions is, like contributing to the from such retaliation.” This is cer- tures like Homo economicus, or vac- common pool, a pure waste of tainly in line with empirical observa- uous entities that don’t really explain money, and so refrains from doing so. tions: people do produce public anything? H. L. Mencken once defined con- goods, they do observe normative Though most find Homo econom- science as “the inner voice that restraints on the pursuit of self-inter- icus an implausible caricature of warns us somebody is looking,” and est (even when there is nobody human behavior, for want of any social scientists and biologists have watching), and they will put them- replacement, he has had to do. often interpreted apparently gener- selves to a lot of trouble to hurt rule- However, one of the most exciting ous acts as self-interest in disguise. breakers.

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 17 PHOTO: BENN MITCHELL/IMAGE BANK Besides the fact that typically by the experimental results: if, for their fitness was reduced by uphold- there have been no other alterna- instance, you get to be the proposer ing norms, so long as reciprocators tives, another objection to abandon- in the ultimatum game by passing were more likely to find themselves ing Homo economicus is that he is, some test, even a trivial one, you with other reciprocators in well- in his own way, a reliable standard. offer less, and the other player ordered groups capable of surviving There is just one way of being a accepts less.) Even when norms bouts of material scarcity and hyper-intelligent hedonistic allow for inequality, they still enjoin attacks that norm-flouting groups sociopath, but there are at least as some reciprocity; the players accept could not endure. many ways of being someone mostly mutual, if not equal, obligations. Since norms differ from place to inclined to follow norms as there are Readers familiar with evolution- place and time to time (sometimes norms to follow. A single, definite, ary psychology will remember the even within a single society at a sin- unambiguous prediction is, for some, elegant experiments of Leda gle time), we would like a theory that superior to an endless series of Cosmides and John Tooby on “cogni- explains what determines differ- “maybe” and “it could be this tive adaptations for social ences in norms. Such a theory would norm...on the other hand it could be exchange.” They showed that most also help close a significant gap in that one” predictions. There are two people can solve certain kinds of current individualist models. They ways out of this, and Bowles, char- logic puzzles when the problem is regard agents’ preferences as acteristically, takes both. phrased as one of detecting people “exogenous,” as fixed, given—and First, which norms a given group breaking rules, even if those same inexplicable. The question is how to of people follow is a factual ques- people cannot solve formally identi- make them “endogenous,” to bring tion, which can be investigated. For cal problems which are presented them within our models? example, to see whether the results abstractly or with different subject The acquisition of norms and of the ultimatum game are uniform matter. This suggests that the capa- preferences (and other bits of cul- across very different types of soci- bility for reciprocans-type behavior is ture) is an abiding concern for eties, Bowles, along with anthropolo- something very deeply wired into our Bowles; his first book with Gintis, on gist Robert Boyd, experimentalist brains. the role of public education in mod- Ernst Fehr, and Gintis, have orga- This only makes more pressing ern America, was, precisely, a study nized field experiments of the ultima- the question, which will have already of how people are acculturated, and tum and public goods games in a occurred to readers familiar with a study of the ways in which what is dozen simple societies around the sociobiology, of how (if at all) Homo learned is affected by economic cir- world, including some, like the reciprocans can evolve and sustain cumstances. Acculturation is an indi- Machiguenga in Amazonian Peru, itself in a population which contains vidual-level process: we get it not with very limited exposure to mar- some exploiters. In the presence of from our culture, but from parents, kets and other modern institutions. such exploiters, natural selection will siblings, other relatives, neighbors, (No noble savages have turned up so tend to eliminate organisms which playmates, colleagues, and, of far: the Machiguenga, in fact, are the engage in unprofitable behaviors, course, teachers. Sometimes, as in closest approximations to Homo eco- such as helping others or engaging in formal schooling in developed coun- nomicus yet discovered.) costly punishments. One of the stan- tries, this is a very deliberate Second, norms do not vary in dard theories of the evolution of process, and people are taught cer- arbitrary and indefinite ways; there cooperation between relatives tain skills, beliefs, norms and prefer- are certain patterns which appear to evades this by postulating “assorta- ences, because these are economi- be common across societies. In tive” interactions between kin—if cally useful. In other cases, there is experimental games, subjects organisms tend to interact with their a mutual influence between social explain their acts by saying that self- close relatives, which carry many of and economic organization and cul- seeking behavior would not be “fair.” the same genes, then altruistic ture: people are apt to imitate those Fairness need not mean equality, but behaviors can establish themselves, who achieve social success. inequality does have to be justified even in populations which contain This kind of “replicator dynamic” somehow. They must have reasons many exploiters. Applying such rea- is actually easier to model than is for it; a person may be rewarded for soning to non-kin, one of Bowles and deliberate instruction, using exten- skill or effort; for virtue; or (a sur- Gintis’ most recent papers shows sions of models from population biol- prisingly common move) because that Homo reciprocans could have ogy originally developed by Marcus they are more than human, at the evolved during the Paleolithic era in a Feldman with Luca Cavalli-Sforza, as very least a different and much bet- similar way: reciprocators could have well as the related evolutionary mod- ter kind of human. (This is borne out proliferated in a population, even if els of Robert Boyd and Peter

