Kshetri, Nir (2004), “Economics of Adoption in Developing Countries,” IEEE , 21 (1), 74-81. Made available courtesy of Institute of Electrical and Electronics Engineers: http://www.ieee.org/ ***Reprinted with permission. No further reproduction is authorized without written permission from the Institute of Electrical and Electronics Engineers. This version of the document is not the version of record. Figures and/or pictures may be missing from this format of the document.*** focusdeveloping with software

Economics of Linux Adoption in Developing Countries

Nir Kshetri, University of North Carolina, Greensboro

here was once a time when developing software required only three main ingredients: a user with a need, a computer, and a programmer. Once these three ingredients came together, the re- T sulting team could largely ignore everything in the outside world except the occasional need for pizza. Later, both the sizes and goals of such teams grew considerably, but often the insular attitude remained. Needs, computers, and software developers still went into a big black box, and

sometime later, if the project was fortunate, the two decades, software development has gone required software emerged as if by magic. from an esoteric discipline dominated by For most software developers, those days American and European white-collar workers of splendid isolation disappeared long ago. In- to a significant new global driver of emerging creasingly, software developers became de- national economies. This globalization can af- pendent on their choice of external software fect build-versus-buy decisions at every level products, which supported them but over of software development. Software managers which they had no direct control. A wrong and developers who ignore this expansion do choice could spell disaster for a project, no so at their fiscal and productivity peril. matter how well the actual development went. Linux, an open source , is Recent globalization of software has made an interesting example of a globally available this choice even more critical. Over the last software product that has nontrivial choice and productivity implications for software projects in developing countries. Its low cost, Software globalization affects the build-versus-buy decision availability, and range of features process at every level and phase of software development. One of make it an intriguing option when developing the most interesting and controversial issues is the expansion of and deploying new software using limited re- sources. However, software managers must Linux into software development efforts in developing countries. examine the full range of costs and benefits in-

74 IEEE SOFTWARE Published by the IEEE Computer Society 0740-7459/04/$20.00 © 2004 IEEE volved in selecting Linux over proprietary op- tions, because they can have a powerful im- 100 Developed countries pact on both individual developers and devel- 89.7 90 87.4 Developing countries oping countries. A deeper and richer 81.6 understanding of the facilitators and hurdles 80 75.9 of Linux adoption in developing countries 70 could help software developers and marketers devise appropriate strategies to accelerate its 60 diffusion and to recognize cases where a pro- prietary solution might be a better choice. 50 Percent This article examines the positive and negative 40 effects of adopting Linux and ways to achieve the greatest overall benefits for developing 30 24.1 countries and their policy makers. Because 20 18.4 ’s Windows operating system, used 12.6 10.3 in over 90 percent of the world’s desktops, is 10 the major proprietary system competing with 0 Linux, we compare Linux with Windows on 1998 2001 1998 2001 several dimensions. Linux Year The current status of Linux use Open source software is often touted to be ideal for accelerating the growth of low-income Figure 1. Percentage of countries’ IT sectors, with the expectation that developing and developed countries seem Linux and Internet it will increase their propensity to innovate. more disparate in Linux adoption and use users from developed Developing countries seem to have adopted than in other modern information and com- and developing Linux, the “flagship” of open source,1 faster munications technologies. countries.3–6 than developed countries, according to numer- ous publications. For instance, a study con- Microeconomics of Linux adoption ducted by the United Nations Conference on Software managers must examine four mi- Trade and Development indicated that Linux croeconomics issues of Linux use and selec- has been a significant force behind some devel- tion: ownership, effective use, learning and oping countries’ increased IT ascendancy in re- switching to Linux, and compatibility (see cent years.2 (In this article, the term “devel- Table 1). These economics evaluate the costs oped countries” indicates the 30 member and benefits of Linux versus proprietary soft- countries of the Organization for Economic ware at the level of an adopting unit. Cooperation and Development, and “develop- ing countries” indicates all other countries.) Ownership Because users can purchase Linux from dis- Ownership includes investment, mainte- tributors and also freely download it from nance, and the costs of downtime and obso- various FTP Web sites, it can be difficult to lescence. These costs are a much higher pro- reasonably estimate developing countries’ portion of income for a computer user in share of the Linux market. But the number of Burundi, which has an annual per-capita gross registered users at the Linux Counter domestic product of US$99, than for a user in (http://counter.li.org) and their countries sug- Belgium. Open source advocates argue that gest an approximate distribution of Linux Linux has a lower total cost of ownership users worldwide. In 2001, people living in de- (TCO) than its competing products. For in- veloping countries (81 percent of the world stance, the price for Microsoft’s entry-level population) accounted for 24.1 percent of In- operating system was US$5 per computer in ternet users but only 18.6 percent of Linux 1981 and $90 in 2002. Taking $600 as a PC’s users.3,4 Moreover, from 19985,6 to 2001, the average price, Microsoft’s entry-level operat- proportion of Linux users in developing ing system amounts to 15 percent of the PC’s economies grew more slowly than the propor- total cost. Although the operating system’s tion of Internet users (see Figure 1).3–6 Thus, cost decreases when manufacturers buy it in

