The Economics of the Hackers' Revolution

Permanent Link:

Material Information

Title: The Economics of the Hackers' Revolution How Open Source Challenges Notions of Innovation and Value in the Age of Free
Physical Description: Book
Language: English
Creator: Davidson, Jonathan
Publisher: New College of Florida
Place of Publication: Sarasota, Fla.
Creation Date: 2012
Publication Date: 2012


Subjects / Keywords: Open Source
Creative Commons
Free Rider
Intrinsic Motivation
Genre: bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: This paper addresses the open source development model of information goods production. Features of its non-proprietary and free nature appear to clash with economic and legal assumptions, so this paper aims to reconcile them by reflecting a paradigmatic shift in intellectual property ownership and use, sparked by the user-innovators called hackers. Contrary to the notion of creating more incentives by restricting intellectual property, rights in a collaborative and modern information goods economy can be overly restrictive and reduce production, leading us to consider alternative ownership styles with software as a public good. As open source communities evolve, altruistic motives give way to other intrinsic motivations such as enjoyment and learning, transforming antagonism towards commercial enterprise into healthy competition of similar-quality products. Not surprisingly, governments show increasing interest in the open source sector, both as an alternative product and a production and innovation model. This paper argues that governments and economists should not reject open source out of hand, for although on its face threatening, open source is an example of creative destruction that commercial firms have adapted to and now contribute to.
Statement of Responsibility: by Jonathan Davidson
Thesis: Thesis (B.A.) -- New College of Florida, 2012
Bibliography: Includes bibliographical references.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The New College of Florida, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Local: Faculty Sponsor: Coe, Richard

Record Information

Source Institution: New College of Florida
Holding Location: New College of Florida
Rights Management: Applicable rights reserved.
Classification: local - S.T. 2012 D2
System ID: NCFE004567:00001

Permanent Link:

Material Information

Title: The Economics of the Hackers' Revolution How Open Source Challenges Notions of Innovation and Value in the Age of Free
Physical Description: Book
Language: English
Creator: Davidson, Jonathan
Publisher: New College of Florida
Place of Publication: Sarasota, Fla.
Creation Date: 2012
Publication Date: 2012


Subjects / Keywords: Open Source
Creative Commons
Free Rider
Intrinsic Motivation
Genre: bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: This paper addresses the open source development model of information goods production. Features of its non-proprietary and free nature appear to clash with economic and legal assumptions, so this paper aims to reconcile them by reflecting a paradigmatic shift in intellectual property ownership and use, sparked by the user-innovators called hackers. Contrary to the notion of creating more incentives by restricting intellectual property, rights in a collaborative and modern information goods economy can be overly restrictive and reduce production, leading us to consider alternative ownership styles with software as a public good. As open source communities evolve, altruistic motives give way to other intrinsic motivations such as enjoyment and learning, transforming antagonism towards commercial enterprise into healthy competition of similar-quality products. Not surprisingly, governments show increasing interest in the open source sector, both as an alternative product and a production and innovation model. This paper argues that governments and economists should not reject open source out of hand, for although on its face threatening, open source is an example of creative destruction that commercial firms have adapted to and now contribute to.
Statement of Responsibility: by Jonathan Davidson
Thesis: Thesis (B.A.) -- New College of Florida, 2012
Bibliography: Includes bibliographical references.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The New College of Florida, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Local: Faculty Sponsor: Coe, Richard

Record Information

Source Institution: New College of Florida
Holding Location: New College of Florida
Rights Management: Applicable rights reserved.
Classification: local - S.T. 2012 D2
System ID: NCFE004567:00001

This item is only available as the following downloads:

Full Text


The Economics of the Hackers' Revolution: How Open Source Challenges Notions of Innovation an d Value in the Age of Free By Jonathan Davidson A Thesis Submitted to the Division of Social Sciences New College of Florida In partial fulfillment of the requirements for the degree Bachelor of the Arts Under the Sponsorship of Richard Coe Sarasota, Florida May, 2012


ii ACKNOWLEDGEMENTS Special thanks to Professor Richard Coe for sponso ring this thesis, allowing me to follow my passions for open source and providing th e guidance and edits that this thesis required to bloom. This thesis also might never hav e been without Dr. Catherine Elliott, who sponsored my first January inter-term ISP (inde pendent study project) on the subject and nurtured what were merely budding interests. My attendance at graduate library and information science programs stems entirely from my interests in open source, for which I am content and thank both of you from the bottom of my heart. I would also like to thank Professors Patrick Van Horn and Duff Cooper for serving on my committee and offering insights that improve my thesis in its final form. For her constant and loving support, I humbly ackn owledge my NCF sweetheart, Rachel Atwood. We go our separate ways, but I trust we’ll be there for one another when it counts.


iii Table of Contents Acknowledgments.................................... ................................................... .................ii Abstract........................................... ................................................... .........................v Introduction ................................................... ................................................... .........1 Thesis Map......................................... ................................................... ..................9 Chapter 1 ................................................... ................................................... ............14 Public Goods Literature as Gateway to Discussion... .............................................14 Software as a Public Good?......................... ................................................... .......17 Why the Tragedy of the Commons Does Not Apply...... ..........................................18 The Tragedy of the Anticommons..................... ................................................... ...22 Software as Intellectual Property—Copyrights and Pa tents....................................25 “Copyleft” as the Happy Medium?.................... ................................................... ...27 Open Source Arising from Copyright Strengthening... ............................................28 Licensing Applications............................. ................................................... ............37 Why Patents Still Matter—Future Restrictiveness.... ...............................................39 Cheating the System................................ ................................................... ...........41 Formal governance by scrutiny in courts............ ................................................... .42 Are Sharing Licenses Contractual?.................. ................................................... ...45 What a Non-contractual License Means to the Commons ......................................47 Restructuring Copyright to Account for Sharing Lice nses.......................................49 Chapter 2 ................................................... ................................................... ............51 Intangible Assets--Organizing and Generating Surplu s without Market Exchange..51 Reframing Profits and Costs........................ ................................................... ........52 Opportunity Costs Approximate Costs of Production.. ............................................53 Costs to Programmers vs. Costs to Firms............ ..................................................5 7 Investments--Capital and Human..................... ................................................... ...59 Human Capital Investments and Sunk Costs........... ...............................................62 Intrinsic v. Extrinsic Motivation.................. ................................................... ...........66 User Need.......................................... ................................................... .................69 Caveats to Direct Action........................... ................................................... ...........72 Learning and Community-Based Motivations........... ...............................................74 Enjoyment-Based Motivations........................ ................................................... .....75 Chapter 3 ................................................... ................................................... ............80 Where Do We Go From Here?.......................... ................................................... ..80 Value-Added Service Enabling....................... ................................................... .....81 Loss-Leading....................................... ................................................... ................83 Dual Licensing..................................... ................................................... ................84 Leveraging Collective Software Development......... ................................................86


iv Gated Source....................................... ................................................... ...............88 Pro-Social Incentives.............................. ................................................... .............92 Transition to Government Policy.................... ................................................... ......93 Imperfect Information, Market Failure.............. ................................................... ....96 Mandates, Strong-arming Adoption................... ................................................... ..98 Government Funds, a Third Policy Arm.............. ................................................... 99 Static versus Dynamic Measures (of Innovation)..... .............................................103 Challenging the Open Source/Commercial Binary...... ..........................................106 Conclusion ................................................... ................................................... .......107 Bibliography....................................... ................................................... ...................115


v The Economics of the Hackers' Revolution: How Open Source Challenges Notions of Innovation an d Value in the Age of Free Jonathan Davidson New College of Florida, 2012 ABSTRACT This paper addresses the open source development mo del of information goods production. Features of its non-proprietary and fre e nature appear to clash with economic and legal assumptions, so this paper aims to reconc ile them by reflecting a paradigmatic shift in intellectual property ownership and use, s parked by the user-innovators called hackers. Contrary to the notion of creating more in centives by restricting intellectual property, rights in a collaborative and modern info rmation goods economy can be overly restrictive and reduce production, leading us to co nsider alternative ownership styles with software as a public good. As open source communiti es evolve, altruistic motives give way to other intrinsic motivations such as enjoymen t and learning, transforming antagonism towards commercial enterprise into healt hy competition of similar-quality products. Not surprisingly, governments show increa sing interest in the open source sector, both as an alternative product and a produc tion and innovation model. This paper argues that governments and economists should not r eject open source out of hand, for although on its face threatening, open source is an example of creative destruction that commercial firms have adapted to and now contribute to. X _______ d Richard Coe, Thesis Adviser s


1 Introduction A subtle tremor is reshaping the continent of busi ness in new and exciting ways. The digital economy, the age of information, the In ternet generation--all have been used to describe the transition of the last few decades to account for the dynamics of a fastchanging, increasingly interconnected world of info rmation. Communication industries have exploded in leaps and bounds during the digita l revolution. As the American economy shifts away from manufacturing into a servi ce industry, the new major commodities are information goods--the software tha t aids an accountant, the online publishing of a professional writer, the receipts g enerated by a cashier that are incorporated into the database of store managers. M arket forces are affecting basic tenets of business practice, as outsourcing becomes an imm inent concern to the United States labor force’s competitiveness, driving prices down. Value is being reexamined, and worthwhile services like email hosting are regularl y given away free of charge. In this context, economists are always quick to po int out that There Is No Such Thing As a Free Lunch. Even donated time is worth s omething, but communication industries diminish the costs associated with produ cing and marketing through the Internet. With increasing ease, individuals connect themselves to the resources and social structures that enable them to transcend being mere consumers and to provide value back. Traditional distributors of content are learning to harness the latent energies of consumers through social networking, and more businesses are operating as organizers of userproducers through communities like YouTube or Wikip edia. User-producers are not the only audience of the new frameworks, as copyrighted works are regularly and sometimes illegally posted on YouTube and Wikipedia articles on arcane subjects are often lifted


2 verbatim from other sites, but rampant piracy does not overpower the dynamic of legal and collaborative exchange of information. Works be get future intellectual property when they teach and inspire others, highlighting how the common resource of knowledge benefits all. Arguably the larger movements, mentioned above, al l originate from an innovation paradigm called open source, the subject of this thesis. Open source is the collective efforts of software writers who freely d istribute their works and place few restrictions on how code is used, adapted, or sprea d. Computer programs operate on a machine-language code, while programmers compose it out of human languages, primarily English, and translate it afterwards into the binary that computers understand. Open source software (OSS) makes its human-language source code freely available in addition to the standard binary machine-language co de, while proprietary companies traditionally guard source code as corporate secret s. At the most basic level, open source inherits its name from this technical divergence, b ut the movement superseded the definition of simple access to the source code. Ele ments of free and unlimited redistribution of source code are incorporated into OSS, for the original code base as well as modifications and derived works. In turn, the or iginal author must give up rights to restrict his code to particular technologies/hardwa re or to uses, and individual users are granted the same rights to redistribute as any othe r vendor (Open Source Initiative). Empowering users to self-innovate proves to be an e xtremely lucrative reward system for an innovation model, but it challenges standard eco nomic principles involving payment structures for organizing production which must be addressed.


3 Occasionally described as a gift economy, what doe s it mean that there is a sustainable system of individuals producing things of substantial value only to offer it away freely? More intriguing still, what happens wh en you give away not only the product, but the means of production? Instead of be ing gifted a word processing program as an immutable “black box,” receiving the ability-through source code--to improve its functionality or reproduce it for another person is undoubtedly of value in addition to the program itself. These intriguing questions inspire the economic analysis of open source computer programs that follows. By examining the op en source movement, we can start to unravel the paradoxical practices of relinquishi ng the control granted to intellectual property creators and how these practices and motiv ations form a community of sharing that accounts for the interdependence of a modern i nformation society. As the sharing culture expands beyond code to othe r forms of intellectual property, open source software projects exemplify a n innovation model that does not rely on traditional market mechanisms such as a price sy stem nor on formal hierarchies or alliance agreements of a command-focused economy (O sterloh and Rota 2007, 157). Although open source does not appear to follow the allocative signals of a price system, it retains the freedom of markets in enterprise and ch oice: individuals and private business are free to produce open (and closed) source softwa re as they see fit (McConnell et al 2011, 31). Work on open source is not compulsory in any fashion and largely selfdirected, because forcing anyone to volunteer on so mething they do not care for requires massive enforcement costs. Enjoyment is crucial to the volunteer basis of open source contributions and concepts like fun and “playful cl everness” aptly describe characteristics and attitudes embraced by hacker participants (Mood y 2001, 16).


4 Coders solve their own coordination and incentive p roblems, as this thesis demonstrates, by self-selecting their work in ways that embrace interesting but nonstandard facets of economic theory. Some critics qu estion whether open source software development is actually revolutionary or largely a myth constructed and perpetuated by early, hopeful participants (Fitzgerald 2006, 588). Either way, it is not myth without basis because even what I have described above is signifi cantly different from the normal commercial models that stress secrecy and keep cons umers from simultaneously becoming innovators. At a minimum, sociologist Chri stopher Kelty convincingly identifies how many exciting models--such as Wikipe dia, free culture, Open Access, and peer production of research--are all linked to the widely felt reorientation of knowledge and power that starts with open source and its pred ecessor, free software (Kelty 2008, 301). Whether or not it is an entirely new paradigm is debatable, but it is certainly worth exploring in order to capture the benefits of alter native styles of information production. Von Hippel and von Krogh (2003) are among the auth ors who take an optimistic stance by offering a new “private-collective” model of innovation observed in open source development. An exact definition for the ter m never forms, but essentially it is the acknowledgement that private investment generates p rovision of a public good, which normally only occurs by collective action. The clos est von Hippel and von Krogh come to a definition of “private-collective” follows: Participants in open source projects use their own resources to privately invest in creating novel software code. I n principle, these innovators could claim property rights over the cod e, but instead they freely reveal it as a public good. (213)


5 Just as the term “private investment” need not but does assume returns on investment through sale of proprietary information, the term “ private-collective” is more a placeholder that von Hippel and von Krogh crafted t o distinguish their model from the others. They propose that the compound model brings the best elements of private investment and collective action together, a coexis tence of--or moderation between--the models. In fact, however, a scholar named Robert C. Allen (1983) started documenting cases of “Collective Invention” in midto late-nin eteenth century England which closely correspond to the innovation arguments surrounding OSS development, leading to questions about the novelty of this model. Collective invention is another case of privately allocated resources that are utilized by agents other than the firms that discov ered particular innovations, suggesting a free rider problem where the optimal strategy might be to merely incorporate competitors’ innovations freely without using the f irm’s resources on innovation. Allen suggests that this problem did not arise in the cas e of blast furnish designs in England’s Cleveland district between 1850 and 1875, which all owed the formally uncoordinated improvements in size and temperature of the furnace s. His answer to the dilemma was that inventions/improvements were too incremental-not ‘novel’ enough to receive patent protections--and so the accumulation of technical k nowledge was a byproduct of experimental design tweaks that were part of normal business operation (Allen 1983, 23). The modern software industry suffers almost the opposite problem, where grants of patents seem too ready to monopolize non-innovative steps. Therefore, I intend to show that this makes creating the commons more important as a check against inefficiently restrictive patents. Prior art of code in the commo ns can reverse the trend of Cold War-


6 style stockpiling of patents by tech companies like Microsoft, Google, Apple, and Motorola, that can escalate to mutually assured des truction (MAD) litigation strategies (McMillon 2012). There are several other aspects of Allen’s work tha t echo the motivations association with contributing to open source, creat ing historical precedent for the modern movement. First, profits of proprietary information may be forgone in order to advance professional ambitions that only accrue to those sh aring information, so it behooves one to play the game even if it requires showing one’s cards, as was the case for Allen’s furnace engineers. Open source literature acknowled ges the reputation rewards of contributing to important projects, benefiting deve lopers socially and economically beyond mere use of the program. Relatedly, the cost s of keeping information secret may be large in comparison to the benefits of sharing, because contractors building the furnaces by trade learn the non-patentable designs. Code can be patented and copyrighted, but the benefits of sharing are appare nt in how open source communities foster reciprocity, for the more people opening the ir software, the more common resource of code and programs are available (17). Because th e primary cost of knowledge is in its creation and not its replication, innovation is a f ixed cost with benefits accruing across output, and small firms benefit less from particula r innovations than larger or more consolidated firms because by definition they produ ce less. Therefore, over time, collective invention becomes overshadowed by propri etized Research & Development departments as oligopolies become more predominant in twentieth century markets through mergers (21).