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 19 Richerson. Individual rewards in such can’t have Bill Gates assassinated), markets, intergroup bargaining, models are affected both by material but these have very little culture-spe- firms, and other economic organiza- circumstances and by economic and cific content, and continual efforts tions—shape the evolution of individ- social organization—“who meets are made to remove any lingering ual preferences, and in turn how whom, to undertake which joint activ- specificity. It’s not so much that mar- these preferences shape the evolu- ities, with what payoffs and opportu- kets make people into Homo eco- tion, and in particular, the emer- nities to acquire new traits,” as nomicus, but that they present situa- gence of new economic organization. Bowles puts it. This structure of tions which evoke behavior that As part of the Institute’s continuing social interactions, Bowles shows, resembles his, and reward it. (In program on evolutionary dynamics, controls the rate and direction of the experimental Prisoner’s Dilemma Bowles is planning subsequent work- differential replication of tastes, games, subjects tend to cooperate if shops including one on the role of habits and norms, whether the repli- the game is called “Community” and group formation, the distinction cation is genetic, cultural, or some defect if it’s “Wall Street.”) between insiders and outsiders, and combination of both. As Adam Smith knew, the institu- group extinctions in evolutionary What happens when material cir- tions of the market can only work if processes. cumstances and social structure many people (e.g., change? In particular, what happens police, judges, par- when those who follow previously ents, soldiers) do accepted norms are liable to fail, not, in the line of whereas those who break them, who duty, act like Homo adhere to some other set of norms, economicus at all, Position may be seen to prosper? Such a sit- but instead act uation is unsustainable; ultimately, more like Homo rec- scan either the norms have to change, or iprocans. Balancing the material situation does. Because two different, most people are merely Homo recip- incompatible sets strip in rocans, and not “rebels in defense of of norms—one for tradition,” it’s safe to bet on the the marketplace, seps norms changing, but this can be a the other for the lengthy process. One very common home, and for rela- reaction, for instance, to introducing tions with friends markets, trade, and a money-econo- and workmates—is not an easy task, Bowles’ papers are available at my into societies which previously and there is a natural tendency for http://www-unix.oit.umass.edu/~bowles didn’t have such institutions is to the balance to tilt in one direction or produce fringe groups of people who the other, for the domain governed are “rootless”—cut off from the by one set of conventions to grow at older social groups, much more the expense of the other’s. Nobody given to the pursuit of individual self- really knows whether this will happen interest. (Bowles’ interest in cultural in our case, or whether we’ll contin- evolution was awakened by witness- ue our uneasy impersonation of ing precisely such a sequence of Homo economicus, or even whether events as a teacher in a remote part we’ll hit upon new rules. Bowles of Nigeria in the early 1960s: “What quite openly hopes for new rules happened in two hundred years of more conducive to humans flourish- European history unfolded before my ing throughout the planet. eyes in the course of a couple of Bowles will coordinate a work- years.”) shop at the Institute next January. In highly marketized societies, “Coevolution of Institutions and careful, intelligent, competitive maxi- Preferences” will bring together mizers of personal gain, unfettered economists and other social scien- by sentiment or scruple, can do very tists to discuss the dynamics of insti- well for themselves. Certain restric- tutional and individual behavioral tions on competition are enforced evolution. The goal is to understand (CEOs of other software companies how social interactions—defined by

20 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 PURSUING COMPLEXI- PHOTOS: JULIE GRABER