January/February 2004 IEEE SOFTWARE 75 Table 1 Macro- and microinfluences on choosing Linux in developing countries Factor Positive effects Negative effects Microeconomics: Ownership Slower obsolescence of basic infrastructure features Lack of supports to deal with security vulnerabilities. help reduce total cost of ownership. Effective use Amenability to modification makes localized Ordinary users can’t custom-configure the system. customization easier. Costs of supporting custom changes can escalate dramatically over time. Learning/switching Switching costs might be lower compared to Undertaking to use the full complexity of Linux utilities and source developed countries. code can lead to higher learning and switching costs. Linux communities provide supportive environments for transition. Compatibility Linux has high levels of compatibility and portability Likely to be incompatible with business partners’ technologies. for old and used hardware. Hardware-OS incompatibility if Linux device drivers are unavailable. Macroeconomics: Enforcement of Linux IP rules encompass both aggressive sharing Proprietary versions of Linux depend on IP laws. intellectual property laws of basic resources and support for business growth. National security Linux provides an easier basis for global sharing of Microsoft’s opening of codes increases Windows’ relative security infrastructure. attractiveness.

bulk, it is still likely to be a significant pro- Electronics sells Linux-based desktops in portion of a PC’s total cost. A cost comparison that are priced lower than brands using com- based on data from the Tech Zone (http:// mercial operating systems. The company says thetechnozone.com) indicates that although it lowered the cost by legally saving the license Mac OS X seems cheaper than Windows XP, cost of proprietary operating systems. upgrade costs over time make it twice as ex- Linux traditionally has a strong reputation pensive. Likewise, according to Internet among developers for its ability to continue Week’s 3 April 2000 issue, Linux is less ex- running under adverse circumstances, leading pensive than BSD/OS. to lower costs due to less downtime. Another Ironically, proprietary operating systems’ powerful distinguishing factor in downtime higher prices often aren’t a significant issue in costs is the relative prevalence of malicious developing countries because piracy of propri- software. According to the Symantec Corp. in etary software is so endemic. However, the an 18 September 2003 article in the Wall Street World Trade Organization’s Agreement on Journal, more than 4,000 viruses and other Trade-Related Aspects of Intellectual Property malicious codes were launched against Mi- Rights (TRIPS) requires developing countries to crosoft Windows during the first eight months provide adequate legal and enforcement tools of 2003, compared to 11 attacks against Unix to prevent such piracy. Enforcement will be and Linux combined for the same period. staggered, with the last mandatory compliance Moreover, by using communications such as being 1 January 2006 for the poorest coun- email, bug reports, bulletin boards, and vari- tries. Developing countries’ full compliance ous tracking mechanisms, Linux communities with this agreement will thus likely strengthen have gained a strong technical reputation for the relative value proposition of Linux adop- their ability to address and process most at- tion over time. tacks and errors rapidly.7 On the other hand, Whereas prices so far haven’t been of much rapid discovery of errors or attacks in Linux concern to individuals using pirated propri- doesn’t necessarily translate into adequate han- etary software, they’ve contributed to the high dling of the attacks at the individual user level, prices of PCs that are bundled with propri- whereas Windows users have a wide range of etary operating systems. So, one way to bring online and commercial resources available to down manufacturers’ PC prices is to use Linux help combat virus attacks. as the operating system. For instance, the Although a study commissioned by Mi- South Korean multinational company LG crosoft found that Windows’ TCO was