7 Given Allen’s conclusions, then, it is anomalous th at open source emerged in the latter half of the twentieth century and stuck. Ost erloh and Rota establish a pattern of collective invention occurring during an exploration phase of a new technology in which it is not well understood, such as the statistical and theoretical limitations of furnace engineering in nineteenth century England, followed by an exploitation phase as the technology matures (2007, 162). The university mode l of shared research lay at the heart of computer science, and academics regularly traded programs and theory freely in the 1960’s and 70’s. Stanford’s Homebrew computer club was a perfect example of an open community sharing and exploring the potential of ap plications for microprocessors, attended by Apple founders like Steve Wozniak. Howe ver, as the opportunities to proprietize information arose, sharing stopped as p revious collaborators transitioned into competitors, and the club atrophied and then disban ded entirely in 1986 (161). The mid-1980’s continue to be a pivotal time period for this thesis, as the seed of an open source paradigm starts to take root. After copyright law reflected explicit protection of software code in 1980, firms gained u nprecedented means by which to control and profit from their code, sparking the ex ploitation phase of the computer industry. A community of trust among software users --primarily academics, because computer science was not well established in civili an life--received jolts in 1984 when AT&T seized back intellectual property control of t he UNIX operating system. Up to this point, antitrust laws kept AT&T from fully exercisi ng monopoly control and thus UNIX previously functioned as a jointly developed projec t between the firm and academia (Osterloh and Rota 2007, 164). Appropriation concer ns over UNIX directly birthed the free software movement preceding open source; MIT a cademic Richard Stallman took an


8 ideological stance in trying to create a replacemen t, UNIX-like operating system named GNU, a self-referential acronym stating “GNU’s Not Unix” (Moody 2001, 20). To counteract the proprietization/exploitation phase, Stallman founded the Free Software Foundation to support software that gave rights bac k to users rather than owners, creating the distinction between free as in gratis, versus liberty --“free as in beer and free as in speech,” respectively. Today, concepts of liberty s till exist in the term free, but freeware is also a generic term for software distributed wit hout cost to the user and, thus, open software is more descriptive. An all inclusive term free and open source software (F/OSS) is seldom used in this thesis except to ret ain sources’ original meaning. Stallman’s protest evolved into the response of a c ommunity that permitted the collective invention regime to survive the transiti on from exploration phase to exploitation. Elements of the idealistic and rebell ious spirit of contributing to free and open systems remain, but the open source paradigm n ow stands apart for its economic efficiencies. The value of open source software was not immediately apparent, more a hobbyist’s dreams and a hacker’s ideals than a legi timate movement, but that changed in the 1990’s. Open source’s quality and scalability, from personal computers to enterprise machines, is evidenced by the number of commercial development tools and software used by companies like IBM, Hewlett-Packard, Red Ha t, Oracle, and the late Sun Microsystems, now merged with Oracle (Mockus et al. 2002, 331). Study of open source began in earnest following a 1999 essay by programm er Eric Raymond titled “The Cathedral and the Bazaar,” drawing a large distinct ion between the commercial method of producing code and the free-flowing but ultimate ly successful open source model of volunteers. Raymond asserts that every good project starts with a developer scratching an


9 itch, giving programmers (rather than marketing dep artments) the ability to drive development in an informed manner (25). From this f unctionality comes an initial prototype that can be improved by subsequent contri butors to fine tune and expand the programs into things capable of rivaling commercial products. Thesis Map As a new production paradigm, it is extremely impo rtant to understand how open source communities and practice work, developing al ongside the Internet and hacker culture and pushing the boundaries of business. The refore, the first two chapters devote themselves to integrating the paradigm with modern economic theory. The first chapter of this thesis traces the movement through its earl y and ideological beginnings under Stallman’s Free Software Foundation in the first ch apter, establishing a framework for the unusual sharing structure that disregards tradi tional methods of extracting value from intellectual property like software. In response to the restrictive structuring of traditional copyright law, Stallman’s GNU licensing selectively relinquishes elements of copyright control to allow users considerable liberties of th e use of software code, costless redistribution of said code, and the ability to alt er it as the users see fit. Because of its vastly different philosophical appro ach of encouraging the common resource of intellectual property rather than priva tization, the GNU license encourages a sharing regime that is applicable beyond software, into books, music, hardware design, etc. I take the opportunity to explore the greater potentials of treating information as a public resource and to explore the legal complexiti es that shape the GNU license and descendant licensing schemes. Many scholars observe the legal sophistication of hackers and open source participants as a group, a necessar y trait for traversing the legally


10 complicated field of software production and use. I ntellectual property rights are not static, so there is both opportunity and responsibi lity for managing rights with deftness and care while establishing the open source paradig m. While the first chapter sets up the general dynamic s of sharing and points to the applicability of a sharing regime to other intellec tual property, its solid theoretical work is only a foundation and does not fully explain OSS’s success. Nothing influences the macroeconomy without actions taken on the microor individual-level, and having a framework by which to contribute to a commons does not explain why anyone is so inclined. Therefore, the second chapter adopts exac tly this approach by examining the motivations of contributors on a more practical lev el, less altruistically and idealistically. When structuring community, Stallman’s idealism and altruism surrounding free software give way to a new mythos surrounding Linus Torvalds and the Linux operating system, based on pragmatism. Thanks in large part to Torval d’s elevation to heroism by Eric Raymond’s “The Cathedral and the Bazaar” (1999), Li nux development becomes the poster child of successfully organizing individuals into the contributions of large open source projects. Torvalds intuitively grasps the lo gics behind intrinsic motivations of hackers, providing them with a workable prototype s oftware that can be incrementally improved and rewarding or encouraging their involve ment by making their contributions tangible and instantaneous. The final chapter shifts its focus towards the futu re, from explanation to implication. Producing information and code is expe nsive, especially if every firm develops everything independently, so most successf ul technology firms are learning how to take advantage of common resources of informatio n. In contributing to open source,


11 they are also giving back, so a virtuous circle pro mises some efficiency from determining what sorts of information can be public without har ming businesses. Open source contributors are often employed by businesses and e ven tasked with helping develop open source at work, while many are volunteers on t heir own time/dime. The third chapter lays out the methods of creating value from commercial involvement in open source, primarily through selling either support se rvice or complementary products. IBM, for instance, can sell servers more cheaply and com petitively with free, open source software powering them than is possible if IBM deve loped the software independently. Network effects often sway technology and software markets into winner-takesall mentalities, and open source has the potential to always inject competition into the market by the nature of its noncommercial propertie s, offered at a zero price. As copyright protection periods historically lengthen, open source may be the push companies need to continue innovating instead of be coming complacent while consolidation monopolizes the software market. Of c ourse, open source competition does not imply second best, as many open source projects are enormously important and are the dominant software examples of their type; Apach e web server software hosts the majority of web sites on the Internet, and Sendmail handles email as a “category killer” with no commercial alternative (Lakhani and von Hippel 2003, 925; Must onen 2003, 101 footnote 3). Such programs are likely to remain dom inant as long as they keep up with innovation, because proprietary competitors will fi nd it difficult to recoup development costs on proprietary alternatives when competing wi th open source programs’ free cost (Olson 2008). However, these successes contribute t o the identification of open source products with technical, back-end applications rath er than consumer-friendly user


12 interfaces, yet strides in the last decade indicate that this is a default preference rather than prerequisite. Finally, governments are the largest single consume rs of software and stand to gain greatly from treating software as a public goo d. Many European governments already claim support of open source, and smaller n ations like Peru and Singapore are proposing far-reaching legislation to institute str ategies to use Linux and F/OSS, by offering tax breaks and other economic incentives. Presumably these strategies limit dependence on foreign programmers and encourage dom estic programmers to influence software development and gain some proficiency (Eva ns and Reddy 2003, 378-80). Because a significant amount of the cost of good pr ograms is the human capital invested in training programmers, the government subsidizes a fair amount of the cost in addition to purchasing licenses of proprietary software. The refore, the thesis concludes by exploring the government’s current role in open sou rce as well as what role it could have if governments adopt or support OSS. Government inv olvement could manifest in many ways, not least of which by changing the intellectu al property laws that sparked Stallman’s movement. Several valid concerns present why government should refrain from official sponsorship, including not being the best judge of emerging technologies or destroying the dynamics that make open source a goo d environment for volunteers. Measuring these pros and cons helps show where open source has room to grow and where it is better left to its own devices. Ope n source is only gaining traction, but that does not mean that efforts to propel the movement b eyond its own limits is beneficial. Enthusiasm for new models cannot erase the economic benefits bestowed by the traditional intellectual property regime of copyrig hts and patents, but there is also no


13 reason to ignore critiques or improvements to the r egime. In an increasingly interconnected and complex business environment, es tablishing a commons of information may be more equitable and perhaps a bet ter platform for future growth.


14 Chapter 1 Far removed from the cottage industry style of prod uction of single households or artisans, the modern economy relies upon a highly c ollaborative nature of production that invites some new ways of thinking about property. T his chapter explores how software sparked a revolution for pluralistic ownership that has enormous implications for software’s production. Software has special propert ies that require an understanding of different types of goods, reflected in the economic theory that marks the first half of this chapter. After grounding the thesis in theory, it b ecomes possible to move into the legal reality of open source: to examine how it operates and whether this is a good and sustainable system. However, because open source is so new and unusual, that legal reality is neither clear-cut nor immutable; there i s room to consider how it might be improved and legitimized. Public Goods Literature as Gateway to Discussion There are a number of select goods and services tha t private markets will fail to provide adequately, because sometimes market prices do not adequately account for all social costs or benefits. Economic literature ackno wledged externalities for many generations, famously Arthur Pigou’s analysis in th e early 20th century. Pollution is a classic social cost affecting sizeable third partie s not immediately benefiting from the transaction of producers and consumers, so these co sts represent negative externalities on society. But occasionally there are benefits that t hird parties stand to gain by the transactions, or positive externalities, where more production means that society as a


15 whole is better off than without accounting for the externalities. One classic example for a positive externality is education. Positive externalities help justify the government’ s provision of education (at least at the elementary and secondary levels) because sta tistically there is an inverse relationship between education and crime suggesting that educating the populace leads to a safer environment for everyone. The private marke t, then, is not the only mechanism through which goods and services are provided in a modern political system. The terms collective goods and public goods define those good s and services provided for citizens from a government looking for a just and equitable distribution of resources, often financed through taxes (McNutt 1996, 178). Governme nts traditionally address private market failures springing from micro-level assessme nts of value, by taking the broader view of societal or macro-level valuations. As a wh ole, this thesis aims to demonstrate some of the beneficial effects of open source code that anyone can learn from and rewrite, viewing the open source community as an ex ternality to another type of development process. To underscore the contrast of the classic examples of private and public goods, economists present the distinguishing features of e xcludability and rivalry. Many items are consumed when used, like cheeseburgers, so one person’s use precludes another from consuming the same product. Resultant rivalry and c ompetition over who uses the goods are well allocated by private market prices where i tems go to those who place the greatest value in them and are willing/able to pay. There ar e exceptions called nonrivalrous goods which continue to exist for multiple users. Ideas a nd innovations, such as those organizing and empowering software programs through its code, are intangible and


16 therefore completely nonrivalrous. Highways are als o quintessential public good examples, even though they ultimately suffer from c ongestion or gradually depreciate over time and use. Software, and particularly the o pen source variety discussed in this thesis, constitutes a public good in the classical sense because it is both non-rivalrous and non-excludable (Osterloh and Rota 2007, 158). Highways are expensive to pave across an entire nat ion spanning thousands of miles, representing large upfront costs that govern ments can bear better than independent companies. Upfront costs theoretically spread over the many people using the roads or benefiting from less costly transportation of goods and millions of traveling cars cause asphalt to wear very slowly therefore translating t o miniscule marginal costs. Software production, differentiated from code production, is similarly characterized by high fixed costs and low marginal costs that make it an ideal candidate for public goods provision at first glance. However, before the national highway systems, there were privately owned turnpikes and toll roads, at least hinting towards another type of good between the standard public and private goods. Originally, economists only paid attention to pure public goods and pure private goods, but since Paul Samuelson’s work in the 1950’ s, the literature started acknowledging a category in-between known as impure public goods (Cornes and Sandler 1986, 3). One way to conceive of the logic of impure public goods is to use the alternative name of club goods: groups can exclude outsiders from use of the nonrivalrous good, common to those with membership in the club (McNutt 1996, 182). This broadened category of goods--for which collect ive provisioning is not essential to production but potentially beneficial--permits me t o explore whether software could be


17 optimally produced as a public good. Commercial com panies regularly produce excludable software, and they control use through c lubs where membership is determined by licensing intellectual property. Software as a Public Good? Since most tech companies charge for their products rather than giving software away free, experience proves that much software is a private, proprietary good. The business computing revolution owes much to the netw orking effects resulting from ubiquity of a unified operating system from Microso ft and its office suite of productivity programs for a range of PC systems. Microsoft’s suc cess at securing market share made former CEO Bill Gates one of the richest men on the planet (throwaway citation needed). This historical trend of proprietary software conti nues to support the US economy as other countries import more software than they expo rt to the United States (fact check and citation needed). Yet, corporations make health y profits despite the fact that there is nothing inherently rival about consumption, one cop y being sufficient to install to any number of computer systems. Indeed, a quick install from a CD can replicate the value to consumers on a new machine at no further cost, for hard drive space is a sunk cost to the consumer/user. The Internet further reduces the difficulty of passing along the code, eliminating the need to physically transfer information on a disc or using limited shelf space for advertising. Costs of hosting servers for consumers to connect t o in order to collect the code are comparatively small to the former costs of sending CDs to consumers, creating efficiency as more households access the Internet. Sharing is often so easy and practically costless that piracy sites host material at no cost to the d ownloading party and do not even require


18 membership. So, how does Microsoft, or any propriet ary software company, make money from this? By holding the rights to the intellectua l property of software, companies retain legal control over how their property is used and b y whom, extracting rent from consumers in the form of licensing payments. By examining open source software through the lens of externalities, this paper argues that positive externalities of exchange and free use of code are worth capturing despite the competition of free software against pr oprietary software. This chapter lays out the logic of considering code as a public good rather than a private one, detailing the implications for how focusing on software as goods freely available to all increases the ultimate value of such goods across the community o f sharers. Intellectual property rights enable private markets to function by allowing prog rammers to capture returns from their work, but an innovation in copyright law coupled wi th the powers of the Internet to distribute almost costlessly permits a type of publ ic production and provisionment that might transform the proprietary model. Later chapte rs address the sustainability of such a sharing model by applying a more micro-level approa ch to the incentive structures facing open source contributors. The third substantive cha pter explores potential government subsidization in considering whether the club good status of current commercial development is optimal. The rest of this chapter lo oks at the macro-level implications by examining just how open source communities operate in relation to proprietary models in intellectual property management. Why the Tragedy of the Commons Does Not Apply Public goods literature routinely focuses on the di fficulties in optimizing consumption of that good. If software were treated as a public good, for the reasons given


19 above, a necessary question is whether any problems will arise from over-use or underprovisionment. This is a complex question that reve als particularly interesting analysis in relation to open source software. One standard dile mma introduced in public goods literature is the scenario of the tragedy of the co mmons: every agent has incentives to use the public good as much as one can, but in a standa rd zero sum game with consumable goods this consumption precludes others from also c onsuming. As an illustration, consider that many sheepherders share common pastur es that each can bring sheep upon to graze. Without regulation, no sheepherder can pr event others from bringing in more sheep which deplete the common pastures of edible g rass, so that each add sheep at the other’s expense. Too many sheep eat the grass to th e roots and seeds, depleting the commons for the next cycle. In the next cycle, fewe r sheep overall can be fostered, representing a net loss to society. Information goods such as software avoid the classi c tragedy of the commons in that they are non-rivalrous goods, where one person ’s use does not preclude another’s use. Arguably, more people using the programs will actually generate network effects, or more people using one type of free product can crea te economies of scale by spreading costs across more consumers. The open source moveme nt in software featured novel dynamics in the free use, distribution, and modific ation of programs that have since spread to other forms of digitally mediated intelle ctual property. Websites, in particular, aggregate or distribute these public works, be they computer code or photographs or writings, in the process organizing it into a commo ns that anyone may make use of. Defining a commons that any creative agent may draw upon for materials allows for larger, more collaborative, and more complex pieces


20 Where the intellectual commons is composed of ideas and innovations rather than grass, old ideas will always be available but new o nes may not develop without proper incentivization. The fear is that stifling innovati on on the pasture elements further threatens the dynamic interaction of ideas proposed for feeding the sheep/creative agents. As the economic history of the United States attest s, innovation is extremely important to any economy’s growth and long-term success. Recogni zing this early on, the country’s founders were concerned with adopting the best prac tices for stimulating ideas. Section 1, article 8 of the United States Constitution issues Congress the power “[t]o promote the Progress of Science and useful Arts, by securing fo r limited Times to Authors and Inventors the exclusive Right to their respective W ritings and Discoveries” (Heffman 1997, 1495). Within the Constitution’s grant of pow ers lies an assumption that authors and inventors require incentives to create, and tha t the “exclusive Rights” give them time to extract monetary value for their efforts. Economists view these exclusive rights as temporar y monopolies, bestowed as either patents or copyrights depending upon the nat ure of the work. Because the criticisms leveled against the intellectual propert y rights management above have broadly applied to both patents and copyright, there has be en little formal distinction between the two up to this point. Although different rules appl y to and govern each, they function under the same basic principles: property rights un der either system, for instance, permit the sale or licensing of intellectual property. The se traditional methods allow the appropriation of creative energies put into produci ng the grass/ideas, but it limits the power or utility of the grass because it allows the m to exclude some sheep/creative agents from using the inexhaustible blades of grass alread y created.