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 21 The undergraduate interns toiling in SFI’s offices this Brink is working with Resident Research Professors summer face an interesting challenge. Most of the insti- James Crutchfield and Cris Moore, using classical infer- tutions from which they come offer only limited course ence techniques to attempt to give/reproduce the struc- work in the area of complexity studies, and yet the stu- ture of data produced by a quantum source. dents are hungry for knowledge of this approach. These Meanwhile, Jessica Kleiss expressed a similar experi- undergraduates have chosen to explore this new science ence at her home institution. “At MIT I would say there out of their own initiative and curiosity, rather than as are a lot of studies pertaining to complexity, but they’re part of an educational track laid before them. in small, hidden pockets,” said this junior majoring in So, what will they gain? mathematics and atmospheric sciences with a concen- “We’ll be prophets,” said Russ Tedrake, jokingly, tration in physics. “You can’t minor in complexity stud- when the undergraduates convened for a roundtable ies, but both the Media Lab and the Artificial discussion on an SFI patio overlooking Santa Fe. Russ is Intelligence Lab deal a lot with complexity issues, and one of seven interns who is spending 10 weeks partici- the math department does a good deal of nonlinear pating in this National Science Foundation supported studies.” annual Research Experience for Undergraduates (REU) While at SFI, Jessica’s mentor is SFI External program. Each undergraduate is matched with one or Faculty Member Alfred Hubler from the University of more mentors with whom they work on a Illinois. Their project involves studying research project; often these mentorships the motions of a spring-mass system result in continuing collaborations. being fed through a viscous medium, Russ tempered his “prophet” joke by like oil. They have found different states adding, “There are in fact quite a few at of motion, like waves, loops and kinks in Michigan who have a broader perspec- the springs, and they’re using the results tive. I’ll be one of them.” Russ’s experi- PURSUING to gain insight about fluid dynamics. ence in his undergraduate work has been Next, Amy Nelson told the group how different from many SFI undergrad exposure to complexity science has uni- interns. At the University of Michigan he fied her interests. She comes from benefits from the presence of such schol- Stanford University, where she’s a junior ars as John Holland and Rick Riolo. “But majoring in philosophy with a concentra- all the complexity courses are graduate COMPLEXI- tion in metaphysics and epistemology. courses,” he said, which means that even Since she encounters little exposure to at an institution where complexity is a complexity science at Stanford, she’s common word, courses are not immedi- eager to explore the approach during this ately available to undergraduates. For summer experience. “I’m identifying this reason his experience at SFI has TY patterns in areas I’m interested in and been rich. “It’s opening my eyes,” he finding relationships that I might not said. have seen otherwise,” she said. While at the Institute, Russ is work- Amy is also working with Tim Hely ing with SFI Postdoctoral Fellow Tim Hely attempting investigating the role of the synchronicity of neuronal to further the results from a recent paper by Petr firing rates in the perceptual task of feature binding. Marsalek, Christof Koch, and John Maunsell on the The project investigates the strategy of temporal bind- relationship between synaptic input and spike output ing in the selection and organization of information from jitter in individual neurons. highly parallel and distributed functional areas in the Looking enthusiastically around at the group, James cortex. Brink, a sophomore majoring in cognitive science, com- The group sat silent for a moment as if pondering puter science, and mathematics at Indiana University, Amy’s notion of unifying interests. Then Matt Bell, a said his experience at his home institution has involved sophomore at Stanford who is majoring in computer sci- a considerable amount of complexity science. “A lot of ence and psychology, offered an interesting perspective issues we deal with here at SFI are touched on by peo- on why complexity science isn’t popular in computer ple at IU. But there’s not a complex systems department science labs at his home institution. “For better or for there. A few of the undergrad cognitive science pro- worse, the computer science department at Stanford is grams discuss issues being discussed here, so I guess closely tied to Silicon Valley. So most researchers are you could say it’s not a unified program but is scattered very concerned about applications. Consequently, around different departments.” there’s little interest in research that does not have