76 IEEE SOFTWARE www.computer.org/software cheaper by at least 10 percent in four of the speech recognition, and text-to-speech conver- five most common tasks,8 surveys of open sion. Time magazine described Simputer as source users tend to give different results. For among the 10 best technological innovations An individual’s instance, an article published by the NewsFac- of 2001. India, which has two official lan- tor Network (http://newsfactor.com) on 7 No- guages (Hindi and English), 18 major lan- capability to vember 2002 quotes a survey conducted in the guages, and 418 officially listed languages, has enjoy the US, , France, Germany, Sweden, and benefited from Simputer’s low cost, efficiency, economics of Japan in which 40 percent of the respondents and local-language capabilities. For example, perceived Linux as having a lower TCO. Some India’s Karnataka state accountants use Sim- effective use organizations from developing countries that puters to collect and upload village data to the is a function have adopted Linux have also reported a sub- government servers, cutting data collection stantially lower TCO for Linux. For instance, time by an estimated 80 to 90 percent. Simi- of skill level. Univates (a university in Lajeado, Brazil) es- larly, Brazil’s US$250 Computador Popular, a timates that switching to Linux saved Linux-based Internet appliance with no floppy US$130,000 in one-time and software or hard disk drive, features many of the at- procurement costs, plus an annual savings of tributes you’d expect in a moderately priced $70,000 on upgrades and maintenance. PC. Consumers can also buy inexpensive hard disks and other peripherals. The first shipment Effective use of Computador Populars is planned to go to If we view open source as a form of tech- schools, libraries, and health centers to pro- nology transfer, developing countries’ capabil- vide access to the Internet. ity to employ it effectively is critical to the What’s more, Linux’s transparent and col- transfer process’s overall success.9 For every- laborative nature will let developing countries day data-processing applications, open source provide affordable technology education and and deploy with similar skills development, which will help build local effectiveness in developing countries. How- programming skills and avoid spending hard ever, when it comes to customization and currency on foreign technologies. Addition- adaptation to highly specific local technology ally, the Simputer might help train students on transfer, the availability of source code pro- software in countries where there are few op- vides important and obvious advantages.10 portunities to use a computer because of Eng- Adaptation to local languages is a good lish language constraints. example. The most powerful adaptation ad- There’s one caveat: An individual’s capabil- vantages show up not on large language ity to enjoy the economics of effective use is a groups such as Asian character sets but on function of skill level. Substantial skills might smaller groups for which there might not be be needed to custom-configure some aspects sufficient market for a proprietary company of Linux systems to suit particular needs. to create and support a localized version. For example, the Open Source Software Transla- Learning and switching to Linux tion Project in South Africa (www.translate. Linux’s extreme configurability has created org.za) has produced Xhosa language packs a major disincentive for learning and switching for Linux to counter the lack of support for to it. Josh Quittner described Linux’s complex- Windows systems. ity in his Time magazine article of 24 May Linux’s affordability has allowed some de- 1999: “The [Linux] interface is user friendly veloping countries to develop low-cost tech- only if the user happens to be a [computer sci- nologies that meet their unique needs. For in- ence] PhD.” Since then, Linux programmers stance, India-based Encore Software has have worked hard to make the software more designed a handheld Internet appliance, called user friendly, and many of its users consider Simputer, based on Linux. Simputer uses the Quittner’s assertion an overstatement now. Intel StrongARM chip (known for its low Compared to commercial software, however, a power consumption), 64 Mbytes of RAM, 32 small-budget Linux project typically has limited Mbytes of flash memory, and a modem to con- support and staff knowledge and fewer usabil- nect to a telephone line. At a cost of less than ity labs, user surveys, and outside experts such US$200, Simputer provides Internet and email as technical authors and graphic designers. access in local languages, microbanking, Moreover, whereas Linux-friendly commer-