21 The traditional system of intellectual property ri ghts management in patents or copyrights foresaw these difficulties to an extent and created the public domain. Much like the intellectual commons, anyone may draw upon the ideas and innovations set aside for the public domain in creating new ideas and inn ovations. Indeed, whenever the copyright or patent expires, those intellectual pro perties instantly revert to the public domain. Yet, a problem persists because technology has characteristically rapid development and implementation, thereby diminishing the value of waiting for the expiration of IP protection and in fact leaving ver y little hope for ever catching up with the progress. I would argue that the same applies t o other fields like music, which sprouts new genres faster than copyright expires. At the very least, the public domain is an inadequa te response for promoting the progress of the sciences and useful arts as directe d by the Constitution. Because of how intellectual property claims are structured, merely releasing a work into the public domain does not ensure that another cannot appropri ate the work into another commercial product, thereby limiting the commons. Part of the strength of a public goods-style approach is looking at all of the dynamic possibili ties of different ideas feeding into new products and ideas that grow the commons rather tha n merely make up for the loss of appropriation. For instance, if a program were rele ased into the public domain and someone modified the program and made it better, th at person could establish proprietary rights to the modified version and prevent others f rom using the modified program. If the modifications are very strong, this can divert prod uction away from the commons and reestablish the club goods approach of private prov isions through excludability. The end


22 of this chapter will explore how open source effort s stem this outflow of good ideas from the commons to proprietary efforts. The Tragedy of the Anticommons If overuse of a software commons is not a problem, under-providing can be. Whether software as a public good is underprovision ed can be interpreted in two ways. First, standard economic theory and the copyright a nd patent systems certainly suggest that no rational agent would provision software des ignated as a public good, citing classic free-rider collective action strategies. Rather tha n expend energy and effort towards provisioning software, all would benefit from waiti ng for someone else to write code and merely use the program that others designed. The ne xt chapter challenges this assumption extensively by empirically finding that open source programmers do participate in collective production of software and explores cont ributors’ motivations. This chapter primarily lays out how open source operates, only t aking why into account in the big picture sense of maximizing the commons. Scholars Michael A. Heller and Rebecca S. Eisenberg (1998) noticed a different, second issue with underprovisionment resulting from the structure of property rights in the biomedical field. Biomedical research relies he avily on prior research and patents; when too many intellectual property owners hold rig hts in previous discoveries, obstacles in obtaining those rights legitimately made future research brutal and slow. Holding a dark mirror up to the metaphor of the tragedy of th e commons above, in which a lack of property rights caused commons to be overutilized a nd depleted, Heller and Eisenberg proposed the tragedy of the anticommons to highlight the underutilization resulting from too many property rights.


23 “The problem we identify,” Heller and Eisenberg wri te, “is distinct from the routine underuse inherent in any well-functioning p atent system... The tragedy of the anticommons refers to the more complex obstacles th at arise when a user needs access to multiple patented inputs to create a single useful product.” For instance, high transaction costs finding the owners of patents or making a dea l with each of them could mean that certain productive activities are avoided. Introduc ing anecdotal evidence of searching in the Lexis patent database for the fairly specific t erm “adrenergic receptor” returned more than 100 results, presenting a challenge in determi ning exactly which would need to be licensed for any given project. Some universities a nd firms started exercising rights on intellectual properties even before the USPTO offic ially granted the patents. In general, they concluded that projects seeming to require fewer patents would undoubtedly be preferred and that potential patents created the specter of rights that may be even larger than the actual rights (Heller and E isenberg 1998, 699). Although Heller and Eisenberg believe their anticommons to be fairl y specific and apply to biomedical patents rather than all patents or copyrights, and they do not wish to be misinterpreted as pseudo-critics of intellectual property rights mana gement, I argue that a broader view is called for because there are issues cropping up in other fields. Technology firms today develop portfolios of broad, suspect patents in ord er to sue and counter-sue opposing firms, erecting barriers of entry against new firms with insufficient capital to withstand lawsuits. Oligopolistic power may develop, and taki ng any firms to court in order to dismantle bogus portfolios costs a lot more money t han constructing them does. Librarians have also brought attention to the diffi culties in finding copyright owners though programs like Orphan Works, which try to det ermine protocols and processes for


24 determining ownership in cases when corporate owner s go out of business or individual owners die without making provisions for the work. To provide some scope to this claim, the Orphan Works FAQ cites the executive director o f HathiTrust Digital Library estimating that 50% of the volumes in its digital l ibrary may be orphan works (University of Michigan Library). Bargaining, even when intellectual property owners are found, can fail for many more reasons than collective action problems with p atent owners holding out. Contrary to Nassim Nicholas Taleb’s Black Swan theory about bad events, Heller and Eisenberg believe that cognitive biases consistently cause pa tent owners to “ over estimate the likelihood that very low probability events of high salience will occur,” causing patent owners to overcharge for the patents’ use by ration al expectations (1998, 701, my italics). Owners do not want to lose if their patents are key so they hold out for higher prices that ensure their patents are more valuable to the ones purchasing the rights to use them. Unfortunately, this crowds out more use of the pate nted innovations than cost requires, potentially under-provisioning the IP to those who would use it. In other ways, too, IP owners negatively influence negotiations to develop commercial products. Rather than merely redistributing wealth back to innovators, en ough patents create large transaction costs that actually preclude some avenues of innova tion and production, a Coasian market failure that must be fully accounted for. Ma ny scientists have noted that their professional endeavors require free and open inform ation through peer-reviewed journals that do not hold exclusive publishing and reproduct ion rights. Enter open journals like the Public Library of Science publication that has the express mission of publishing to an unlimited, public readership rather than a club (PLoS). While 1998 may have only


25 provided Eisenberg and Heller hints of the restrict iveness of intellectual property rights management, the first decade of the twenty-first ce ntury highlighted the importance of coining and introducing the “anticommons” to academ ic readership. Against the standard patent and copyright system of appropriation, I con sider open source to be an alternative filling the void between public domain and standard intellectual property management. Software as Intellectual Property—Copyrights and Pa tents In order to provide depth to the analysis that foll ows, it becomes important to outline the standard intellectual property rights r egime practices regarding patents and copyright in more detail before delving into the sp ecifics of open source. Isolating whether computer programs fall under patent or copy right protection can be difficult, because programs seem eligible for protection from either. On one hand, programs are composed of code, which qualify as a unique express ion worthy of copyright protection. Consider the similarities between the written expre ssion of poetry and the meticulously composed code that are operating instructions to a computer. On the other hand, however, programs operate by generating innovative functions on computers and thus seem eligible for patent protection. Program functions are create d with algorithms, however, and neither algorithms nor mathematics are eligible for patenting (Mustonen 2003, 103). Different institutions arose to grant the exclusive rights mandated by the Constitution, and rules differ between the two. The United States Patent and Trademark Office (USPTO) handle patents, awarded for inventio ns or innovations, while the U.S. Copyright Office protect unique expressions such as literary works or pieces of visual art through copyright management (Mustonen 2003, 103). To receive a patent, would-be intellectual property owners need to file at the Pa tent and Trademark Office and pay a fee


26 for consideration. Copyrights, however, are not spe cifically awarded by the Copyright Office and therefore require no special action on t he part of the copyright owner in order to take effect. Protection generates at creation, w hether or not the author of an email or a photographer capturing a snapshot is even aware tha t these works constitute copyrightable material. Copyrighting is both free and lasts longer than pat enting, so it made a logical choice for software to originally pursue protection from. The Copyright Act of 1976 clearly articulated acts constituting infringement of protected properties, including the unauthorized copying of material, so commercial sof tware developers initially sought protection for their goods under copyright. By 1980 Congress amended the Copyright Act of 1976 to officially make programs copyrightab le, cementing copyright’s status as historically and legally significant to open source software development (Heffan 1997, 1496). Patents now affect programming and code1, but an open source community has more difficulty responding to patent protection, fo r reasons this chapter deals with below. At any rate, open source licensing operates exclusi vely on the level of copyright--patents are too expensive for individual volunteers to reas onably apply for on a regular basis-and thus copyright composes the majority of this th esis’s intellectual property system analysis. 1 Diamond v. Diehr, 450 U.S. 175 (1981) deserves the distinction of being the first strong support by the courts for software patenting (Kelty 2008, 2 00). The Supreme Court ruled that although algorithms are themselves not patentable, a physica l machine or process using the algorithm is eligible for patent protection, hence applying to s oftware.


27 “Copyleft” as the Happy Medium? Along a spectrum with the most liberal use of a wor k in the public domain and the most restrictive uses ensured by copyrights, the co mmunities fostering intellectual commons have struck a middle ground often denoted a s copyleft. Copyleft typically establishes strong proprietary rights for the autho r or originator of the work, on par with copyright, but with the intention of relinquishing certain rights, such as allowing users to copy the work without reproach of copyright infring ement. Commentators note the “irony” of using copyright in such an unintended wa y: not to limit distribution of the IP but specifically to promote it freely (Heffman 1997 1507). Copyleft is the subversion of a copyright system of appropriation control into co ntrol for the purposes of sharing. As the similar names indicate, copyright and the pl ay-on-words “copyleft” are closely related. Indeed, copyleft could not exist w ithout copyright law, because the former derives its legal power from enforcement of the latter. Mikko Mustonen writes about the positive correlation between copyright st rength and copyleft adoption as a “paradoxical” relationship: If society does not support a high level of copyrig ht protection the copyleft communities are likely to restrict the distribution of programs to consumers because the risk of copyright violations increases. (2003, 101) If originators of IP gained no more control over ho w their products are used when copyrighted as opposed to being left to the public domain, there would be no incentive for claiming proprietary rights. Therefore, to unde rstand copyleft, it is helpful and worthwhile to consider the history and legal power of copyright as its own entity above.


28 Open Source Arising from Copyright Strengthening Once developers could incontrovertibly invoke copy right law in 1980, a number of rights become applicable to software, protected by statute. Among the most important are the exclusive rights to reproduce the works and generate copies, to distribute copies at their own discretion, and to create derivative work s based on the original (Loren 2007, 280-1). Firms routinely limit the number of copies and devices on which software can be simultaneously installed, sometimes even using DRM to tie the software to a single, particular device. Copyleft communities, meanwhile, grant users of their intellectual properties permission to reproduce the works as muc h as they wish and to distribute copies, traditionally acts of copyright infringemen t. Technological and legal history is extremely import ant to the evolution of open source, because in many cases the programmers who s ocially engineered the new production paradigm were driven by specific needs a nd circumstances that shaped their movement. Even the name--open source--owes itself t o technological and practical necessities. Computers operate in binary, but very few people today can even code in that esoteric and time-consuming way and it is certainly inefficient to do so. Instead programmers rely on compilers to translate their or iginal, “source” code from (primarily English) human language to the binary “machine lang uage.” Learning from or changing program functions in binary is very difficult becau se reverse engineering from machine to human language is impractical (Moody 2001, 27). The profile of open source contributor, commonly known as a hacker2, tinkers with the underlying systems and can best 2 Traditionally, the media has a hard time distingui shing between hackers and crackers which lead to misuse, according to the self-policing Jarg on File (2004). The word hacker generally connotes a “person who enjoys exploring the details of programmable systems and how to stretch their capabilities, as opposed to most user s who prefer to learn only the minimum


29 accomplish this with access to the human language c ode rather than the binary. Importance of human language/source code translates to a community call for the openness and transparency of the aforementioned sou rce code. Open source as a movement traces back to academic i nstitutions and research, because computers were originally too large and exp ensive for individual consumers to demand. Computers filled rooms and users shared com mon computing resources through different terminals. Personal computers (PCs) were available, but not very strong or useful to most of the population. Many hackers of t his era were computer science majors or graduates, playing around with potential applica tions of computers and often writing their own programs or code. Programs were originall y more tied to the hardware and different across machines, meaning software became a newly distinct term for uniform code across hardware systems. When portability beca me important, so did the commercial potential of software which influenced t he switch to protecting and appropriating intellectual property. This developme nt spelt the end of the sharing culture of the academic system, and hence the beginning of the open source movement through its forebear, the free software movement. Under copyright protection, commercial software mar kets matured at the expense of the free trade of code. Prior to the 1980s and s pecific copyright protection for code, students and faculty at universities were the prima ry users of software and consequently shared code in approximately the same fashion as re search: if one university wrote a program or improved some code, the university freel y shared that with others. Also, necessary” to operate the program. A cracker, by co mparison, is “[o]ne who breaks security on a system. Coined circa 1985 by hackers in defense aga inst journalistic misuse of hacker” in the sense given above. By definition a cracker performs illegal actions, but in actuality hackers may also break the law in playing with commercial and p roprietary intellectual property (


30 hardware manufacturers created programs capable of being shared without many legal restrictions, but this changed after the 1980 amend ments to the Copyright Act of 1976. After commercialization occurred, IP owners forced universities to license code under terms that prevented users from giving code freely amongst themselves. This caused a backlash of discontent by some academics and hacker s, most notably Richard Stallman at MIT, father of the Free Software Foundation which p receded the open source changeover (thereby making Stallman the staunch grandfather, o f sorts, of the current movement). Stallman longed to return the spirit of reciprocati on and easy sharing that predominated research institutions. Commercial efforts began draining the talent from t he AI lab that Stallman worked at, as programmers heeded the siren’s call o f better wages than academia offered. To this day, living wages present a danger to Stall man, who works with an older personal computer and invested a $230,000 MacArthur Foundati on fellowship he won to continue covering his ascetic lifestyle while he works on fr ee software (Moody 2001, 27). In an interview with Rebel Code author Glyn Moody, Stallman tellingly revealed the deep roots of his disgust of appropriation of intellectu al property: “I live in a city where you don’t need to have a ca r. I don’t want to own a house, I don’t want to spend a lot of money. If you spend a lot of money then you’re the slave of having to make money The money then jerks you around, controls your life... [Stallman h as never married or had children.] That takes a lot of money. There’s only one way I could have made that money, and that is by doing what I’m asha med of. If I had been


31 developing proprietary software, I would have been spending my life building walls to imprison people.” (Moody 2001, 28 -9) Abandoned and probably a little embittered at the e nd of what has been termed the “Golden Age of hacking,” the lab became a “ghost to wn” that lost the spirit of “playful cleverness” that epitomized the hacker mentality in the university system (Moody 2001, 14 and 18). At first Stallman’s dedication to creating free ver sions of commercial efforts was an attempt to punish commercial firms, but it evolv ed into a moral crusade to re-institute the community he had lost at the lab by trying for a free operating system that anyone could use (Moody 2001, 19). If proprietary code ere cted walls that kept the community from playing and building, free code leveled the fi eld and created vast stretches of pastures or commons that enabled cooperation and co mmunity. To accomplish this, Stallman drafted a new, very prevalent license to f acilitate this copyleft community’s rules for the commons called the GNU3 General Public License, known by its acronym GNU GPL, or GPL for short (Heffman 1997, 1508). The GNU GPL acts as the default license describing copyleft community structures, m aking it an ideal case study for this thesis. Although the story of the rise of the GNU G PL is decidedly anti-commercial and the license forms the cornerstone of the current op en source movement, the efficiencies gained by open code are not anti-commercial. For al l of the good Stallman wrought spearheading and defining the hacker community, his abrasive pursuit of radical idealism for “free software” ultimately made him a pariah to the open source movement (Moody 3 Initiated by Richard Stallman, the GNU Project aim ed to create an operating system composed entirely of free software that could handle everyth ing the proprietary Unix OS did while being better for the nature of being open and free. As me ntioned in the introduction, GNU selfreferentially stands for “GNU’s Not Unix.”


32 2001). When debates rage about the ultimate incompa tibility of open source with competitive firms, Stallman’s spirit is probably ho vering nearby. There have always been a multitude of options for l icensing, as each product can have its own license, but the GNU GPL was the first widespread one that could apply to a range of products. Before the GPL, Stallman experim ented with other licenses, but those were unsatisfactory. Drafting the GNU General Publi c License with a lawyer to make the document as legally sound as possible, Stallman ens hrined values of freedom that had always existed amongst hackers but more as unspoken custom than doctrine. Expressed in concentrated form in the GPL, however, Stallman had constructed “a kind of written constitution” enabling rapid adoption by hackers an d factions and making him forefather to many of the ideals of open source (Moody 2001, 2 7-8). As Linus Torvald’s success story for how open source appeals to programmers on an individual level informs the second chapter, Stallman’s story lays the foundatio ns for the logic behind open source licensing and legal logistics on a community level. According to Stallman, the centrality of free use (unobstructed and costless) is more imp ortant than any subsequent efficiency arguments: code should be free as a principle rathe r than a conclusion from economic reasoning. His hard-lined, unyielding commitment to values over finesse generated a GNU GPL that is fairly restrictive compared to othe r open source licenses. Eventually, these bottom-line values created a split between St allman’s Free Software movement and the emerging open source community. The term “open source” debuted at the 1998 Freeware Summit as an alternative to the free software espoused by Stallman and the FSF. Christopher Kelty documents this split from an anthropological point of view in Two Bits: The Cultural Significance of


33 Free Software (2008), dividing the terms as cultural and economi cal. The head of the open source agenda was Eric Raymond, “a libertarian pro-business hacker” who wrote about the “economic value and cost savings that Ope n Source Software represented” (Kelty 2008, 99). Kelty convincingly argues that th e split is largely irrelevant because the two groups have approximately the same practices an d behaviors and only differ in the stories and mythologies they develop. Although Kelt y clearly sympathizes with the former term, this paper shall continue with the mod ern spirit of optimism for open source coexisting with commercial, proprietary code. Kelty proposes that open source proponents see themselves as concerned with pragmat ism and seeking efficiency gains and Free Software as a dogmatic philosophy, pursuin g freedom of code above all economic concerns. Glyn Moody, who wrote the other popular history Rebel Code also talks of open source’s Linus Torvalds as pragmatic and would agree that Stallman is dogmatic, and other authors agree that the splinter ing occurred as practical and ethical viewpoints clashed, so I accept this reason to use open source as my term of choice (Moody 2001; Ulhoi 2004, 1100). Chapter 3 models th is coexistence more extensively through the various firm strategies to contribute t o open source software; however, it is worth noting at this point that open source is a mo re flexible concept than “free” software, so there is a marked evolution around the 1990s when the term open source was adopted as a better signifier of efficiency gain th an free software. The GNU GPL in Detail After the abstract descriptions of copyleft presen t in this paper thus far, it becomes worthwhile to note specific licenses and re strictions in copyleft communities. By 2005, 65-70% of all open source programs were li censed with the GPL, so it