22 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 financial value. However, there is one course in genetic concrete and the abstract. Many commented on the algorithms taught by John Koza.” schedule at the Institute, where work goes on day and While at SFI, Matt is working with SFI Research night. “You might come in at one a.m. and find all these Professor Walter Fontana examining the evolvability of people here talking,” said Amy. RNA structures, both in terms of understanding the They were somewhat surprised that everyone eats evolution of RNA itself and as a framework for looking lunch together. This allows the interns to converse with at evolutionary theory in general. “major scientists.” They are encouraged by this kind of Next, Hunter Fraser told the group he’d heard about freedom at the Institute. “You can even show up at SFI from his father and then followed up with his own someone’s door and ask what they’re working on,” said research. “When I told people in the Biology Amy. “I’m struck by how people communicate across Department at MIT that I was coming to SFI, they disciplines here. I’ve never seen people working and thought I was weird. They thought I was going to study talking together to that extent.” math for the summer,” said this sophomore, whose This comment led to discussion about where the SFI interests include biological networks, the origin of life, experience might lead them. James said his time at the and immunology. Institute will help him explore new areas when he returns His mentor is SFI Visitor Charles Sidman from the to IU. “From this experience, I’ll be inclined to talk to University of Cincinnati. They people from different projects.” are exploring epigenetic interac- Meanwhile, Russ introduced tions and how they may cause the the question of getting too broad genome to act as a complex non- too fast. “I have a professor who linear system in determining an would never have gotten broad organism’s phenotype, instead of until he got a grounding in one the traditional “one gene-one field. So while I’m here I try to trait” idea of genetics. listen to what people say and Like Hunter, Roger Turner, a take away what relates to me, to sophomore majoring in the histo- what I’m interested in.” ry of science at Brown University, Jessica countered, “Some say found SFI on his own, rather than in grad school you’ll learn almost through his home institution, and everything about almost nothing he’s impressed with what he’s or almost nothing about almost experiencing. “It’s exciting to see everything.” the nature of the place—all these Most of the group chuckled people who majored in one field, at this, while some nodded their got their masters in another, and heads. their Ph.D. in another,” he said. “But the undergrad years “Things have become so splin- are the best years to explore,” tered in academia. People have clockwise beginning in the upper left: Hunter Fraser, said Matt. “It’s then that you one small thing they know about. James Brink, Amy Nelson, Jessica Kleiss, Roger have a chance to get a general But here we learn to see patterns. Turner, Matt Bell, and Russ Tedrake. foundation.” We can study them and apply Amy, the philosophy major, them to different systems.” threw in a joke, “What I’ve been exposed to here has His mentor is SFI Visitor Douglas White from the heightened my prospects beyond bartending when I University of California at Irvine. Roger is assisting with graduate.” the Institute’s secondary school computer modeling Everyone laughed at that one. workshop. This workshop introduced high school stu- All of the undergraduates agreed their experience at dents and teachers to the concepts of computer model- SFI would expand their ways of thinking and help them ing using the Starlogo programming environment. He’s design their own tracks into the future. Said Roger, also been studying how philosophical and ideological “There’s a wealth of interesting ideas flowing around critiques of science, particularly feminist critiques, can here.” suggest improvements in science education. Roger’s comment led the students to talk more Lesley S. King about their experience as interns. In some ways they see SFI as a world without boundaries, both in terms of the

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 23 New Members Named to SFI Boards, External Faculty PHOTO: DAN BARSOTTI

TRUSTEES SCIENCE BOARD

John Koza is a consulting professor in the Medical Frances Arnold is professor of chemical engineering and Informatics Section and the Electrical Engineering biochemistry at the California Institute of Technology. Arnold Department at Stanford University. He is president of Third received her Ph.D. in chemical engineering from the Millennium Venture Capital Limited and of Genetic University of California at Berkeley in 1985. At Caltech, Programming, Inc. Koza is the author of numerous papers Arnold’s group is applying state-of-the-art methods to and serves on the editorial boards of Genetic Programming address central issues in protein design and the evolution and Evolvable Machines, IEEE Transactions on Evolutionary of enzymes and biosynthetic pathways. This research Computation, Evolutionary Computation, and Artificial Life. requires contributions from a variety of disciplines, including He is the author of three books on genetic programming biochemistry, molecular biology, chemical engineering, including Genetic Programming III: Darwinian Invention and chemistry, and applied physics. Problem Solving, published this year. Marjorie Blumenthal is executive director of the Ford Rowan is a former NBC news correspondent and Computer Science and Telecommunications Board of the host of International Edition, a weekly program on public National Research Council, which is part of the National television. He is the principal author of Crisis Prevention, Academy of Sciences. One of the purposes of the Board is Management and Communications (1991). Rowan has writ- to foster interaction among computer science, computing ten widely in the media and communications field, including and telecommunications technologies, and other pure and articles on topics such as news ethics and information tech- applied science and technology. Last year Blumenthal was a nology. Rowan has taught at Northwestern University’s visiting scientist at the MIT Laboratory for Computer Science Medill School of Journalism and at the University of where she designed and taught a graduate seminar on com- Southern California’s Washington Public Affairs Center. puting and public policy.