January/February 2004 IEEE SOFTWARE 77 cial companies such as , VA Software, Valley Toxics Coalition, for instance, esti- and IBM provide complementary services and mates that the US will discard 500 million A significant products not supplied by open source commu- computers between 1997 and 2007, 80 per- nities,11 such measures are heavily oriented to- cent of which will be shipped to developing proportion of ward developed countries. As with other tech- countries. Linux tends to be smaller than the Microsoft’s nologies, commercial distributors often find latest versions of Windows, allowing it to run US$400 million developing countries unprofitable for their even on outdated machines powered by Intel markets. Pentium 90 and even older 486 chips. Older investment in However, Linux communities worldwide versions of Windows, however, are smaller India is being and collaborations among them have con- and run on older machines and can still be tributed to the economics of learning open bought and used. However, Linux device driv- spent on source. They’re helping educational institu- ers might not be available in developing coun- computer tions, government agencies, private busi- tries for some components, which worsens the literacy nesses, and individual users worldwide to economics of Linux ownership. adopt Linux. Kathleen Sibley, in an article Second, Linux is experiencing a vicious cir- programs and published in Computing Canada on 21 July cle of low penetration of both its operating localization of 1997, compares Linux communities with the system and applications. Although some ap- Red Cross and the United Nations. Similarly, plications such as and VMware can be its products. Thomas Malone and Robert Laubacher, in a used to run Windows applications on and Harvard Business Review article published in within Linux, many of the Linux and Win- September 1998 (pp. 144–152), view Linux dows applications don’t work on each other’s communities as a “model for a new kind of platforms. So, Linux can’t become widely used business organization that could form the ba- unless its applications become popular, but its sis for a new kind of economy.” Linux user applications will become popular only if groups are located in many places—from Al- Linux is widely used. Because Linux is free, bania to Armenia—where less experienced Microsoft isn’t likely to gain from the demand users get installation and other help from for hybrids (for example, using Microsoft’s knowledgeable and experienced users. applications and Linux operating systems), so On the other hand, Microsoft programs in it has little economic incentive to create them. some developing countries are helping to in- The resulting absence of products such as Mi- still Windows-centric computer habits. For in- crosoft Office on Linux thus encourages desk- stance, a significant proportion of Microsoft’s top users to stay with Microsoft or other pro- US$400 million investment in India is being prietary operating systems. spent on computer literacy programs and lo- Third, Linux users are also encountering calization of its products. Microsoft is also compatibility issues with business partners’ contributing its software to schools and is pro- standards. Trading relationships between firms viding training to 80,000 schoolteachers and is a function of their technologies’ degree of 3.5 million students in government-run Indian “fit,” or the “technological distance.” Because schools. This acclimatization increases the rel- of their lower bargaining power, firms from de- ative cost of learning and switching to Linux. veloping countries are forced to comply with the technologies used by their trading partners in Compatibility advanced countries. For instance, a 1995 study Networked technologies such as Linux de- found that organizations from developed coun- rive much of their value from compatibility. tries accepted new suppliers only if they could Corporate resources such as hardware, applica- demonstrate an electronic-data-interchange ca- tions, and existing files and technologies used pability.12 Similarly, pressure from American by trading partners largely influence this com- multinationals such as Wal-Mart and J.C. Pen- patibility. There are three important issues here. ney require their foreign suppliers to transact The first is compatibility between operating on the Internet. The suppliers, mainly from de- systems and hardware. Because of the unaf- veloping Asian countries, adopted the Internet fordability of new PCs and thanks to charita- because of such pressure. ble donations from developed countries, a Software upgrading thus tends to diffuse much higher proportion of developing coun- from big to small companies and from compa- tries’ computers are used and old. The Silicon nies in developed countries to those in develop-