34 represents the best license to consider in detail ( Kumar 2006, 3). Other big names are the Mozilla license and the BSD from Berkley. Succinctl y phrased, the GPL grants “everyone the right to use, modify, and distribute the program on the condition that the licensee also grants similar rights over the modifi cations he has made ” (Mustonen 2003, 101). In more detail, users accepting the license a gree to the following: 1) not to establish proprietary rights in the softw are; 2) to provide the source code to anyone to whom they give the object code; 3) to include in the software notice o f the applicability of the GNU GPL; and 4) to accept the software without warranties of any kind. (Heffman 1997, 1508 ) Providing the human-language source code ensures th at others with sufficient skill can learn from or modify the program easily. The lack o f warranty absolves copyright owners in exchange for licensing to anyone free of charge, which erases a liability for programmers and thereby encourages more programmers to release code under the GPL (Heffman 1997, 1509). This risk-shifting from guara ntees of quality to users is fundamental to sharing dynamics because limiting li abilities for the programmer thereby lowers barriers to entry by encouraging anyone to c ontribute code, even smaller and less wealthy individuals who cannot weather a lawsuit (G omulkiewicz 1999, 191). These features promote knowledge and learning in ways tha t the Constitution and the institution of copyright intended (Loren 2007, 279). Scholarly analysis indicates that the GPL “does not restrict use of the software in a manner beyond what is permitted under the Copyright Act” and “any claim under the GPL should not be pre empted by federal law,” but confirmation of these claims is too extensive for t his paper (Heffman 1997, 1511). In


35 light of the previous discussion of licenses as con ditional permits, the GPL appears legitimate. Stallman and the GPL focus almost entirely on code and software. Because copyright is broad and applies to more than softwar e, the unique paradigm of sharing can be replicated beyond the scope of programs and code to other unique expressions like writing or photographs (Heffman 1997, 1491). Stanfo rd professor Lawrence Lessig served as the focus of a similar effort “to help ar tists and authors” achieve the freedom to collaborate or incorporate one another's works that the copyleft communities created, more broadly known as the Creative Commons (Loren 2 007, 285). There are six different Creative Commons licenses with different restrictio ns to fit the variety of copyright owner's preferences regarding derivative works, whe ther uses can be commercial, or whether copies must be shared without modification (290). The close link between the Creative Commons and the GPL endures, for the Creat ive Commons keeps pre-existing legal code from the GPL for software copyright owne rs who wish to license their works with that (Carroll 2006, 48). As the GPL requires software to bear a copy of th e license or otherwise indicate the applicability of the sharing license, the six d ifferent Creative Commons licenses all necessitate that works under the license identify t hemselves with a notice and symbols and a link to the corresponding license online. “Wh en a work is marked with a notice that it is licensed under a Creative Commons license, th e public is informed that instead of the default rules of copyright law, some uses that copy right law would prohibit are instead permitted” (Loren 2007, 295). Again, works under th ese licenses signal greater accessibility and frictionless legality of use than standard copyrighted works, to the


36 extent that restrictions on such works can become c ommon knowledge like legitimate uses of parks and other public goods. Although ther e are six different Creative Commons licenses, the notices include distinctive icons des cribing at a glance which restrictions apply to a work. For instance, licenses that do not permit commercial uses of their work without express permission by the copyright owner s port a dollar sign inside a circle with a slash, similar to no smoking signs for cigarettes Signaling prompts collaborative uses that would go unexplored or underutilized if collab orators had to bargain with copyright owners to obtain a license. Beyond the fact that the Creative Commons license c an apply to many different types of copyrightable mediums, I also want to unde rscore the (perhaps inevitable) progression from completely nonproprietary to a mor e lenient use of copyrighted materials based on individual licensors’ preference s. The hardline idealist core of the GPL is not necessary and does not define open sourc e, although it is vital to tracing the evolution of open source economics. The BSD license mentioned early is another example of a license that allows copyright owners t o choose whether future users can use the works in a proprietary effort. Linus Torvalds, father of the Linux operating system and a celebrity of open source, expressed a sound v iew in his rule that “he who writes the code gets to choose his license, and nobody else ge ts to complain” (Moody 2001, 266). Granting rights to decide how an intellectual prope rty is used incentivizes its creation as copyright schemes always have, and the range of opt ions between the GPL’s version of copyleft and traditional copyright allows flexibili ty for creators to optimize their comfort in deciding which rights to retain and relinquish.


37 Licensing Applications Neither community of copyright owners--commercial n or copyleft--transfers title of the software to users. Doing so would effectivel y transform users into owners capable of making any choice about what to do with the inte llectual property. Commercial firms would be unable to control how programs are distrib uted and hence lose their method of appropriating returns from the intellectual propert y. Copyleft communities, meanwhile, are more concerned with others appropriating their work and profiting by selling the same code to someone who does not know or care that they can get it for free. Instead of transferring title, copyright owners license their software, basically the equivalent of renting the software rather than selling it. Licensing software gives copyright owners control o ver who uses software and how. Because copyright automatically assumes the gr eatest level of restriction, licensing allows owners to selectively ungrip the stranglehol d over their properties and thereby allow others to use their programs. In commercial s oftware, this loosening often happens when a customer pays to use the software and agrees to the terms of use offered by the copyright owner, while open source and copyleft com munities only require the terms of service be met. Just as renters can be evicted from properties, copyright owners generally retain the power to rescind their licenses and make it illegal for users to continue operating the software. Contractual issues may mean that commercial entities might have to refund consumers, so it is not quite the ideolog ical threat of losing out on something you had yesterday without compensation. In open sou rce contexts of free, the threat is more theoretical than real thus far. However differ ent the goals of copyright and copyleft


38 software, they work on similar legal grounds and re ly heavily on terms of service to control code. Copyright’s default terms are more restrictive tha n copyleft licenses like the GPL or Creative Commons, which relax some conditions of infringing behavior, so would-be users cannot simultaneously assert that they have p ermission to use that intellectual property without accepting the copyleft terms of pe rmission as valid (Kumar 2006, 13). The US Court of Appeals, 7th District, ruled in ProCD Incorporated v. Zeidenberg (86 F.3d 1447 (1996)) that a licensor can structure acceptance of the ter ms of service by the users’ conduct. Using the copyrighted work in a man ner that the licensor proposes to treat as acceptance counts as acceptance. In the Zeidenberg case, the buyer accepted terms of service simply “by using the software after having an opportunity to read the license at leisure. This Zeidenberg did. He had no choice, bec ause the software splashed the license on the screen and would not let him proceed without indicating acceptance.” The same acceptance of terms should be applicable to open so urce software, which regularly splash their respective licenses on the screen before inst alling or permitting use of a product. However, some open source programs are less technic ally sophisticated or invasive, and they do not have splash screens that force users to accept the terms of service before starting the program. Rather, some p rograms merely include the license in the program files and point users there as necessar y. However, the Zeidenberg case states that the terms need not be visible at the time of a cquiring the software, partially because it would not be conducive to advertising. Quoting t he UCC (Universal Commercial Code), the Court of Appeals specifically addresses the difference between acceptance of an offer--terms of service--from acceptance of a pr oduct (86 F.3d 1447 (1996)). In other


39 words, downloading a program is one act, and using the program is another. As long as there is the opportunity for the user to read the license before using and can return the software if the terms are unacceptable, that should be enough to please the 7th District Court of Appeals. Therefore, as long as the license is included somewhere with the software, like package inserts that are inside the box with commercial drugs, it is the user’s responsibility to read the information there in. Why Patents Still Matter—Future Restrictiveness More recently than the 1980s, however, the United States granted patents to programs, and the European Union has an established history of awarding patents to programs (Mustonen 2003, 103)., Inc.’s c ontroversial patent on one-click buying for virtual shopping carts, US Patent No. 5, 960,411, is a perfect example of a US patent for code (Hartman et al. 1999) The Free Software Foundation criticized the patent for being obvious and the company, itself, for brin ging a lawsuit against competitor Barnes & Noble (Free Softeware Foundation 2011). Im plementation is very easy to work around and was replicated on many other sites that circumvented the patent by requiring two clicks instead of one (O’Reilly 2000). Indeed, criticisms regarding obviousness appear valid, because the European Patent Office ne ver granted protection to Amazon’s claim (European Patent Organization 2007). As karma would have it, patent wars have escalated and Amazon now faces a lawsuit against its new Fire tablet from Acacia Res earch Corp—a publicly traded firm that pools patents and is sometimes regarded as a p atent troll—for infringing some of its frivolous patents (Roberts 2011). Regarding the pat ent in question, U.S. Patent No.


40 6,956,562 seems to describe a basic function of all touchscreen devices, including smart phones: According to the method, a graphical feature having a surface area is displayed on a touch-sensitive screen... To control software executing on the processor, a user-supplied writing on the surface area is received and the software is controlled responsive to the writing. (O’Hara et al. 2005) What may have started as defensive posturing by com panies collecting broad and fairly obvious patents now threatens entire industries. Ev en commercial firms may have limited prospects of bringing new products to market, so th e extortionary pressures on nonmonetary open source development is predictably eve n more significant if threatened by legal action against them. Patenting matters to open source because patents c an restrict the very nature of what software can legally do. Most importantly, pat ents apply to functions more than expressions, so while code can achieve the same out come by different commands, it is exponentially more restrictive and costly to deny m ethods of achieving things. As a case study, Linux would probably never be what it is tod ay if functions of operating systems were patentable back in the 1980s. Stallman’s dream of free software led him to attempt a free operating system that could run all of the pro grams that people had already written for the proprietary UNIX operating system developed by AT&T employees at Bell labs. A UNIX-like system required him to reproduce work o n many small pieces with functions similar to those in UNIX without relying on the source code of UNIX to code the replacement pieces. The source code was a trade secret that if convicted of stealing


41 might have doomed the project (Moody 2001, 20-1). T he result of all of this hard work was the GNU/Linux operating system (often acknowled ged only as Linux, even within this paper to conform to normal use). Yet, if paten ts on the UNIX processes existed rather than copyright, it likely would have precluded any efforts to rewrite the operating system because anything close enough to approximate a UNIX environment would fall on the wrong side of the patent. While communities can generate copyright for nothi ng, patents require formal grants from the government that require fees to app ly for protection. Given the noncommercial nature of open source collaboration, it is unlikely that anyone will apply for patents in order to foster the commons. Rather, the only legal protection that open source can afford on a shoestring budget would be r eliance on prior art, meaning that open source can cause patents filed by proprietary seekers to be rejected if they can prove that the supposed innovation has already been done or is obvious to anyone in the field. Unfortunately, this is not likely to be very helpfu l, for one of the greatest criticisms leveled against open source is that many of its big gest successes in recent years were derivative in nature of proprietary efforts. Modern successes like Linux or GIMP, derivative of UNIX and Photoshop respectively, are from the 1990’s rather than the original internet-based successes like Sendmail. Fr ee and open source software may be doomed to lag decades behind the fast-moving techno logy industries if software patents are adopted in force. Cheating the System While the license organizes the rights of original copyright owners and userinnovators respectively, it does not of itself ensu re that the system functions. There will


42 always be cheats, and some entity must punish defec tors so that cheating does not become widespread. While the courts are the obvious choice for punishing defectors, there are a number of issues that may limit their e ffectiveness as arbitrators of justice: copyleft licenses do not have good precedence in co urt, and there are problems concerning the valuation of damages for freely dist ributed programs. Each shall be addressed in turn, but sometimes the court does not measure up to its charges as well as it should. Legal sanctions are not the only recourse to punis h defectors. Informal governance structures within the open source commun ity, itself, can use stigma to enforce compliance, ostracizing defectors and destr oying their reputations in ways strikingly similar to academia. Because this strain of analysis is better explained in the context of how open source communities govern thems elves, I leave that for the next chapter. The third substantive chapter also discuss es how commercial firms’ relationships with the open source community can help or hurt the ir bottom line, where forging a good relationship can be monetarily beneficial and not p laying by the rules can spark animosity with IBM and Microsoft as relevant examples. My foc us below is exclusively on the legal responses for defection, leaving informal gov ernance for the later chapters. Formal governance by scrutiny in courts The GNU GPL is copyrighted 1989, but as of 1997 it had not gone to court (Heffman 1997, 1515). According to Richard Stallman at that time, the Free Software Foundations sent out several letters a year demandi ng compliance with the GPL, but never once have the organizations promoting the GNU GPL been forced to litigate against infringers (1509). Even now, in the second decade of the twenty-first century, the


43 legal waters are murky, for most suits relating to the enforceability of the GPL end in settlement and prevent precedence from forming (Kum ar 2006, 5). Similarly, the first Creative Commons licenses were available in Decembe r 2002 (Loren, 2007, 286). As of 2007, no United States court addressed the Creative Commons license in particular, but the Netherland gave an affirmative ruling and Spain appeared to accept the license in one of its cases (277). The rapid adoption of copyleft and Creative Commons licenses make it somewhat odd, then, that they have not seen a real day in co urt. Apache web servers and the Linux operating system are both copyleft software license d with the GNU GPL that have “gained significant market shares in a short time” (Mustonen 2003, 102). In 2006, there were 16 million digital works--including photograph s and music, and up to 149 million unique webpages in 2007--linking back to the online Creative Commons licenses (Loren 2007, 286-7; Carroll 2006, 47). Such rapid adoption points toward future litigation regarding infringement, according to several source s. Settlements and lack of precedence with copyright infringement of works under sharing licenses like the Creative Commons or GPL prevent complete knowledge of how courts wou ld handle consequences of infringement; there are certainly issues in determi ning damages when a licensee frustrates a licensor’s expectation of zero profits, and speci fic performance or injunction may or may not be appropriate remedies (Kumar 2006, 11). F urthermore, the development of software, in particular, takes place across the Int ernet, where hundreds of contributors can work independently on the same project, making asse ssment of the value of contributions difficult or impossible (15).


44 When infringement occurs, lawyers will undoubtedly pursue multiple interpretations of the licensing terms to achieve t he greatest effect and protection for the copyrighted materials. For instance, one avenue tha t lawyers explore in the literature would make copyleft licenses contractual to attain maximum protection and legal remedies to copyright infringement. Even without a contract (or the possibility of one), liability for copyright infringement remains if use rs do not obey the terms of service. Many in the open source community, and particularly the Free Software Foundation started by copyleft founder Richard Stallman, inter pret the GPL as a non-contractual license (Kumar 2006, 35). So, why do lawyers and at torneys pursue this line of legal reasoning? “Hammering the GPL into a contract-shape d mold for legal stability is very tempting. Contract law is more developed than licen sing law and offers a wide range of legal remedies that are not available under the Cop yright Act.” (Kumar 2006, 24). As one example of a new legal remedy, specific performance could require infringers to follow through with the stipulations of the license, namel y that they make their additions to the code base open source as well. This is the extreme interpretation of the “viral” nature of restrictive copyleft licenses like the GPL. In shor t, the distinction between contractual and non-contractual licenses has important ramifica tions for legal governance that can shape open source’s legal future. Therefore, the le gal system should make a determination of whether the sharing licenses of th e GPL and Creative Commons are actually contracts between the copyright owners and the user-innovators, and then trace the effects of that decision on the legal landscape


45 Are Sharing Licenses Contractual? Cited earlier in the chapter, cases such as ProCD Incorporated v. Zeidenberg have created the traditional interpretation of software licenses as contracts, making the idea of copyleft licenses as contractual at least plausible (Kumar 2006, 6). Although the 7th District Court of Appeals acknowledges the legal di fferences between contract and license law, it saves the subject for another case where the distinctions apply more and proceeds with its analysis (86 F.3d 1447 (1996)). A lthough the case cites the Universal Commercial Code and applies to a commercial intelle ctual property while copyleft programs are not commercial, I believe that the gen eral principles of the case would stand in other contract formation cases. The judge specif ically rejects monetary considerations as the sole way to create a contract when saying th at the contract is formed after the transaction at the store occurs; rather, the contra ct becomes binding when the licensee uses the software, agreeing to terms of use at a sp lash screen that ensures the users acceptance of terms (257). Therefore, copyleft lice nse can be a contract by specifying that using the software or source code forms a cont ractual relationship between the IP owner and the user. Of course, interpreting the GPL as contractual is t ricky when those who created the prominent license specifically reject the contr actual interpretation. Like any contract, open source needs to prove the requirements for the formation of a contract before that purported contract is valid and enforceable. In thr ee parts, there must be the licensor’s offer, the licensee’s acceptance, and consideration for both parties. The act of applying the GPL or Creative Commons license to a copyrighta ble work represents the offer, and the terms in that license propose that exercising r ights as defined by the license


46 demonstrate acceptance (Loren 2007, 311). Copyright assumes that all rights are reserved, meaning that the default copyright terms are still more restrictive than even the most restrictive copyleft licenses, so infringers c annot assert that they have permission to use that copyrighted work without accepting the ter ms of the license as valid (Kumar 2006, 13). Clearly the release of copyrighted mater ials constitutes a consideration on the part of the copyright owner. However, what the user or licensee offers in return is more tricky. One might interpret the intentions of the GPL and C reative Commons to grow the sharing community or the commons, from which all (i ncluding the owners of any individual intellectual property) may benefit. In t his case, an expectation of reciprocation may be what the intellectual property owner gains f rom his licensees. This open source notion of reciprocity can translate to future expec tations, constituting consideration, but such an argument is unclear, vague, and ill-defined If the copyright owner licenses the work, conditioned on the licensee’s actions, and th e licensee does not do what he or she agreed to by accepting the licenses, a breach of co ntract seems legitimate (Loren 2007, 305). Requirements for proper attribution of origin al copyright owners in derivative works represent positive, as opposed to negative, o bligations on the actions undertaken by licensees, conceivably contractual consideration s. As previously mentioned, however, Stallman and much of the open source community do not support the idea of the GPL as a c ontract; this means the most prominent open source license is a document that fa ils to detail what benefits the licensor receives in exchange for its grant (Kumar 2006, 24) .4 Kumar notes that the GPL—and 4 Because the GPL fails to list consideration, it is also difficult to prove “a meeting of the minds” where the parties understand their roles in the con tract (19-23).