24 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 Manfred Eigen, 1967 Nobel Laureate in chemistry, pio- oratory and in organisms as far ranging as ciliates or try- neered and remains a leader in the field of molecular evolu- panosomes. Her work on “gene unscrambling” and RNA tion. His interests in biochemistry range from hydrogen editing in these organisms offers a fresh way of thinking bridges of nucleic acids, through the dynamics of code about the construction of functional genes from encrypted transfer, to enzymes and lipid membranes. Biological control pieces of the genome, as biological computation. and regulation processes, and the problem of the storage of information in the central nervous system also occupy his Harold Morowitz is Clarence J. Robinson Professor of attention. In 1957 Max Planck Institute in Göttingen appoint- Biology and Natural Philosophy at George Mason University. ed Eigen a Scientific Member, where in 1964 he became the Morowitz became a Robinson Professor after teaching at head of the Institute. In 1967, he was elected managing Yale University as professor of molecular biophysics and bio- director of the Institute for a period of three years. At the chemistry and serving for five years as master of Pierson same time he was appointed to the Scientific Council of the College. The author of several books, Morowitz has written German Federal Republic. Eigen is recipient of numerous extensively on the thermodynamics of living systems, as prizes in addition to the Nobel; these include Foreign well as on popular topics in science. In his current research, Honorary Member of the American Academy of Arts and Morowitz is investigating the interface of biology and infor- Sciences and the Linus Pauling Medal of the American mation sciences and continues his exploration of the origins Chemical Society. of life. His books include The Origin of Cellular Life: Metabolism Recapitulates Biogenesis and The Facts of Life Hans Frauenfelder is director of the Center for Nonlinear (co-authored with James Trefil). He is director of the Studies at Los Alamos National Laboratory. During his more Krasnow Institute for Advanced Study and editor-in-chief of than 50 years of research in physics, Frauenfelder has the journal Complexity. moved through a number of different fields. He notes, “I started by studying nuclear energy levels, explored the sur- Mitchel Resnick is currently associate professor at the face effects with radioactivity, discovered perturbed angular Media Laboratory at Massachusetts Institute of Technology, correlation, helped elucidate parity violation in the weak where he holds the LEGO Papert Chair. Resnick’s research interactions, used the Mossbauer effect, and finally began interests include the role of technology in learning and edu- to investigate the physics of proteins. I find biomolecular cation, design of computational systems for nonexperts and physics as fascinating (or more challenging) than any other children, decentralized systems and decentralized thinking, branch of physics and continue to do work in this field. One informal learning environments, and learning in virtual com- of my goals at the Center for Nonlinear Studies is to foster munities. Resnick co-developed LEGO/Logo, a computer-con- more interaction between theory and experiment, and trolled construction set and PlayWrite, a reading/writing com- increase work in biological physics.” puter program. He also developed Starlogo, a programmable modeling environment designed to help students explore Mimi Koehl is a professor in the Integrative Biology decentralized systems and self-organizing phenomena. Department at the University of California, Berkeley. Her research involves the application of fluid dynamics and solid Shripad Tuljapurkar is president and chief scientist at mechanics to the study of biological structure. The aim is to Mountain View Research (MVR) a population-science better understand basic physical rules that apply across research firm in California. MVR creates innovative tools for taxa and to provide tools for understanding and predicting population forecasting. Tuljapurkar has a 1976 Ph.D., and how organisms interact physically with each other and with has held faculty positions at Portland State University, the their abiotic environments. Koehl’s work places a strong University of California at Berkeley, and Stanford University. emphasis on field work as well as on laboratory experimen- His research focuses on uncertainty in human, ecological, tation, and investigates structure and function on various and evolutionary processes. Tuljapurkar is author and co- levels of organization including tissue, organismal, and envi- author of numerous scientific papers and two books. In ronmental. 1990, he was elected Fellow of the American Association of Arts and Sciences, and in 1996 received the Mindel Sheps Laura Landweber received her Ph.D. from Harvard award from the Population Association of America. University in 1993. Her area of study was biology in the Department of Cellular and Developmental Biology. She has been an assistant professor of biology in the Department of Ecology & Evolutionary Biology at Princeton since 1994. A recipient of a Burroughs Wellcome Fund New Investigator Award in molecular parasitology, her main interest is the evolution of biological information processing, or complex molecular systems, both in test-tube experiments in the lab-

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 25 EXTERNAL FACULTY OTHER EXTERNAL FACULTY MEMBERS BEGINNING TERMS ARE:

Two former members of the Institute’s residential Elizabeth Bradley received her Ph.D. in electrical engi- research community have become members of the SFI neering and computer science from MIT in 1992. She is External Faculty: associate professor at the University of Colorado at Boulder, holding a joint appointment in the Computer Science and W. Brian Arthur is Citibank Professor at the Santa Fe Electrical and Computer Engineering Departments. Her Institute and PricewaterhouseCoopers Fellow. From 1983 to research interests focus on artificial intelligence, or com- 1996 he was Dean and Virginia Morrison Professor of puter tools that autonomously analyze and/or design things, Economics and Population Studies at Stanford University. and on techniques for characterizing and exploiting the He holds a Ph.D. from Berkeley in operations research, and unique properties of chaos. Bradley, currently a Packard has other degrees in economics, engineering and mathe- Fellow, also holds the 1999 John & Mercedes Peebles matics. Arthur was the first director of the Economics Innovation in Teaching Award. Program at the Santa Fe Institute in New Mexico; and he currently serves on the Institute’s Board of Trustees. Arthur Stephen Lansing is Professor of Anthropology, Natural pioneered the study of positive feedbacks or increasing Resources, and Environment at the University of Michigan. returns in the economy—in particular their role in magnify- Lansing’s research has involved the study of the relation- ing small, random events in the economy. His current inter- ship between Balinese religious systems, farming, and ests are the economics of high technology, the “new econo- human ecology. He has demonstrated that ideology has my” and how business evolves in an era of high technology, strong control over land use and other environmental factors cognition in the economy, and financial markets. on this island nation. In 1995, he was the recipient of the J.I. Staley Prize from the School of American Research for Melanie Mitchell received a Ph.D. in computer science his book Priests and Programmers: Technologies of Power in from the University of Michigan in 1990. From 1992 to the Engineered Landscape of Bali. He currently has a book 1999 she was research professor at the Santa Fe Institute, in preparation titled Ecology, Complexity, and Social Theory and directed the Institute’s program in adaptive computa- for Princeton University Press. tion. She is currently a technical staff member in the Biophysics Group at Los Alamos National Laboratory. University of Iowa economist Scott Page’s work involves Mitchell’s research interests include intelligent systems and the application of computational and complex systems machine learning; evolutionary computation and artificial methods to economics and political economy. Some of his life; decentralized parallel computation in spatially extended work has focused on how models with interactive agents can systems such as cellular automata; and cognitive science, illuminate political phenomena such as party platform char- particularly computer modeling of perception and analogy- acteristics and the advantages of incumbency. Page has making, emergent computation and representation, and also done research into diversity and problem-solving and is philosophical foundations of cognitive science. developing a theory of how different skills produce syner- gies that result in powerful aggregate problem-solving capacity. With John Miller, Page has been co-director of SFI’s Graduate Workshop in Computational Economics since 1995, and he serves on the steering committee of the Institute’s Fellows-at-Large initiative.

John Reinitz is associate professor at the Brooksdale Center for Molecular Biology, Mount Sinai Medical School. Reinitz performs both theoretical and experimental studies on mechanisms of segmentation genes expression during early Drosophila embryogenesis. The qualitative characteristics of segmentation genes expression patterns are used as variables in a dynamic model for genes expression pattern formation. The validity and utility of the model has been demonstrated by its successful application to several important problems of embryo development.

David Stark is the Arnold A. Saltzman Professor of Sociology and International Affairs and chair of the Department of Sociology at Columbia University. Stark is

26 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 currently doing research on new organizational forms among firms in Manhattan’s Silicon Alley. In postsocialist Eastern Europe, he studied how interfirm networks facilitated and impeded economic restructuring. His recent publications include Heterarchy: Distributed Intelligence and the Organization of Diversity, forthcoming from Princeton University Press. Postsocialist Pathways: Transforming Politics and Property in Eastern Europe, with Laszlo Bruszt, Cambridge University Press, is a comparative study of the opportunities and dilemmas posed by the simultaneous extension of property rights and citizenship rights.