78 IEEE SOFTWARE www.computer.org/software ing countries. This is commonly known as the measures of institutional intervention. In Asia, rank effect in industrial economics literature. for instance, almost every country has a “na- When trading partners from developed coun- tional” Linux and a number of high-profile Some tries follow the Microsoft standard, as many do, open source projects. For example, under In- firms in developing countries are less likely to dia’s Indlinux.org, a vast army of program- governments, adopt Linux. In this aspect, Linux attracts gov- mers are working on Linux to make IT’s ben- such as China, ernments, especially municipalities, because they efits freely available. The Chinese government perceive are not as connected with the outside world and established the Beijing Software Industry Pro- don’t have to exchange information with cus- ductivity Center to organize Linux develop- proprietary tomers that follow Microsoft’s standards. ment. Yangfan Linux, a version developed by software’s the center, has already been installed in many Macroeconomics of Linux adoption government computers. hidden Software managers must also examine two Developing countries’ orientation toward protocols as macroeconomic issues: enforcement of intel- IP protection and the requirement to comply threats to their lectual-property laws and national security with TRIPS, however, will have a differential (see Table 1). impact on the penetration of free and propri- national etary versions of Linux. Whereas free versions security. Enforcement of IP laws might diffuse faster in these countries, propri- The WTO’s TRIPS agreement has obligated etary versions of Linux depend largely on the developing countries to provide and ensure the availability and enforcement of IP laws. adequacy of new antipiracy enforcement tools.13 Strengthening IP protection in devel- National security oping countries arguably creates incentives for Policy makers around the world are realiz- domestic innovation. Nevertheless, developing ing the increasing influence of technologies in countries feel that the TRIPS agreement has cre- the national-security game. The impact and ated a “disproportionate burden” on them distinctions between proprietary and open without “tangible development benefits.”14 source development are a fascinating compo- Societies in many developing countries nent of this debate and have become an im- don’t support the concept of private owner- portant element in some countries’ national- ship of ideas and hence don’t support IP pro- security decision making. An article published tection laws.15 Compounded by low income in China Economic Times on 12 June 2000 and proprietary software’s ever-increasing discusses the views of Xu Guanhua, China’s costs, a high proportion of software products vice minister of science and technology at that in developing countries are illegal copies. For time, on how high technology affects national instance, 96 percent of software used in China security militarily, economically, and cultur- and Indonesia, 97 percent in Thailand, 95 per- ally. Regarding military security, Guanhua cent in Turkey, and 99 percent in Vietnam are said that developed countries have used many estimated to be illegally copied.16 hi-tech arms in battle and that technology- Changing the prevalent social beliefs has exporting countries might have installed pro- been a costly proposition for such governments. grams for “coercing, attacking, or sabotage” Citizens also perceive such enforcement tools as into the software they’re selling into China. supports to foreign software companies. For in- Some governments, such as China, perceive stance, Taiwan’s attempt to force students using proprietary software’s hidden protocols as pirated versions of to pay threats to their national security because it’s was perceived by the students as supporting a hard to know what the software is doing or foreign company rather than its own citizens. whether data is being shared inappropriately. What’s more, enforcement of IP laws also re- The Chinese government in particular has ex- sults in spending hard-earned foreign currency pressed concerns that Microsoft and the US on software imports. government might spy on Chinese computer Rapid open source diffusion in developing users through secret back doors in Microsoft countries will likely lower economic losses and products. Ironically, the truth or falsity of such reduce administrative costs for IP law enforce- claims is less relevant from a marketing per- ment. So, these governments have opted to spective than the fear itself, which can be very promote open source by applying various damaging to sales. Microsoft has recognized