47 hence the Creative Commons inspired by the GPL—base s itself on real property licenses and gives the following example to disprove any con sideration: Suppose that a land owner grants a revocable licens e to the public to cross through a strip of the landowner’s propert y to access a public beach. The landowner does not explicitly receive an ything in return from the public. Through the landowner may limit the pub lic’s access to certain times of the day, these “burdens” on the public do not serve as consideration for using the landowner’s property. T hey are merely limitations on the access that the public is receiv ing. (2006, 20-1) Along that line of reasoning, the sharing licenses merely restrict what purposes intellectual property owners will allow--e.g., deri vative works and reproduction must be non-commercial only, or restrictions not to provide a warranty in case the software is defective or harmful to the user’s computer system-and so they qualify as limitations, not consideration. Any promises to abide by restric tions in sharing licenses lack value “because they are merely promises to not engage in actions that are otherwise prohibited by law” (Loren 2007, 312). Only if user-modificatio ns are released to the commons in the form of derivative works or bug fixes does any tang ible return benefit the copyright owner as consideration. As it may be supposed, this cannot always be expected to happen and there are many free riders who lack the motivat ion or technical skills to offer such consideration (Kumar 2006, 21). What a Non-contractual License Means to the Commons Remedies for copyright infringement available under the Copyright Act include damages, injunctions, and sometimes attorney’s fees (Kumar 2006, 15). Damages are


48 difficult to quantify for free products like copyle ft computer software, but courts may disgorge profits from the illegitimate sale of copy left software. Copyright infringement cases are handled exclusively by the federal courts while states typically handle breach of contract claims, so immediately jurisdiction is decided by rejecting the notion of a contractual GPL, presumably applying to the Creativ e Commons license as well (Loren 2007, 302-3). Federal courts tend to require more s pecialized attorneys, handle higher value cases, and command higher legal fees, making it potentially less costly to treating the GPL and other sharing licenses as contractual a nd stay out of federal courts (Person). Perhaps more importantly to the nature of the commo ns, without a contract and consideration the licensor can freely revoke permis sion to use the work, just as the boardwalk owner can increase the limitations by fur ther restricting the hours of day he allows neighbors to use his boardwalk. Nothing forc es the copyright owner to make the work accessible perpetually, as it could if the use r or licensee could show consideration through something like a sale (Kumar 2006, 13-4). A t most, the GPL and Creative Commons licenses are conditional gifts (Loren 2007, 312). A large degree of the strength of the commons comes from an understanding of stabi lity and permanence in the ability to use materials in the commons whenever and howeve r much the user desires. Copyright owners can undermine the stability of the commons b y revoking the permission granted in the licenses, per their legal rights under a non -contractual sharing license. Whatever current law permits, copyright owners shou ld not take back the rights offered to users and policy can prohibit such actio ns if it wants to foster an effective commons. If someone decides to contribute code to a program, it is in their best interest to do so only if they can be sure that the base cod e will not be snatched away from them


49 in the future. The second draft of the third versio n of the GPL adds language that issuing terms under the GPL becomes irrevocable, meaning on e cannot reclaim rights rescinded by adoption of the GPL, but the clause’s legal vali dity is debatable (Kumar 2006, 14). Creative Commons and GPL scholars Lydia Loren and S apna Kumar advocate alternative change in copyright law to give legitim acy to the new structure of copyleft rights. Restructuring Copyright to Account for Sharing Lice nses Loren (2007) wants to institute a new category of “ limited abandonment” to mark the permanence of some-but-not-all-rights-reserved, which would apply to the Creative Commons and other sharing licenses. Limited abandon ment would expressly address the revocability issues undermining the commons above b ut would not apply to individual licensing agreements, as between normal commercial firms (323). A category of limited abandonment would help frame the issues in a produc tive way by eliminating termination rights, which were meant to protect copyright owner s who may have been in a poor bargaining position at the initial transfer of righ ts. Copyright owners face no bargaining pressure when releasing under a sharing license, so the call for protection is unnecessary and potentially harmful to the stability of the com mons (325). By eliminating the possibility of taking the intellectual property awa y from the commons after it is there, the logic is that people can rely on the commons always remaining the commons for new users in the future (as opposed to only original us ers). Kumar, meanwhile, wants to see a focused amendment to the Copyright Act recognize open source licenses and define causes of action and remedies. This would clarify the risks for parties about to use the comm ons, thereby eliminating the uncertainty


50 without affecting other legal arrangements and prop rietary approaches, but it is also a much more intensive legislative undertaking. Preced ence in legal interpretations of copyleft licenses is important, but variation from one jurisdiction to another raises transaction costs of adopting open source programs because the rules can change between different parts of the country. A statute amendment would reduce courts’ leeway to interpret broad arguments and ensure a more uniform treatment across the nation (Kumar 2006, 27-8). Of course an amendment of this mangitude would requ ire extensive Congressional support, but the political feasibilit y is less important to my analysis than the benefits from instituting some type of national recognition of open source dynamics in copyright law. The following chapters demonstrat e some of the efficiency and promise of open source software development, but all would be for not without grounding the movement in the existing legal fabric of intellectu al property rights management. Copyleft licenses already prove to be fairly robust without official government acknowledgement, but the open source and sharing mo vements deserve some consideration as new manifestations of intellectual property rights that will greatly affect the legal system in years to come.


51 Chapter 2 Sharing licenses, discussed extensively in the pre vious chapter, lay out the group dynamics of the open source development paradigm in a macroscopic sense, observing how the constituents and legal rules organize the c ommunity of contributors, known as hackers. The series of solutions to intellectual pr operty issues exemplified by Stallman’s General Public License outline the foundation for c ommunity involvement. Yet, such analysis on its own is insufficient for explaining the continuance, except on altruistic grounds which are limited and unstable. As Osterloh and Rota wryly observe, “Even among intrinsically motivated donators, martyrs and saints are in short supply” (2007, 166), to which I add that Richard Stallman is an ex ception who proves the rule. For an overall perspective of how the paradigm functions, economists must strive to understand it at both the macro and micro levels. Therefore, t his chapter identifies individuals’ motivations for contributing to the public good of open source code, which are often separate from the structure of sharing made by lice nses. Intangible Assets--Organizing and Generating Surplu s without Market Exchange Because open source development diverges so radica lly from traditional coordination mechanisms used by commercial developm ent--such as central planning, systems-level design, and schedules--comparative st udies of the merits and usefulness of this novel development cycle must understand how op en source empirically operates (Mockus et al. 2002, 310). There is no single open source community, nor a single organizational method to open source projects, but examining at least one is instructive for the larger system. Whenever possible, this chap ter outlines general rules and


52 characteristics while employing historical examples to give context and proof of concept. While the first chapter periodically reached beyond software into other realms of intellectual property, this chapter sticks close to the programming movement because doing so allows more direct comparison to commercia l software firms. Both open source and commercial firms allocate time and resources in a decentralized environment, but the aim of this chapter is to discover exactly how allo cations occur without traditional markets and monetary compensation. Following the model presented in the last chapter, I start with an examination of the theoretical economic basis of understanding the nonmonetary nature of open source, namely a discussion of profits, before carrying thi s analysis through case studies of open source software. Time is money, as they say, and th e resultant output of code is worth something, even though the difficulty of ascribing a particular value is hefty. As seen in the previous chapter, ascribing damages to copyrigh t infringement of an open source product is an unruly task, because programmers are not paid for their contributions to the free software. Some companies do hire programmers a nd donate help to the open source software (OSS) movement, with cases such as IBM emp loyees developing for the Linux operating system in Chapter 3, but damages are stil l imperfect because the product draws in no revenue. The economic concept of profits will not provide a magic formula for valuing open code, but it will demonstrate a basis for assuming that benefits substantially outweigh the costs of contributing. Reframing Profits and Costs Profits are defined in traditional economics as re venues minus cost, where it is self-evident that production is the appropriate cho ice when benefits exceed costs


53 (Osterloh and Rota 2007, 166). Open source contribu tors will rationally stop producing code if they derive no benefit from their activity, so if contributors choose to remain in the open source sector. However, it is probably bec ause of a net benefit to the individuals, themselves, and not ideological or altruistic motiv ations. Of course, the difficulty applying profits to open source is that OSS code is freely available for download which means that revenue is zero. Because costs of produc ing code is decidedly not zero, another interpretation of profit may be more useful to OSS production decisions. A more Marxian conception of profit is “the form of the surplus product... it is what is left over... after wages, the costs of mate rials used up, and wear and tear on machines have been paid” (Bowles et al. 2005, 144). Surplus product is a good term for the portion of output that is in excess of what is strictly required to ensure production, in this case the production and distribution of code ( 93). Because the easy and virtually costless replication of code minimizes much of the associated cost of supplying and distributing programs, the benefits to society grow with additional users even if those benefits are not estimable based on how much each u ser is willing to pay. Open source differs substantially from market systems precisely because its surplus product is not related to the monetary revenue resulting from sale s, and all users of the software are beneficiaries of the surplus product, not only deve lopers. Because users do not pay, however, the demand side of the equation is almost irrelevant to equilibrium production and the allocation of resources, and it is certainl y secondary to supply dynamics except in gift economy or reputation literature and the cases of pro-social or community motivations for contribution. Opportunity Costs Approximate Costs of Production


54 Potentially limitless benefits conferred by public use of code does not, of course, mean that it is always in the best interest to trea t code freely and openly, or the intellectual property rights management of patents and copyrights would lack a raison d’tre as an incentive to produce. The first chapte r treated this as the promise of open source development, as a reason to consider a new f ormulation of intellectual property rights, but here I want to focus on how it makes se nse at all. If it is impossible to systematically value the benefits, what matters to the decision to produce OSS or not is whether benefits at least meet costs of code produc tion. Costs can be either explicit, easily valued by bills listing the necessary paymen ts for maintaining a web server, etc., or they may be implicit, inherent in the activity but without incurring payment. A classic example of implicit costs is the leisure/work dicht omy stressed in introductory economics texts: leisure is defined as allocating one’s time to some other purpose than earning money--relaxing, having fun, or cleaning the house-which has an implicit cost of not making money from work (Krugman and Wells 2005, 298 -9). This represents an opportunity cost because using time in one way forc es one to forgo employing it elsewhere (Varian 2006, 335). Implicit costs are a large component of costs that prove central to understanding the logic of open source e conomics. Because open source does not generate revenue, all contributions are, essentially, donated. Donors are more willing to contribute if t he private costs of contributing are not high, which helped collaboration accelerate exponen tially in the age of cheap, global communication (Osterloh and Rota 2007, 166). Commun ication and coordination of open source projects is regularly facilitated through th e Internet, in message boards and email lists primarily. Mailing lists are very open, and a nyone may sign up without any explicit


55 costs like membership dues. Participants do not nee d to meet one another in person, and it is not uncommon for many ancillary members of a project to listen in on conversations out of interest rather than involvement (Mockus et al. 2002, 310; Fitzgerald 2006, 589). Mangaement tools and hosting servers and code depos itories are also free, thanks to sites like that host tens of thousands of f/OSS (free and/or open source software) projects without charge ( Geeknet ). The primary cost in OSS development is time. Since the beginning of scholarship on open source, commentators and scholars have sugg ested that “the only really limiting resource is skilled attention” (Raymond 1999, 43). That is, the lack of interest and willingness starves OSS development into stasis, bu t contributors typically have the requisite machines, office space, and Internet acce ss at home to keep a favorite project alive. Whether contributors participate “on the clo ck” at work or on their own time does not matter. For volunteers, working for pay or usin g leisure in other ways make obvious substitutes, and thus it becomes more important for this type of contributor to have fun, to enjoy the activity of coding on its own. Enjoyment is less necessary for those who work on open source at work, but the opportunity cost ca n still be measured in time because they could be pursuing other work instead (Lerner a nd Tirole 2005b, 102-3). For those who are fortunate enough to work for a company that donates its time and effort on open source, workers maintain a balance between open sou rce and other work important to their firm or university employers. Considering rationally whether to contribute to ope n source, economists are trained to consider the opportunity costs, both exp licit and implicit, directly into their calculations. Accountants adopt a more simplistic v iew of profits, taking the difference


56 between revenues/benefits and explicit costs like s erver maintenance, but economic profit acknowledges implicit costs like donated time (McCo nnell et al. 2012, 142). Thus, in economics, rational agents are guided to act accord ing to what maximizes their utility in terms of all possible choices. Where open source wa s a mystery to the business world and media, economists have a more intuitive sense of th e open source economy and are in the best place to translate its logic to others. Working for pay obviously gives a conceptually simp le valuation of time spent on open source. To calculate the donated time from wor kers merely requires summing all of the individual wages by the number of hours worked at each wage, per the original formula of costs above. In terms of the volunteers, the cost of the opportunity lost includes the option to be paid, in which case their volunteered time is as valuable as if they were being paid. To understand something of th e scale of open source contributions, it helps to look at hourly wage data from the Unite d States Bureau of Labor Statistics. Computer programmers earned a mean hourly pay of co mputer programmers of $37.13 in May 2010 ( Bureau of Labor Statistics ). In order to better understand the composition of co ntributors, both paid and unpaid, Lakhani and Wolf (2003) conducted a pair of separate but identical surveys by emailing individuals listed on as o fficial developers on open source projects. I shall use their data extensively throug hout the chapter to quantify claims and abstract concepts of open source development. A lit tle over a third of those contacted filled out the survey, 34.3% (8). The average numbe r of hours spend per week on all


57 F/OSS projects was 14.3 (sd=15.7)5, with paid contributors giving a average of 17.7 hours/week and unpaid only 11.7 hours. Time spent o n the focal project, for which the survey participant is listed as official developer, was less, at 7.5 (11.6) for all, 10.3 for paid, and 5.7 for volunteers (Table 3, 20). Because specifics are less important to the thesis than the general outline, I shall not spend much time estimating how large the open source sector is, but a year (52 weeks) times the n umber of hours per week (14.3) times the mean hourly pay ($37.13) times the number of co ntributors (say, 1000 in a given week) conservatively enters the tens of millions of dollars (eight digits) on an annual basis.6 On the other hand, it is difficult to determine ho w volunteered time will affect labor markets: a large volunteer presence might dra w down wages as labor supply increases. Therefore, this is a mere estimate but t he literature I found does not tackle issues relating to the size of the OSS sector. Costs to Programmers vs. Costs to Firms Unlike a normal, commercial environment where firms pay all costs out of revenue or credit, volunteerism necessitates that m ore costs in the open source model are borne by individual contributors. This fact is actu ally quite illuminating because it explains the fluidity of the OSS development scene: participants who do not make their living from OSS respond to circumstances when deter mining the level of their 5 These standard deviations are quite large. The flo w of contributions is extremely skewed, with more than a third reporting no time spent on projec ts that week, strengthening my case that contributors transition regularly and with ease. 6 1000 contributors per week is an offhand estimate Bonaccorsi and Rossi estimates a community of 120,000 developers from 12,000 project s averaging 20 developers, each of whom typically work on at least two projects (2003, 1246 ). Those assumptions are much too large for my purposes, because there are plenty of small proj ects with only one developer that do not work every week on his/her project, underscoring how bac k-of-the-hand 120,000 seems.