Andreas Wagner is assistant professor in biology at the University of New Mexico. Wagner received his Ph.D. from Yale University in 1994. Following fellowships at the Institute MOTOROLA’S GALVIN TAKES SFI POST for Advanced Study in Berlin and then with Leo Buss at Yale, Wagner Robert W. Galvin was named chairman of the Santa Fe Institute Board of was a postdoctoral fellow at the Trustees at its May 1999 meeting. Galvin started his career at Motorola in Santa Fe Institute from 1996 to 1940. He held the senior officership position at the company from 1959 until 1998. While at SFI Wagner pursued 1990, when he became chairman of the Executive Committee. Currently, he research projects on the evolution of serves as a full-time officer of the company. genetic redundancy and developed Galvin attended the University of Notre Dame and the University of Chicago, mathematical and computational and is now a member and was the recent chairman of the Board of Trustees techniques for the analysis of whole of the Illinois Institute of Technology. genomes. His SFI redundancy work He has been awarded honorary degrees and other recognitions, including rests on a population genetics election to the National Business Hall of Fame and receipt of the National model based on the fact that redun- Medal of Technology in 1991. dancy provides protection against Continuing as vice-chair of the Board is Robert J. Denison, founder and deleterious mutations. Within this chairman of First Security Management, Inc. framework he was able to show that Galvin and Denison replace David Liddle and Robert Maxfield who have selection cannot only maintain but served as chair and vice-chair since 1994. Both Liddle and Maxfield remain on increase genetic redundancy in large the Board of Trustees. populations. “I shall always be grateful to David Liddle and Bob Maxfield for their help in moving SFI forward,” said SFI President Ellen Goldberg at the May meeting. “I look forward to their continued involvement as active members of our Board. “I am delighted that Bob Denison has agreed to continue on as vice-chairman of the Board. His knowledge of fiduciary issues provides all of us with sound advice, and his breadth of knowledge in a number of fields provides the Institute with a wonderful resource for counsel. “As to having Bob Galvin agree to become chairman of the Board, I am truly thrilled. He is a visionary who will help move SFI to new heights. I look forward to continuing to work with the entire Board of Trustees, and I thank all of the members for their commitment to SFI. Their guidance has been and continues to be very much appreciated.”

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 27 Language Issues

BY KEN BAAKE

Issues of language surface frequently at the Santa Fe Institute, where discussions among members often lead to debates about word meaning. One of the biggest challenges in this arena is finding words and metaphors that can stimulate productive scientific thinking without distorting that thinking. Dynamic new concepts are constantly emerging at the Institute. Language scholars define metaphor action has nothing to do with laws, Just ask librarian Margaret in various ways, but most say that it he said. Clearly, the semantical Alexander about her list of scientific is a rhetorical device for transporting debate centered around the specific “buzz words” that she uses when knowledge by using a word that use of language in different scientif- shopping for books to add to the brings connotations from one field ic disciplines. As Ostrom explained library. The list is always changing, into play in another field. The word later, “rules” for biologists might be she says. has at its roots the Greek word observed regular behavior or strate- The word “complexity,” one of “phora,” which means locomotion. gies, while “rules” for political sci- the central concepts at the Institute, SFI theoretical chemist Walter entists might be enforceable laws presents possibly the biggest defini- Fontana explains the power of such agreed upon by members of a tion challenge. A number of scien- locomotion in SFI science. group. tists acknowledge that they do not “Metaphor makes people realize But terminology can be prob- G & V CHAPMAN/IMAGE BANK ILLUSTRATION: really know what it means. The that certain old questions can be lematic even when confined to a apparent vagueness of the term, cast as new ones. It triggers new single scientific discipline. however, may be what makes it so thoughts and speculations.” Theoretical biologist Michael valuable as a catalyst for thought. A It can also trigger a lot of discus- Lachmann calls attention to the word like complexity is new and sion. Take for instance the public term “signaling” in biology as an unresolved; it is not an inert tool of lecture by Indiana University politi- example of a metaphor that is rich, scientific description, but rather an cal scientist Elinor Ostrom at last but also defies precise definition. idea whose meaning evolves as it summer’s Integrated Themes What does it mean when we say a interacts with researchers. Workshop. Ostrom used a game the- yellow bee is signaling that it is dan- SFI President Ellen Goldberg oretical approach to look at public gerous to eat? “Does that mean that says that complexity involves policy questions. She examined the a brown bee is not dangerous to “interacting parts with very simple ways in which a community decides eat?” Lachmann asks. Signaling rules,” but she is quick to add that public policy, such as the allocation implies an intentionality on the part the term does not reduce to a con- of irrigation rights among farmers. of some living agent, but the bee’s stant definition across disciplines. Much of the question and color is purely the product of evolu- For that reason Goldberg likes the answer portion of her talk con- tion. Can we use a metaphor like term; it is flexible and permits mul- cerned debate over the word signaling, with all its connotations of tiple definitions. Former SFI presi- “rules,” a term Ostrom used to intentionality, to refer to a DNA- dent George Cowan also accepts the describe how members of the com- coded process? “Who intends the polysemantic nature of the word. munity would interact with each bee to be yellow?” Lachmann asks. “Its chief value is that it embraces a other and what policy decisions they The good news is that such number of possible systems,” would make. But one researcher questions lead to further explo- Cowan says. questioned the use of the word ration. SFI scientists are constantly The definition challenges “rules” because it implies legal reg- stretching terms to help explain increase when you bring metaphors ulations, while much human inter- new insights. For example, theoret- into the rhetorical equation. inside SFI