January/February 2004 IEEE SOFTWARE 79 the danger of such fears and embarked on a hard to increase its user-friendliness so as to program of revealing source code to help alle- extend its reach outside technical and knowl- Exclusive viate such fears. However, proprietary compa- edgeable individuals and those in academic nies will always have more difficulty balancing computing environments. Further improve- statewide IP concerns with security than do open source ment in its user-friendliness can help attract the contracts systems with no hidden IP secrets to protect. critical mass of users required for broader suc- with either National security concerns from the US and cess. Linux communities and Linux-friendly its allies in the form of a Coordinating Commit- commercial companies thus should place more proprietary or tee for Multilateral Export Security (COCOM) emphasis on improving Linux’s simplicity and open source put restrictions on high-technology exports to ease of use as opposed to more traditional countries such as China, which resulted in the quality dimensions such as accuracy, complete- operating avoidance of proprietary software in such coun- ness, features, and structuredness. systems are tries. Despite the disbandment of COCOM in Compatibility issues associated with busi- seldom best for 1994, the US law still restricts the sale of com- ness partners’ technologies have the least in- puters that exceed specified performance limits, fluence on governments in developing coun- a country in the measured in “millions of theoretical operations tries, because they exchange relatively little long term and per second.” Not too surprisingly, the most sig- data with more established groups and sys- nificant impact of such restrictions in the case tems. However, because governments are the should be of software operating systems has been to en- biggest single user of hardware and software avoided. courage rapid growth in China of Linux-based in most developing countries,18 any decisions supercomputing clusters. they make about adopting a specific operating In general, the Chinese government sees system can have a powerful secondary com- Linux as a powerful opportunity to catch up patibility effect on the OS’s subsequent spread and even pull ahead in the global technology and dissemination nationwide. When they race. Some signs of success are beginning to ma- choose proprietary operating systems, the re- terialize and should be taken seriously by na- sult can be an unfortunate exclusion not only tions more concerned with protecting past in- of open source operating systems but also of vestments than with fully exploiting the range other proprietary software systems such as of options made available by open source tech- Apple. Conversely, a decision to use Linux can nologies. For example, in 2002 China an- yield a powerful incentive to use it more nounced a mobile version of Linux developed broadly, including in circumstances where it by Eforce, Culturecom, and Mobile Telecom. might not be the optimal solution. There is no To contribute to national security, develop- easy and fully equitable resolution of this ef- ing countries are enacting new laws and pro- fect, but from the perspective of individual viding guidelines to facilitate Linux develop- governments, it’s important always to focus ment. By mid-2002, Latin American countries both on the ability to access all resources and such as Brazil, Mexico, Argentina, and Peru on keeping future options open. So, exclusive had proposed bills mandating the use of open statewide contracts with either proprietary or source in government organizations. open source operating systems are seldom best for a country in the long term and should be avoided. Instead, multiple contracts that take advantage of the strong points of both open or the most part, Linux isn’t being ex- source and proprietary solutions will more ploited as fully as possible in develop- likely lead to stronger long-term national po- F ing countries, and there are more im- sitions in the global software market. pediments than unfair advantages to Linux Linux communities and Linux-friendly deployment. Given that perspective, how can companies can also help promote diversity we encourage the effective and economically and flexible options by working to make pro- fair use of Linux? prietary and open source operating systems Of the four micro and two macro issues, more interoperable. The Chinese government Linux probably fares the worst in terms of has, for example, launched its own version of learning and switching. However, users that Wine, an open source API (application pro- have touched Linux codes, which exceeded 15 gramming interface) that lets Microsoft appli- million in number by 2001,17 have worked cations run on Linux.