58 involvement. Volunteers participate when they can, whenever it is pleasurable or personally profitable to do so. Some open source de velopers like Miguel de Icaza and Matthias Ettrich suggest that turnover may be as fr equent as six months to two years, as school, marriage, job changes, or family circumstan ces constitute important life changes that alter the cost of contributing (Moody 2001, 31 6). Paid contributors give 51% more time to a project than volunteers, but there is eno rmous range in both groups, from zero to 85 hours. Distribution was skewed towards no work o n current projects, with 25% of respondents spending no time on their focal project and 11% spending no time on any current projects (Lakhani and Wolf 2003, 10). Because of the uncoordinated nature of personal cir cumstances, there is an ebb and flow of productiveness on individual projects, but the OSS development scene as a whole enjoys relative stability in a similar way to a national economy. The unlikelihood of significant numbers of major contributors leavin g simultaneously, effectively disrupting a project’s development, diminishes the effect on projects of individuals leaving medium or small projects. Even Apache, whic h at 400 individual contributors has a much smaller contributor base than Mozilla or Lin ux, withstands transitions well because of staggered contributions. The Apache Grou p stands in the double digits (1225), but the core contributors may only be 4 to 6 o f the active Apache Group and 2 or 3 on the cusp of membership with something to prove, not even totaling to the number of members with committing power to change the source code base (Mockus et al. 2002, 317). Furthermore, because current computer science major s and volunteers continue to be a significant contingent of OSS development, bus iness cycles and economic recessions


59 may not endanger as many projects as other forms of collective action. Indeed, many contributors are heavily concentrated in the inform ation technology sector, where 58% of respondents held jobs in IT, and those contributors are most likely to be paid for their efforts and consequently give more time to OSS. How ever, 19.5% of those surveyed by Lakhani and Wolf, listed as official developers of an open source project, were students, and another 7% were academic researchers (Lakhani a nd Wolf 2003, 9). Despite their lack of experience, students have started several i mportant projects, the most famous of whom is Linus Torvalds, a young computer science ma jor who birthed the open source operating system Linux. Another graduate student at the University of Berkeley named Eric Allman created Sendmail in the 1970’s, which f reely handled 75% of all email traffic in 2000 (Lerner and Tirole 2002, 211-2). Ac ademia alone provides a quarter of the contributing base to OSS and promises some protecti on against widespread dissolution of the open source paradigm. Investments--Capital and Human Continuing our analysis of the costs, there are ve ry large fixed costs associated with producing code, which I shall refer to as inve stments. First, software requires hardware capable of operating it, which represents a capital investment on the part of all users. Previously, I mentioned that users must have room on their hard drives to store the instructions for whatever program they want to use. Unsurprisingly, this includes most, if not all, OSS contributors. When surveyed about thei r top three reasons to participate in open source projects, 58.7% cited a need for the pa rticular code/program, either for work or non-work purposes. It was the most common motiva tion, as measured by the survey, followed by 44.9% of respondents saying that the co de was intellectually stimulating to


60 write (Lakhani and Wolf 2003, Table 6, 23). Many c ontributors purchase their own computers, in addition to any allocated by work, me aning that users bear a significant portion of the capital investment of OSS developmen t.7 Also, people periodically update their PCs or purchase new rigs, but most programs s till run on older hardware, particularly open source software, meaning the capi tal investments may be inefficient in terms of cost savings. Just as explicit costs are secondary to the opport unity cost of time, capital investments in software are dwarfed by human capita l investments. Whether formal or not, schooling or training is an important prerequi site to coding successfully, just as important as composition and grammar is to fiction writing. More than half (51%) of the official developers leading an open source project on received university-level computer science schooling, althou gh not specifically for OSS-related purposes (Lakhani and Wolf 2003, 9). I expect this number to be much higher for larger projects that coordinate more contributors because of the complexity of the code base. Openness demands a program’s source code is availab le to anyone, allowing one to read, replicate, and/or alter in whatever way th e individual sees fit, but that is not the end of the story. Where sharing licenses govern the legal and legitimate redistribution of source materials, other programs provide the tools to change or copy the code or to compile it into a usable program. The sharing licen ses that grant the rights of any user to perform these actions are not predicated on the exi stence of such tools, but in many cases they do exist as open source programs, themselves ( Moody 2001, 229). Likewise, the 7 Although rare, there are some projects like Mozill a that use corporately sponsored equipment to test daily builds of the software across different platforms (Mockus et al. 2002, 332). This represents a cost not borne by individuals, and my analysis also neglects the relative costs of PCs and servers, the latter primarily purchased by firms instead of individuals.


61 capacity to claim the rights in sharing licenses th rough technical knowledge is not predetermined, either, making technical literacy es sential to contribution. Despite the capacity of any individual to meddle w ith source code, contributors are a motivated and self-selective group. As libert arian programmer Eric Raymond (1999, 30) famously wrote in his early essay on the open source paradigm, “The Cathedral and the Bazaar,” about the various steps culminating in a bug fix or patch: ...contributions are received not from a random sam ple, but from people who are interested enough to use the softwar e, learn about how it works, attempt to find solutions to problems they e ncounter, and actually produce an apparently reasonable fix. Anyone who pa sses all of these filters is highly likely to have something useful t o contribute. Raymond’s assertions about the random sample are mi sleading because he seems to be indicating a population of all software users rather than technically literate hackers, but the gist of his argument stands. There are three main groups of knowledge amongst users: the largest are individual s who only use programs that are user-friendly, another of hobbyists who have te chnical knowledge but cannot on their own sustain OSS, and a final, more elite g roup often described as a hacker culture that habitually plays around with co de and builds things (Bonaccorsi and Rossi 2003, 1244). The majority of contributors are experienced and/or professional programmers who are capable of making informed coding choices (von Hippel and von Krogh 2003, 212). Of 67 3 respondents to their survey, Lakhani and Wolf calculated a mean of 11.86 years of coding experience for official developers; standard deviation is 7.04 while the minimum and


62 maximum is 1 and 44 years, respectively (2003, Tabl e 1, 18). Enormous variation cannot disguise the broad role of education in eith er a formal or informal capacity. Human Capital Investments and Sunk Costs Source code may require some level of human capital to understand, but practical demonstrations of functioning code is a swift and d irect method of teaching (Kelty 2008, 135). Previous source code acted as a scaffolding f or the Linux project, for instance, which was new code but built on the conceptual mode ls of operating systems from the past (Raymond 1999, 25). In fact, human capital inv estment is a large cost to the programmer in a fixed and definitive sense, for it takes many years to transform babies who have no sense of language at all into serviceab le writers. The first chapter suggested that software is characterized by high fixed costs and low marginal costs, just like other classic public goods. This is true in the sense of disseminating code, wherein broadband infrastructure is monumentally more expensive than delivery through that system, and it is true in the sense of creating code. However, for day-to-day contributions to open sourc e, human capital investments count largely as a sunk cost that cannot be recover ed (McConnell et al. 2012, 158-9). There is no way to directly sell back knowledge for resources spent educating the programmer; one can become a teacher and use one’s time and efforts to utilize the knowledge and provide a service to a new generation of programmers, but as Thomas Jefferson observed, “He who receives ideas from me, receives instruction himself without lessening mine; as he who lights his taper at mine receives light without darkening me.” As such, the human capital investment is largely se condary to the decision of whether to contribute to open source than the opportunity cost of time. In open source, opportunity


63 cost of time becomes almost synonymous with the con cept of marginal cost. Programmers do command a higher wage than most blue -collar professions, averaging $37.13 in May 2011, presumably because of their hum an capital in the way of skills. This means that education inflates their marginal costs, but not to the degree of fixed costs, and it is almost always worth their while to contin ue providing code. Per Raymond’s assertions above, contributors incur large fixed co sts in learning something about how the program works before they provide any bug fixes, bu t these sunk costs are irrelevant to the decision to keep producing after the first fix. Rather, marginal costs decrease as the individual learns more about the code base and how it functions, thereby making contributions relatively inexpensive the more one i nvests in a project (Lerner and Tirole 2002, 220). The key to open source success, then, l ies in making the original sunk costs worthwhile and encouraging participation, which get s less costly as familiarity with the projects increase. To return to the theme of this chapter as the micro -level set of motivations that step beyond the promise of sharing licenses, it is appropriate to build from the historical lessons of the first chapter’s exposition of Richar d Stallman. Stallman’s Free Software Foundation formed the ideological dream of a free o perating system, but it also neglected development of a central component called the kerne l. Working largely alone, Stallman had many other pieces of the operating system to wr ite and neglected ever finishing the GNU Hurd, as the project was known. “That Stallman had left the kernel until last might seem strange given its central nature,” writes Mood y (2001, 25), partly because he was busy with other pieces of the monumental undertakin g and partly because Stallman was self-admittedly “...trying to bypass the hardest pa rt of the job.” What GNU needed, a


64 professor named Andrew Tanenbaum already had, but a ccording to Tanenbaum’s testimony Stallman was too “abrasive” and demanding to negotiate the use of Tanenbaum’s kernel (32). Torvalds eventually filled this void in the early 1990’s by rewriting and reconstituting Tanenbaum’s success, w hich finally transformed the FSF’s dream of a free operating system into something tan gible. Linux has already been attributed to the pluck show n by Finnish student Linus Torvalds, but its reality also owes a great deal to the educational efforts of those like Andrew Tanenbaum, who provided computer science stu dents with a working model of an operating system on which to base their efforts. Often described as UNIX-like, Linux owes much of its architecture to the proprietary op erating system UNIX developed by AT&T in the early 1970’s, but through the intermedi ary step of Tanenbaum’s own operating system Minix. When AT&T stopped being a r egulated monopoly forbidden from commercialize UNIX in 1984, AT&T quickly adopt ed more restrictive licenses that forbade computer science professors from teaching w ith AT&T’s source code for fear of losing trade secrets (Kelty 2008, 120). Tanenbaum h ad to find or create a replacement, choosing to start writing his own in the same year as Stallman: 1984 (135). Between 1984 and 1986, Tanenbaum wrote a barebone but usable UNI X-like operating system for educational purposes, named Minix. He simplified th e details and architecture of UNIX in order to make a lightweight and elegant teaching tool, and he distributed his OS with source code (Moody 2001, 33-4). Where the official UNIX code dominated curricula previously, Minix code was quickly adopted in class rooms around the world in the late 1980’s, thus informing the greater part of a genera tion of hackers about what an operating system consists of.


65 Even though Minix was a glorified teaching tool, 40 ,000 individuals installed it functionally as their OS on personal computers, inc luding Torvalds (Kelty 2009, 135; Moody 2001, 35). As a teaching tool, Tanenbaum pref erred to freeze Minix development to keep it as lightweight and elegant as possible, but non-student users preferred to see it grow. These two audiences--one academic and illustr ative of the concept of operating system, and the second practical and dynamic with u sers’ needs for new hardware released in the 1990’s--caused the fork that birthe d Linux (Kelty 2008, 136). At the time he was a Finnish student, Linus Torvalds bought a P C and needed an operating system to install. GNU was theoretically perfect for a poor s tudent but incomplete without a kernel like GNU Hurd, and Minix was basic and resistant to change, creating a niche in consumer needs for Torvalds’ pet project. So, in th e fashion that Minix was a rewrite of UNIX for students, Linux rewrote Minix and establis hed a pragmatic development style geared toward user needs. In relation to human capi tal arguments, Linux would not have enjoyed nearly the success or revolutionary punch i f a significant cohort had not been educated with the open sourced Minix. Both Torvalds and his cohort benefited from the fam iliar conceptual scaffold of the Minix architecture, even if their product ultim ately contained no Minix code in order to avoid copyright infringement (Raymond 1999, 25). At this point, it becomes clear how open code is more valuable than simply a depository of free code. Although code reuse is powerful, knowledge reuse is ultimately more import ant because it builds on past efforts in the fashion of scientific communities. Sharing a lso enables programmers both to improve their code through feedback by other commun ity members and to gain


66 recognition for their efforts (Bonaccorsi and Rossi 2003, 1245; Koget and Metiu 2001, 250). In the long run it is impossible to overlook the im plications of education as a hefty human capital investment in software producti on, yet low marginal costs (consisting almost entirely of opportunity costs) m ake contributing code an easy choice in the short run. Minix did not evolve with PC hardwar e in the 1990’s, and hackers were willing to put in effort to build an operating syst em that did. The third chapter will question the serendipity of having hackers trained more extensively, when it examines hybrid business/open source models and government s upport of open source into the future; for the remainder of this chapter, I treat such costs as sunk. Assuming that contributors are rational actors, their motivations necessarily match or exceed the opportunity costs of some next-best use of the hack ers’ time. All that remains to be seen is what contributors gain, understanding their curr ent motivations. Intrinsic v. Extrinsic Motivation For such a new field as open source software devel opment, many other scholars agree with this micro-level approach of theorizing and measuring individual motivations. Scholars Josh Lerner and Jean Tirole collect and sy nthesize a lot of material, evaluating the existing literature. The traditional answer has been to take the profit equation at its most literal and derive net benefits from immediate and delayed payoffs, which is a good first attempt (2002, 213). This no non-sense approa ch sheds light on many of the lessimmediately obvious benefits for contributing to a public good that accrue to the individual contributors, underscoring the career or iented incentives of developing reputation and human capital skills and programming experience. In particular, co-


67 developing large and important open source projects often correlate with good future job offers. Lerner and Tirole especially like to trace the capital investments made by companies like IBM, Hewlett Packard, and Sun or the venture capital raised by open source companies like Eric Raymond’s Sendmail (2002 198). Citing start-up companies Sun and Netscape anecdotally, Lerner and Tirole sta te that frequent and voluminous contributors are recognized and rewarded with “read y access to venture capital” (2001, 822). The “rational calculus” approach to motivation, abo ve, has detractors including Lakhani and Wolf (2003, 6). Trying too hard to forc e contributors’ motivations into traditional economic molds threatens to give us a m odel that unintentionally warps reality into an incomprehensive caricature. Reputation is i mportant to contributors, but it cannot serve as a non-monetary replacement for wages or pr ofits, because enhancing professional status and status in the OSS community is a relatively unimportant motivation (Lakhani and Wolf 2003, Table 6, 23). Li stening to virtually any account of the hacker culture reveals that there is an element of joy, creativity, and community that is relevant to the production of open source code w hich is closer to descriptions of leisure. Remove these elements, and it is not clear that the dynamic production of software would occur at the same volume. Money shal l always be an important resource, because the information economy interacts with the national economy instead of operating as an island, but it is not the most impo rtant thing. Open source is also not antithetical to business, as mentioned in the first chapter and expanded on in the third, and this thesis is far more interested in finding t he coexistence and equilibrium state than in engaging an ideological battle between the two. However, understanding the


68 equilibrium between the two requires acknowledging motivations to produce public goods as they exist, rather than how we might ideal ize them. The career concerns previously mentioned are future -oriented investments, meant to generate some favorable outcome when firms recog nize and reward talent or determination. Such career concerns are extrinsical ly motivating because they rely on some indirect satisfaction of needs by outside forc es. Monetary compensation through wages are the quintessential extrinsic motivator of a capitalist society, for instance, because money is exchanged for something else throu gh indirect means. Many sources rely on Deci’s (1971, 105) succinct definition that “one is said to be intrinsically motivated to perform an activity when one receives no apparent reward except the activity itself” (Osterloh and Rota 2007; Lakhani a nd Wolf 2003; Frey and Jegen 2001). In contrast to extrinsic motivations, intrinsic mot ivations often result in immediate satisfaction of needs by undertaking an activity th at is “valued for its own sake and appears to be self-sustaining” (Osterloh and Rota 2 007, 164). Open source received such good play from the media precisely because intrinsi c motivations stemming from enjoyment and pro-social motives of the hacker comm unity were so unusual to business world decision-making, hinting at broader social ch ange. Interviews conducted by Sonali Shah reveal partial evidence against a career building approach. Because career and learning moti vations seemed so disproportionately unimportant to interviewees in relation to hypothes ized theory, Shah asked about them directly before finishing the interviews. Few liste d their open source achievements on resumes, and several hobbyists tempered or informed employers’ perceptions of their open source work to avoid allegations of neglecting their jobs for the unpaid project.


69 (Shah 2006, 1006, 1008). Only one hobbyist listed h is open source involvement on his resume, a student with presumably less job experien ce, yet the same student revealed that the language he used in the open source community w as rarely used in commercial development in his country (1008). For hobbyists an d long-term participants, other motivations were more intimately connected to parti cipants’ contributions, leading Shah to posit the importance of “fun and enjoyment” from contributing as an important primary motivation (1010). Economics as a social science is traditionally succ essful because it deals with the extrinsic motivations, which may elicit particular behaviors by manipulating external incentives. Social sciences like psychology, meanwh ile, devote more emphasis to intrinsic motivations (Frey and Jegen 2001, 591). H owever, public goods literature often details the failure of extrinsic rewards to always raise the level of contribution. Researchers Frolich and Oppenheimer found strong ev idence for the negative effects on contribution games when they instituted random payoffs in a series of contribution games to a pure public good (600). The potential ne gative backlash to rewards seems to come from undermining self-regulation and autonomy, both of which are important to the hacker culture. User Need Need and desire are powerful forces in open source, capable of spurring and sustaining action. Most contributors are users, wit h 58.7% citing a work or non-work (read: personal) related need for the software as o ne of their top three motivations for contributing (Lakhani and Wolf 2003, Table 6, 23). Individually, 33.8% reported only a work related need, and 29.7% cited non-work related needs; the contrast between work


70 and personal needs for the software were striking, however, as less than 5% listed both as a primary reason to contribute (11-12). Due to the way Lakhani and Wolf separate the contingents, other individual motivations outrank e ach contingent. For instance, 44.9% of contributors chose to code because they find the ac tivity “intellectually stimulating,” an intrinsic motivation rather than extrinsic need.Thi s introduces some interpretation into whether need is the most common motivation, but in either case it showcases an exceptional balance between the two types of motiva tion in open source. When empowered to provide their own needs, users ha rness intrinsic motivation that ends up benefiting far more than themselves. T he demand for a free and open operating system was so large amongst hackers that Torvalds found all the codevelopers he needed by posting a message on boards frequented by Minix users: Hello everybody out there using minix—I’m doing a ( free) operating system (just a hobby, won’t be big and professional like gnu) for 386(486) AT clones. This has been brewing since april, and i s starting to get ready. I’d like any feedback on things people like/dislike in minix, as my OS resembles it somewhat (same physical layout of the file-system (due to practical reasons) among other things). I’ve curren tly ported bash (1.08) and gcc(1.40), and things seem to work. This implie s that I’ll get something practical within a few months, and I’d li ke to know what features most people would want. Any suggestions ar e welcome, but I won’t promise I’ll implement them :-) Linus (


71 PS. Yes—it’s free of any minix code, and it has a m ulti-threaded fs. It is NOT portable (uses 386 task-switching etc), and it probably never will support anything other than AT-harddisks, as that’s all I have :-(. (Linux Online) As indicated, Torvalds’ needs were selfish and not altruistic, yet his impetus drew other users who wanted to see Minix grow past its pedagog ical use. His pet project was originally only meant to serve Torvalds’ specific a nd fairly “dinky” hardware, but there were enough peers with similar needs due to high ha rdware prices and abundant human capital (Moody 2001, 56). Although Torvalds did not promise to implement fea tures and code additions that he did not need, his choices in accepting contribut ions reflect great flexibility which led the project to grow so successfully. The noncommerc ial nature of open source puts userinnovators in a unique position to meet their needs through direct action. For instance, one quality that helped Linux catch on in the 1990’ s and 2000’s was the uninvasive capability of installing alongside a Windows operat ing system on one hard drive. Being forced to choose one over the other instead of dual -booting would limit experimental adopters who might find the costs of commitment too high. However, the decision to make Linux dual-booting was not a carefully conside red marketing ploy but rather a user desire brought to fruition by a competent programme r. In truth, Torvalds only wanted to keep MS-DOS (precursor to Microsoft Windows) in ord er to retain the ability to play a video game, the original Prince of Persia (Moody 2001, 56-7). This example from Linux’s earliest days is illustrative of the sponta neous and fortuitous adaptability and pragmatism of letting user needs dictate developmen t.