28 SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 ical chemist Walter Fontana has lead to distortions as well by spawn- Metaphor is a major rhetorical taken the common word “neutrali- ing unintended connotations that focus at SFI, but it is not the only ty” and used it to argue that the influence society far beyond the one. Other problems involve genetic code underlying a com- intention of those scientists who researchers having to modify their pound such as RNA (ribonucleic developed the metaphors as aids to speaking and writing according to acid) can change frequently without cognition. Einstein’s concept of rel- the needs of a lay audience. One of affecting the performance of the ativity, for example, has rippled the most difficult rhetorical chal- compound—until at some point the beyond the confines of physics and lenges scientists face when explain- compound seems to “suddenly” cosmology into the world at large, ing their research to the lay public evolve to a higher level of perfor- where some might argue that it has comes in answering the question, mance. Such “neutral” changes of reinforced a moral relativism in “And so what?” Dulle notes. the underlying genetic sequence Western society. SFI physicist People often wonder what the make evolution possible, Fontana Cris Moore presents another research means to their lives, to says. example—the metaphor of the world at large. Still, Fontana acknowledges genetic mapping, which implies SFI scientists also face a that the neutrality may only be tem- a certain determinism that could unique challenge when writing porary; eventually the compound possibly lead to social eugenics. to an audience of specialists in evolves. So, it seems appropriate to Metaphors are so powerful another field. At times they’re ask if a change can be called “neu- that they can paralyze a frustrated when trying to publish tral” when it has no apparent imme- researcher who becomes so their results in scholarly journals diate effect on a compound or an enamored of a term that she or he that may not be receptive to the organism, yet over time, in combi- loses the flexibility to see the world unique and almost poetic language nation with other changes, it leads outside of the frame of the of complexity science. For example, to a profound impact. Again, it is metaphor. As Fontana says, “You economist graduate researcher obvious that when scientists strug- fall in line with ideas and then you Shareen Joshi says she often uses gle with word meaning they often become brittle.” For example, concepts like “the edge of chaos” are led to new insights—and new Moore notes that metaphors import- when envisioning the economy in questions. ed from biology into economics may her mind, but she would never use Fontana’s research has piqued distort impressions of the economy such terms in a journal article. the interest of social scientists affili- by making it seem more natural Joshi’s compromise with elo- ated with SFI who are considering than it really is. “The economy isn’t quent language reveals the true whether seemingly unimportant a jungle,” Moore says. Economic essence of metaphor as a midwife of “neutral” events in human history systems have governments, which knowledge. Metaphor can open the and human economies could have are not present in the rain forest. researcher’s mind to fresh scientific major consequences over time. The Other times the metaphors are insights that might remain occluded neutrality concept also may apply to so catchy that they imply quick-fix, without the powerful cognitive present-day business. Suppose you real-world applications for a particu- “locomotion” that metaphor entails. are a chief executive officer who lar theory that cannot be delivered, Yet, because metaphor is so powerful wants to reorganize your firm, at least not in the short term. and perhaps intrinsic to human cog- Fontana says. You know that a suc- Suzanne Dulle, SFI director of busi- nition, it also carries the dangers cessful reorganization would require ness relations and external affairs, inherent in any powerful tool; a lot of small changes affecting the says that the language of complexity metaphoric language can distort the way employees go about their jobs, has flooded the popular business- truth and imply a greater sense of sci- but you have no way of knowing in book press, leading to an expecta- entific certainty than the results of advance the best combination of tion among some business people rigorous scientific exploration would changes to make. Neutrality sug- that they can spend a few days at support. Most at SFI seem willing to gests that you can make small SFI and come away knowing how to accept this paradox; they know that changes gradually—even in an save their company money. Some the benefits outweigh the risks. uncertain environment—without business managers hearing about risk that you will sink your firm. self-organization have asked half- Ken Baake is a doctoral candidate in Rhetoric Clearly the way scientists stretch seriously if they should step aside and Professional Communication at New Mexico metaphors and other terms helps and let the company run on its own, State University in Las Cruces. For his disserta- elucidate ideas. But metaphors can Dulle says. tion he is examining issues of language at SFI.

SANTA FE INSTITUTE BULLETIN ¥ FALL 1999 NON-PROFIT ORGANIZATION U.S. POSTAGE SANTA FE INSTITUTE 1399 HYDE PARK ROAD PAID SANTA FE, NEW MEXICO 87501 PERMIT NO. 49 U.S.A. LUBBOCK, TEXAS