80 IEEE SOFTWARE www.computer.org/software Finally, governments should explicitly sup- Open Source,” J. Industrial Economics, vol. 50, no. 2, port and encourage more of the kinds of low- June 2002, pp. 197–234. 12. R. Schware and P. Kimberley, Information Technology cost, minimal-infrastructure developments and National Trade Facilitation, tech. report 317, demonstrated by Simputer. So far, neither the World Bank, 1995, p. 19. Indian information and communications tech- 13. T.N. Samahon, “TRIPS Copyright Dispute Settlement after the Transition and Moratorium: Nonviolation and nologies industry nor the Indian government Situation Complaints against Developing Countries,” has taken Simputer as seriously as the soft- Law and Policy in Int’l Business; vol. 31, no. 3, Spring ware industry has.19 This is unfortunate, be- 2000, pp. 1051–1075. 14. R.L. Gana, “Prospects for Developing Countries under cause such systems could bring thousands of the TRIPS Agreement,” Vanderbilt J. Transnational the poorest villages into the Computer Age. Law, vol. 29, no. 3, Mar. 1996, pp. 735–740. The implications of such actions could be 15. J.D. Mittelstaedt and R.A. Mittelstaedt, “The Protec- tion of Intellectual Property: Issues of Origination and enormous, not just for industry but also for re- Ownership,” J. Public Policy and Marketing, vol. 16, ducing the global cycle of poverty and disillu- no. 1, Spring 1997, pp.14–25. sionment that engenders many of the greatest 16. D.B. Marron and D.G. Steel, “Which Countries Protect Intellectual Property? The Case of Software Piracy,” dangers of the modern world. Economic Inquiry, vol. 38, no. 2, Apr. 2000, pp. 59–- 174. 17. N.K. Katyal, “Digital Architecture as Crime Control,” The Yale Law J., vol. 112, no. 8, June 2003, pp. 2261–2289. References 18. S.R. Nidumolu et al., “Information Technology for Lo- cal Administration Support: The Governorates Project 1. A. Applewhite, “Should Governments Go Open Source?” in Egypt,” MIS Quarterly, vol. 20, no. 2, June 1996, IEEE Software, vol. 20, no. 4, July/Aug. 2003, pp. 88–91. pp. 196–224. 2. Good Prospects for IT Industry in 2003, Predicts UNC- 19. R.H. Wade, “Bridging the Digital Divide: New Route to TAD; Developing World Looks to E-Business for Growth, Development or New Form of Dependency?” Global United Nations Conf. Trade and Development, 18 Nov. Governance, vol. 8, no. 4, Oct–Dec. 2002, pp. 443–476. 2002, http://r0.unctad.org/ecommerce/docs/edr02_en/ ecdr02press1.htm. 3. “Human Development Report 2003,” United Nations Development Programme, www.undp.org/hdr2003. 4. Linux Counter, http://lists.q-linux.com/pipermail/plug/ For more information on this or any other computing topic, please visit our 2001-September/009641.html. Digital Library at www.computer.org/publications/dlib. 5 European Marketing Data and Statistics and Interna- tional Marketing Data and Statistics, Euromonitor Int’l, 2001; www.euromonitor.com. 6. faq.linux.hr, http://faq.linux.hr/index.phtml?kat= razno&poruka=36. 7. W. Scacchi, “Understanding the Requirements for De- veloping Open Source Software,” IEE Proc.—Software, vol. 149, no. 1, 2002, pp. 24–39. 8. W.M. Bulkeley and R. Buckman, “Microsoft Wages Quiet Campaign against ,” Wall Street J., 9 Dec. 2002, p. B1. We’d like 9. G. Sridharan, “Managing Technology Transfer in Con- struction Joint Ventures,” Am. Assoc. of Cost Engineers Trans., vol. 1994, pp. INT 6.1-6.4. 10. F. Williams and D. Gibson, Technology Transfer: A Communication Perspective, Sage Publications, 1990. to hear 11. J. Lerner and J. Tirole, “Some Simple Economics of from you About the Author

Nir Kshetri is an assistant professor at the Bryan School of Business and Economics, Univer- SEND US EMAIL AT sity of North Carolina, Greensboro. His research focuses on globalization of modern information technologies. He received an MA in economics from Tribhuvan University, Nepal, and a PhD in business administration from the University of @ Rhode Island. He is a member of the Pacific computer.org Telecommunications Council. Contact him at the Bryan School of Business and Economics, Univ. of North Carolina, Greens- boro, Bryan Bldg., Rm. 368, PO Box 26165, Greensboro, NC 27402-6165; [email protected].

January/February 2004 IEEE SOFTWARE 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.