72 In hindsight, could this have occurred in a profit -driven, commercial context? The answer is, perhaps. Competition could easily have i nduced the lock-in mentioned above, wherein OS manufacturers might choose to leave cons umers choosing between one or the other. Apple pursues this strategy by only licensin g its Mac OSX on Apple hardware. Exclusivity helps multiple operating systems find n iches in markets, but Linux in particular has nothing to lose from competition. “Q uite frankly, I don’t _want_ people using Linux for ideological reasons,” wrote Torvald s, “I think ideology sucks. This world would be a much better place if people had less ide ology, and a whole lot more ‘I do this because it’s FUN and because others might find it u seful, not because I got religion.’ ” (Kelty 2008, 233). Linux development primarily exis ts to scratch an itch, not driven by licensing revenues and hardly by market share. The original desire to retain MS-DOS for a video game counts as a user need-driven experimen t that yielded a non-obvious, nonmarket-induced benefit that promoted coexistence. One of the most powerful features of open source d evelopment is its new way of allocating resources (in this case, mostly the prog rammers’ time, effort, and abilities) through direct action. Offering open source program s for free means that the usual market mechanisms for determining demand do not exist, bec ause there is no profit to draw in entrepreneurs to supply some willingness to pay. Ra ther than manufacturers driving the innovation process for market share and profit, the users themselves have great influence in setting the agenda and schedule for innovation ( von Hippel and von Krogh 2003, 213). In contrast to the imperfect act of voting with the ir wallets, true user-innovators are in a unique position to meet their own needs through dir ect action. Caveats to Direct Action


73 Normally programmers face a spectrum of possible s olutions and a balance between pragmatism and aesthetics that programmers navigate based on personal characteristics and tolerances. The Do-It-Yourself atmosphere embraced by hackers creates a bias for sophisticated users and programs that are typically not developed for mass audiences. Especially true of the 1990’s, open source development focuses more on technical projects like administrative tools or com pilers or web server software, e.g. Apache, than on intuitive graphical user interfaces (GUIs) or office suites or web browsers, e.g. Mozilla Firefox (Lerner and Tirole 2 001, 823). Because developers work on something they have an interest in, they are inc lined towards a solution that works for themselves in the easiest way possible, which may n ot be particularly user-friendly, especially to non-programmers. Befuddled users fund amentally have the option of making the program more user-friendly themselves, b ut in order to do so they must invest in enough human capital to alter it--by which time their personal impetus may be a moot point. After improving their technical literacy to the poi nt of correcting the usability problem, the problem no longer presents a personal grievance; to focus then on making a GUI more intuitive and user-friendly may indeed be altruistic in nature, and the literature broadly rejects explaining open source success thro ugh altruism. A related claim is that open source is only suitable for those things that qualify as “sexy,” or technically sweet and interesting to write (Raymond 1999, 44). The ch allenge is to determine how open source addresses the unsexy and mundane tasks, whic h is something Lakhani and von Hippel (2003) investigate in their work on support forums for the Apache web server. On forums, people post topics relating to issues with the software, often seeking help in


74 addressing the issue. Others, henceforth called inf ormation providers, post replies with solutions or informed guesses about the nature of t he problem. The punchline to their investigation was that information providers only t ransfer information they already know to questioners, meaning that their replies cost the m approximately 1-5 minutes. Rather, the vast majority of their time on the forum (98%) is spent reading posted questions and their solutions, helping information providers lear n about other Apache users’ issues to manage and update their own websites (Lakhani and v on Hippel 2003, 924). The majority of responses by information providers were cases where the provider already knew the answer (76% for frequent providers and 64% for others) or had useful information but no solution (19% and 28%, respectiv ely) (Table 12, 936).8 Learning and Community-Based Motivations Once more, the cost of contributing is so low (2% o f time on forums) and intrinsically rewarding because it builds a communi ty that freely shares and makes sharing more worthwhile. It is difficult to determi ne how well this case study translates to broader applications, but this example points towar d a broader link between intrinsic motivations of user needs and learning that inform the new open source paradigm of successfully integrating intrinsic with extrinsic m otivations. According to the surveys by Lakhani and Wolf, 41.3% of official developers rate the motivation to improve their programming skills as one of their top three, makin g it the second most common motivation that narrowly trails an enjoyment-relate d intrinsic one (2003, Table 6, 23). As 8 Information providers are largely or completely in dependent of the Apache Group and key developers. Rather, it is another way of incorporat ing another level of technical literacy that is not as source-capable as developers but still more comp etent than other users.


75 previously mentioned, depending on the reason for u ndertaking the human capital investments, learning can either be an intrinsic or extrinsic motivator. The Apache support forum example indicates the former, while b uilding a product line to attract venture capital indicates the latter. Looking at how the costs are low to be an informati on provider on Apache support forums, it is important to realize once mor e that costs decrease as one contributes more. The first time one composes a reply, the post may take longer, but one can save the post in an archive and draw upon it for another inf ormation seeker with very minimal time commitment. Frequent information providers wer e often long-term participants, showing an alumni effect that may translate to othe r open source efforts (Lakhani and von Hippel 2003, 931). At this point, although it d oes not rank as one of the most important or common motivations, concepts of commun ity are important to the hacker culture that develops open source. On the face of i t, community-based motivations did not rank highly on Lakhani and Wolf’s surveys: 28.6 % felt a strong, personal obligation to give back because of previous use of F/OSS, and 20.3% chose to work on open source because they liked the development team. Even fewer take Stallman’s ideological stands against proprietary software; only 11.3% state that they want to defeat proprietary software by contributing to open source projects (L akhani and Wolf 2003, Table 6, 23). By the nature of supporting open source, contributo rs fulfill communityand obligationbased intrinsic motivations. Enjoyment-Based Motivations Many of the motivations listed above can have eleme nts of enjoyment in them, which may be the most archetypal intrinsic motivati on. Because coding is easily a type of


76 work that could command a wage, it may not be immed iately apparent why programming should be a hobby as well. Yet rock climbing is ano ther example of something strenuous that is never the less intrinsically rewarding to p erform, making it important to learn how to measure indicators of enjoyment (Lakhani and von Hippel 2003, 927). Lakhani and Wolf link enjoyment to creativity, following decade s of psychological studies by Csikszentmihalyi which propose that enjoyment is ma ximized when participants enter a state of flow “characterized by intense and focused concentrati on; a merging of action and awareness; confidence in one’s ability; and the enjoyment of the activity itself regardless of outcome” (2003, 4). They establish a baseline of flow and enjoyment in their survey by confirming that programmers lose tr ack of time when programming because they enter this flow and by seeing how many programmers would use a magic 25th hour in the day for programming if given it. 7 2.7% responded that they always or frequently lost track of time during programming se ssions, and 60% said that they would use the extra hour to program (Table 5, 22). Discounting the aggregate of work and non-work need for the open source code, the most common motivation in Lakhani and Wolf’s su rvey of official developers was that the coding is intellectually stimulating to wr ite, shared by 44.9% of all respondents (2003, Table 6, 23). It is entirely possible that c reativity and challenge can also circle back to extrinsic motivations such as reputation. E go gratification is hard to achieve with easy, boring, or repetitive tasks, and challenging tasks or ingenious code work best for garnering the type of recognition that can help car eer advancement (Lerner and Tirole 2001, 823). Even through both enjoyed what they wer e doing, it is undeniable that Torvalds and Stallman both benefited from ambitious undertakings. Torvalds landed a


77 job with the secretive Transmeta Corporation for ad vanced computer chips (Moody 2001, 28-9), and Stallman received a large McArthur Found ation fellowship that he uses to live on and support his freedom to code (117-8). Abundantly clear in Torvald’s original posting abou t Linux quoted above is the lack of very formal schedules--it will be updated w hen there is something to update. Where commercial enterprises thrive on deadlines, n aming and sticking to a release cycle is not conducive to the volunteerism reliant on fac tors like fun or enjoyment. Rather, open source development inspires a creative excitem ent that is more sustained and less extreme than the commercial crunch before releases (Moody 2001, 202-3). As pointed out in the first chapter, one of Stallman’s signifi cant motivations in starting the GNU project and the Free Software Foundation was the lo ss of community and fun that he had experienced at the AI lab. The Apache Group oversee ing the web server software’s birth and committing changes to the code base in the earl y years were entirely volunteers who held other jobs (Mockus et al. 2002, 317). Voluntee rs continue to supply a significant proportion of open source code. A few projects, such as the Ubuntu distribution of Linux, prove the exception to the rule of no schedules, primarily because the sch edule allows them to designate stable releases for long-term support. Such support is nec essary for winning over relatively unsophisticated users who are not ready for experim ental versions that may splice in code that could break the program or cause malfunctions. One of the subversively productive elements of lacking a release cycle is that everyth ing can be released as it is ready, early and often, so that code evolves faster as the exper imental versions constantly induce trials by fire. Accepting the limitations of fast-moving c ode constantly being replaced meant


78 disseminating programs to users who could enjoy the latest features, stress the code to discover issues, and use the source code to identif y and correct issues with unprecedented rapidity (Raymond 1999, 28-30). As Raymond explains “Linus was keeping his hacker/users constantly stimulated and rewarded--st imulated by the prospect of having an ego-satisfying piece of the action, rewarded by the sight of constant (even daily) improvement in their work” (29). By using Source Co de Management systems (SCMs) that could organize and disseminate both experiment al and stable releases of software, open source can experience the best of both models, winning over sophisticated and unsophisticated users alike.9 Although we’ve established that nearly all contribu tors are users, obviously not all users are contributors. Indeed, download counts att est to the disparity of users and contributors, and all users who do not give money o r code or add some other value like information providers on support forums are free ri ders according to the standard economic definition. However, free riders’ presence does not threaten software’s production as it does other public goods, so our qu estion is, why? Non-contributing users cannot take away what the intrinsically-motivated c ontributor already gained, because the reward is intrinsic to the act of contributing, but the OSS code’s widespread use can only increase the product/community’s value to the contr ibutor. On one hand, unsophisticated users are the majority, and their use of open sourc e software increases market share that plays into the extrinsic motivations of career-orie nted individuals by enhancing 9 Commercial companies obviously do not benefit from faster bug fixes, because they do not share source code with users. However, it is not im mediately apparent that the open source method of staggering stable and experimental releas es is necessarily better. Tests benchmarking GNU/Linux against Microsoft by Mindcraft in 1998 sh owed the commercial operating system to be superior (Moody 2001, 282). Conflicting reports hav e resulted since, but tellingly hardware manufacturer IBM announced on January 10, 2000 that all of its server platforms would be made Linux-friendly (289).


79 contributors’ reputations to be leveraged into bett er jobs or venture capital. Simultaneously, free riders often feed intrinsic mo tivations by acting as bug testers who report their issues with software (experimental or stable releases), with the more technically literate best able to help diagnose the problems with the code base. Free riders also inspire the community to unite because success es are contagious, and the development paradigm enjoys so much renown today be cause it learned to address unsophisticated users better and hopes to improve s till. The first two chapters attempt to inspire the reade r with promise for open source development, acknowledging an alternative paradigm for structuring society’s productive capacity. Explained to the best of my ability, open source taps into the non-monetary motivations that impassion and fulfill individuals, and it provides a dynamic way of interacting with intellectual property that seems e volutionary. The next chapter handles the implications of open source on the business and public spheres more deliberately, striving to understand where the paradigm belongs i n a capitalist society. The intersections of hybrid models and government suppo rt provide a crossroads by which I can start to answer that question.


80 Chapter 3 The previous chapters explore what open source com munities are and how their distinctive innovation model differs from what came before. Requiring a structural break from classic software producers--namely commercial, proprietary firms--is useful to an extent, but now I shall reintroduce it to a dynamic environment with an orientation to the future. OSS captures the attention and imagination precisely because of its effects on the business world, reverberating across the business l andscape rather than dying out in a vacuum as a dud experiment. Instead of debating whe ther OSS will continue to exist, the next phase of scholarship is how: by determining wh at the future holds for the movement and whether its effects on software will be as bene ficial as the traditional, commercial firms. Since the turn of the millennium, government s grow more intrigued by open source. As the largest single consumer of software products, their adoption or support of open source would signal a massive restructuring of capital depending on the extent of the policies. Accepting the promise of open source need not dismantle the competitive market, but it would represent a paradigm shift tha t necessitates a more nuanced view of OSS than presented in the previous chapters. Where Do We Go From Here? Even if governments retain commercial software, th e open source sector has and will continue to shift the commercial paradigm. Due to large network effects, the software industry is notoriously winner-take-all, b ut the large payout makes it lucrative to compete (Evans and Reddy 2003, 330). However, margi ns for second-best commercial products are further reduced by free OSS-licensed a lternatives often of comparable


81 quality, leading software and hardware firms to ada pt to the competitive force of OSS. Many scholars are concerned with how open source ma y reduce incentives in the commercial sphere and ultimately hurt social welfar e, but experience indicates that neither the proprietary nor open source model is li kely to extinguish the other, except in distinct categories of software. Many important com panies employ hybrid business models that marry open source and commercial elemen ts in nontraditional ways: specifically, value-added service enabling, loss-le ading, leveraging collective software development, pro-social incentives. The following s ections focus on how firms incorporate open source in their business strategie s, in order to better understand the ramifications of an expanded open source sector. Af ter identifying some coexistence of business and open source ventures, the chapter turn s toward government policy to tease out what public sector involvement would bring. Value-Added Service Enabling First, commercial firms can extract value from serv ices built around open source software: value-added service enabling Although Linux is available free of charge, several companies nevertheless subsist or profit fr om the sale of particular distributions. Red Hat is arguably the most famous and prevalent e xample, garnering publicity from its successful initial public offering in 1999 that rea ched a market value of $3.5 billion on the first day (Moody 2001, 228). This strategy taps into the entrepreneurial spirit to address issues of usability for less technically-so phisticated users and does not diminish the value of Linux as a public good--those who want it for free can still download it themselves, even the Red Hat distribution. As Glyn Moody phrased it, “This was not a question of programming, but packaging...” (2001, 9 7). Customer support rather than


82 new code is Red Hat’s money-maker--like Google’s ad vertising revenue underwriting its programming efforts--because Red Hat releases all o f the code it produces under the GNU GPL, showcasing the broader implications of an open model on business. Commercial entities are particularly vulnerable to malfunctions and thus pay to ensure that their downtime and technical difficulties are quickly resolved, while non-paying consumers still benefit from Red Hat code.10 Beyond technical support, many companies are starti ng to favor reliable and low cost OSS implementations for their business strateg ies. Creating open source platforms on which to offer services not related to software provision eliminates the need to pay licensing fees or develop software in-house. Custom ers experience lower market prices, for competitive forces ensure that companies do not accrue all of the cost reduction as unbridled profit. Amazon and Google are prominent e xamples of this savvy, new breed of company that capitalizes on adapting existing OS S to the firms’ needs (Osterloh and Rota 2007, 165). Even if companies like Amazon conc eal their source code, legally they are in compliance with GNU GPL licensing because th e code is not redistributed but used internally to offer a service (Fitzgerald 2006, 591 -2). Returning to the collective definition of open source maintained by the Open So urce Initiative, a key element is that licenses cannot discriminate against fields of ende avor (OSI). In other words, licensing cannot restrict how or why users choose to operate the program. Most noticeably this includes commercial uses, a functional requirement that gives businesses a reason to use and contribute to the open source sector. 10 Red Hat is, coincidentally, the largest corporatel y affiliated code contributor to the Linux kernel, according to a Linux Foundation white paper (Corbet et al. 2012, 9-10).


83 Loss-Leading Commercial entities also gain in the long term by u sing open source strategies as a loss-leader to grow existing markets or to create n ew ones Software is useless without appropriate hardware to run it, leading IBM to beco me the leader in increasing demand for complementary goods and services like server ra cks, capable of hosting Apache web server software (Osterloh and Rota 2007). By distri buting servers with free and open source software and contributing to make programs l ike Apache the leader, companies can lower the costs to consumers on servers. Even i mproving Apache is less expensive than every hardware manufacturer coupling the hardw are with proprietary software, although Windows and Mac servers prove that there i s a specialty market for proprietary server software. Furthermore, when IBM put GNU/Linu x on all of its enterprise hardware, the unification of systems into a single OSS family simplifies serving all of the proprietary systems it previously managed--doing aw ay with the consequences of incompatible software--and offers scalability from portable computers to supercomputers and servers (Moody 2001, 291-2). Loss-leading also manifests more subtly, by setting a standard and enabling compatibility between products. Windows and OSX ope rating systems are inefficient because they require different code to run the same applications on each OS, but the market is flexible enough to allow each a niche, al though fragmentation does not seem to support more than a few more like Linux. One of the early problems with email, before the Internet existed as a worldwide web, was making mail compatible among different networks, leading Eric Allman, a graduate student a t University of California at Berkeley, to write and open source the simple mail transfer p rotocol (STMP) called Sendmail.


84 Sendmail’s success grew with the Internet, and in 2 000 Sendmail was estimated to handle 75% of all email traffic (Lerner and Tirole 2002, 2 11-2). To support his open source development, Allman started Sendmail, Inc. with Gre g Olson by offering full, enterpriselevel support and commercial versions of Sendmail w ith new features. Much of Allman’s Sendmail Pro success was dependent on the thorough market penetration that accumulated around the free use of his open sourced code, when it became the standard (Fitzgerald 2006, 589). Dual Licensing A more refined method of Sendmail’s commercial vers ion is dual-product licensing, seen in projects like MySQL. MySQL raise d venture capital for a business plan to license the exact same project separately under open source terms and conventional proprietary terms--differing from Sendmail Pro’s ad ditional features. Licensing under two different terms requires that the company retai n full control of their corporate and not accept code contributions from the community, meani ng that the only benefit to the company is a larger user base that uses the free/op en source version. As explained by Michael Olson (2008), “dual licensing is an innovat ive strategy that combines open source distribution with proprietary licensing,” as opposed to open sou rce development (italics indicate my emphasis). Because only the or iginal intellectual property owner has the right to offer the IP under different licenses to different customers, the dual-licensing firm must retain full rights over the code and cann ot accept contributions from an external development community. Hence, open source is actually only convenient as a mechanism for ensuring free prices that draw custom ers and establish a user base, thereby demolishing the dynamics of cooperation and public good this thesis bases itself on.


85 Other important dynamics persist, making the hybrid strategy of dual licensing different from any established model prior to the o pen source movement. Revealing source code automatically changes the game from pro prietary secrets kept by firms, but the legal infrastructure of copyrights and patents protect MySQL and other dual-licensors from blatant infringement or piracy. The choice of open source license proves very important for limiting direct commercial competitor s and for encouraging some users to pay for a different, commercial license. Ironically the firms wanting to commercialize their products are incentivized to release it under the most restrictive licensing terms: the GNU GPL. The distinction between restrictive and non-restric tive licenses is vital to the dual licensing model. Thus far, licensing in the first c hapter focused on the generally restrictive GNU General Public License (GPL). By re quiring that redistributed code or code interfacing with GPLed software be distributed under the same license, code additions can only be licensed under the GPL or not distributed at all. To review, the viral or reciprocal nature of the GPL benefits the dual l icensor because it keeps the code base and additions perpetually non-commercial and open s ourced under strict and potentially costly terms to other commercial enterprises. Non-r estrictive or non-reciprocal licenses-often called academic licenses11--do permit commercial derivative works, which make such rights more valuable to the collective but mor e costly to a dual licensor. Neither restrictive or non-restrictive licenses are mutuall y exclusive to the open source definition, but releasing the software as open source can act l ike a loss leader; the price for most 11 “Academic” because they often originated at resear ch institutions like University of California in Berkeley. Secondarily, such licenses mirror the sci entific academic community’s tendency to allow commercial entities to profit by incorporatin g academic discoveries in their products or production.


86 users drops to zero, often achieving broad market p enetration that dissuades competition for the remaining commercial licensees. Additionally, a restrictive license provides the du al licensor a stick to encourage people to opt to pay for the second commercial lice nse instead of the free/open source one. By introducing painful or restrictive terms of acceptance to open source licensees, a dual licensor segments a niche market populated by commercial enterprises that want to avoid the viral nature of the GPL, that is to inter face with the product without open sourcing their own code under the GPL. The dual lic ensor then profits from willingness to pay to avoid this fate, or as Michael Olson (200 8) explains it, “Open source licensees pay no fees, but make certain promises. Proprietary licensees have different rights, and pay a fee for that consideration.” The dual licensi ng tactic thus works best with certain types of fundamental, foundational software that ot hers want to build upon and redistribute, in contrast to the service enabling a spects discussed above in relation to Amazon. It also requires the consolidation of intel lectual property ownership that eliminates the collective, external development tha t is another major perk of open sourcing code, making dual licensing an interesting but specific application of a business/open source hybrid model. Perhaps most imp ortantly, dual licensing shows that commercial and open source is not as dichotomous as many believe or present it. Leveraging Collective Software Development Unsurprisingly, forgoing community support and mer ely using the open source license as a way to distribute and gain market shar e is helpful in some instances for commercial gains but still costly in a proprietary sense. In relation to traditional proprietary models of software development, to bene fit from external development


87 support is more revolutionary (Osterloh and Rota 20 07). Firms can carefully and strategically select what elements of a larger code base to open source and what elements to keep closed, weighing the benefits of external p articipation against the costs of losing an edge in what is shared. Apple and Google are both excellent case studies of leveraging collective software development to improve their market positi ons, although Apple suffices to be illustrative of the tactics. Apple open sourced the bottom layer of its OSX and iOS operating systems to draw upon the ideas and coding help of volunteers, called Darwin. Extra functionality, either innovative or increment ally improved, only increases the draw for volunteer developers, creating a lower cost vir tuous circle that improves the final Mac OSX product (Fitzgerald 2006, 592). Licensed under the Apple Public Source License, Apple drew upon the esoteric interests of the open source community for the technical and infrastructural. At the same time, by designing their own license rather than using the standard GNU GPL, Apple is also able to specify tha t Darwin code can be linked to commercial and proprietary files, i.e., the other p ieces of the Apple’s OSX and iOS. Apple retains control and secrecy around the user-f riendly graphical interface that gives it a competitive edge against Microsoft’s Windows OS a nd open source alternatives, ensuring profitability. Google’s Android OS similar ly retains effective control over Android-powered smartphones by retaining key code a nd trademarking the Android logo, but its relatively unrestricted terms of service fo r the app marketplace typically buy it more goodwill amongst the public.


88 Gated Source As dual licensing was a specific corporate-open sou rce model presented above, centered around using open source as a distribution point, another hybrid expression is opening the code to a community of users to help in novate but using licensing terms to retain functional ownership of all code created by the community. To maintain profitability, companies cannot release their IP as completely as open source projects do, thus giving birth to middle ground between the prop rietary and open source extremes, here exemplified by Sonali K. Shah’s gated source m odel. Presumably named to play on the themes of gated residence communities, corporat e owners can sell licenses that allow users to view, use, and modify code but stipulates that code-sharing may only occur with other licensees (Shah 2006, 1002). As with all open source models, the underlying intellectual property is fully exposed to prying ey es but ultimately protected from rival uses and piracy because only users who follow the t erms of the license, by paying, are allowed to use the code. Theoretically, what exists in the gated source comm unity could be replicated and rebuilt completely by an open source community, so long as the copyrighted code were rewritten entirely. Even those who use and particip ate in gated source projects substantially prefer open source, if only they can find a product of similar quality; ultimately their decision is overruled by the techn ical proficiency and superiority of the gated source software, for which they accept the re strictive licensing terms (Shah 2006, 1004). Overall, gated source appeals to highly tech nically-literate users who do not have reason to go out and build their own programs from scratch and also do not have the patience to help raise the best open source alterna tive (if there is one) to equal footing.


89 Normally these pressures come from working commerci ally for firms that need the best results at the lowest price and the quickest perfor mance. Therefore, we are talking about a minority of a minority, the original minority of us ers of open source software and another minority inside this group pressured to make things happen. Despite this minority status, the issues are indicative of broader trends that he lp us to predict what will happen as more entities release their source code. Capturing the efficiencies of the open source devel opment cycle, the gated source model reveals source code that users can innovate a nd improve code as the needs and desires suit them. As stated above, all gated sourc e users start as need-driven participants, but Shah hypothesizes that the needs-based particip ants can evolve into hobbyists, who participate in development for enjoyment, because o f their growing expertise with the software. These needs precipitate code creation, be cause the participants will typically solve their own issues if investigation reveals tha t no other members resolved the problem yet (1005). Of course, rather than merely f ix their problems on their own machines and leave it at that, participants reveal their own problems and solutions to the collective for all of the reciprocal advantages men tioned in earlier chapters: receive feedback, start or sustain development in direction s they want, and to communicate their needs to the rest of the community (1005). Nearly every long-term gated source participant cit ed need as their primary reason for continued involvement with that collective deve lopment model as opposed to the fun and enjoyment prevalent in the open source model (1009). In gate d source communities, there are no hobbyists to speak of to handle the mo re mundane tasks like organizing volunteers (in this case, other users of the gated source software). Predictably, this lack of


90 hobbyist-motivation leads the corporate sponsor to hire employees to handle maintenance tasks, which further dissuade larger and more inclu sive volunteer efforts because the firm employees dictate the direction of software develop ment. For an economist, this brings forth many questions relating to the costs of corpo rate code ownership instead of treating it as a public good, because employee wages serve a s a cost of keeping the code proprietary that have to be recouped by the firm ra ther than donated by the individual contributor and sunk costs. Importantly, interviewees in Shah’s study pointed o ut that working on the gated source project is ordinarily solving a small proble m, so participants typically do not develop skills when working in the collective devel opment model of gated source (1006). To the extent that learning occurs, participants te nd to gain specific—rather than general—knowledge about the project, which over tim e can develop into extensive knowledge of the content and structure of the overa ll code (1007). This accumulation of specific knowledge can create more complex and rewa rding challenges for the hobbyists, which leads to a different theory of participation amongst more localized and uncommunicative constituents than in the broader op en source community. A more relevant motive for engaging the collective, accord ing to the interviews, is fun or enjoyment, per the second chapter. More than half o f the long-term open source participants described their work as fun or likened it to a hobbylike activity: a jigsaw puzzle, chess, rock climbing (1006-7). Long term pa rticipants—hobbyists, in particular— tend to be highly skilled and experienced software developers, many of whom hold managerial positions at companies using the open so urce project’s software (1008). In contrast, need-driven participants vary widely by s kill (1005). Thus, the gated source


91 community theoretically will not engage participant s as easily as the open source commonly described, but the gated model is more ope n than traditionally copyrighted software and its source code access benefits some t echnical users. Corporate ownership of the code--as opposed to the effectively collective ownership of open source--prompts very few particip ants to spend their efforts rolling out code that may not be available to them or others la ter if the firm goes out of business or otherwise revokes its licenses (1012). Whereas open source lets select volunteers called committers alter the source code, a gated source co mmunity vests all power to do so with the corporate sponsor and its agents (employees). A cting as the final point of appeal and authority affords the firm considerable leverage ov er the use of volunteer contributions and the official development direction of the proje ct. This does not prevent the userinnovators from sharing amongst themselves, but it does create power dynamics and a formal hierarchy that is not found in the consensus building of open source communities. Although gated source community members contributed code to the project, interviews revealed that many gated source develope rs chose not to contribute code that they had already created, for several reasons (1010 ). Developers mostly worried that their needs were secondary to the firms’ needs, and the l oss of autonomy strongly affects hobbyists’ capacity to contribute in self-defined w ays (1011). Keeping code to oneself leaves other developers with less common code and o f poorer quality, curbing further collaborative efforts. Furthermore, developers’ dis taste of tightly controlled projects was documented even among open source participants, so the control wielded by corporate owners of the code should be anticipated to be offputting (1012). All-in-all, there are


92 significant restructuring of community dynamics fro m this method of appropriating value from revealing source code and leveraging community involvement with code creation. Pro-Social Incentives Finally, firms do benefit from influencing how the y are perceived by the open source community, often by open sourcing their code Much of Google’s popularity amongst techies is due to the company’s broad suite of free tools, and its Android OS for smartphones deliberately courts community developer s and an unrestricted, hands-off app marketplace. Leveraging the open source brand is si milar to using it as a loss-leader, but the implications are deeper than profiting from the wide diffusion of the respective software. On a simple level, revealing code per dua l licensing may only ingratiate users, pandering to public “trust through transparency” th at does not garner goodwill toward the firm, per say (Fitzgerald 2006, 594). Conveying pro -social intentions represents a commitment to a new development environment that ta kes advantage of the OSS community’s involvement. Signaling good “OSScitizen ship” means not merely revealing source but also ceding control, thereby enhancing i ndividuals’ autonomy and feelings of competence and social relatedness and allowing open source dynamics to flow (Osterloh and Rota 2007, 165, 169). Trustworthiness from util izing the OSS brand responsibly can help firms achieve market-leader status (Fitzgerald 2006, 593). As an illustrative case, Netscape and IBM embraced open source in different ways that impacted their reception by the community: Netscape from relative weakness a nd IBM from marketplace leadership. Netscape open sourced earlier than IBM did, normally the definition of a trend setter, but out of necessity and essentially calling on open source communities to salvage its work. IBM open sourced from a position of power that commentators saw as


93 making it “one of the good guys” and a progressive company, even by Netscape CTO Eric Hahn (Moody 2001, 206-10). Transition to Government Policy A base knowledge of open source and commercial hybr id business strategies above helps discern how companies will fare in the future, especially with expectations that the open source sector shall continue expandin g. Creative destruction is likely to result, with some companies succumbing to the chang e, but the basis of competition will survive unscathed. Because government is the larges t single consumer of software, the largest shock to the system may come from governmen t adoption and support of open source (citation). Following Linux’s success at bri nging OSS to the mainstream in the 1990’s, government interest in the model took off a s early as the new millennium. Various policies have been proposed, from mandates to subsidization to active contribution. Fittingly, IBM provides a analogous e xample of widespread, official support, creating a business-world precedent for su ch a mammoth conversion. IBM announced in June 1998 that Apache would be shipped on its servers, which Glyn Moody calls “probably the single most important boost the proponents of free [and open source] software could have wished for” (2001, 206). At the turn of the millennium, January 10, 2000, the company announced that it would make all of its server platforms Linuxfriendly and took Linux to new heights on the might iest and most prestigious corporate mainframe computers of the era (289-90). IBM acknow ledging Linux and Apache were important landmarks in the evolution of OSS as a co mpetitor to commercial software Cautious optimism suggests that government support or adoption of open source would likewise introduce creative destruction to th e software market. Far from


94 cataclysmic, IBM’s lead has not devastated the comm ercial software firms, and the innovative incorporation of open source elements in to business models discussed above suggests that coexistence may be the new norm. Mikk o Mustonen (2003) explicitly crafts a model that allows for the coexistence of open and closed source software under conditions of higher priced commercial licenses. Cu stomers on the fringe, close to indifference between adopting a commercial product or not, will always benefit from a copyleft program on the market, because even when t hey are indifferent between a copyright program and nothing, they will have a pos itive surplus after licensing costs drop to zero under an open source scheme (Mustonen 2003, 110).12 Welfare can be higher from the mere presence of open source compet ition, although in some cases users who may benefit from OSS may not even be aware that a low-cost alternative exists. When consumers are cognizant of the copyleft/open s ource program, welfare will always be higher on account of those deriving surplus from its use, even if prices of the commercial product rise to maintain firm profitabil ity (117). Therefore, more pervasive awareness of OSS that already exists always increas es welfare. Part of a government’s raison d’etre is to maximize citizens’ welfare, so the case for government cognizance of open source is apparen t, if not its outright involvement. What policies governments should pursue is less cle ar cut. Governments have the power to provision public goods like OSS from tax revenue s, but doing so often raises specters of debates about the capitalist system’s superiorit y over controlled economies. Command 12 A caveat to this analysis is that although the pri ce of a license is zero in the open source framework, consumers and providers must still consi der the implementation costs--e.g. the open source project may have high distribution costs or a very unfriendly user-interface that necessitates training. If it is costly to adopt the open source alternative, commercial firms can continue pricing the closed source software as if t here were no OSS alternative (Mustonen 2003, 113). Luckily, the cost of distributing OSS is univ ersally low because of the Internet, meaning that implementation costs can be almost zero sometimes.

PAGE 100

95 economies like the USSR failed, and industrial poli cies have proven lackluster through examples like Japan succumbing to its Lost Decade ( Evans and Reddy 2003, 316). Governments are no better equipped to determine win ners and losers than a market--in terms of either products or standards--so the major ity is in consensus that government involvement in OSS is a dubious or dangerous propos ition. Evans and Reddy are particularly suspicious of the top-down approach of government involvement, writing that, “When legislators get involved, these decisio ns are moved from the strictly technical/economic arena to the political,” indicat ing that the decision-making passes from typical firms’ interest in profits to other co nsiderations like public approval (2003, 372). By taking a more nuanced approach to governme nt involvement instead of the broad command-versus-market approach of provisionme nt, the rest of this chapter argues for some forms of government support that do not di srupt the particular efficiencies of the open source model of innovation and production. As long as a few rationales withstand immediate rejection--e.g., explicit cost savings on free licenses--it is worth considering when government should permit itself to support ope n source and in what fashion. Typically the policies discussed are mandating that government employees use OSS, subsidizing its adoption, funding its developm ent, or providing information about the availability and true characteristics of OSS. B efore any policy is implemented, however, governments and economists both need to un derstand the implications of an increasingly large and healthy open source sector o n social welfare, but sweeping conclusions are difficult to arrive at. Policy reco mmendations, as a result, become more nuanced and complicated by specific instruments of governmental response and particular circumstances and contexts in which they operate (Lerner and Tirole 2005b,

PAGE 101

96 111). Many economists (Evans and Reddy, Smith, Schm idt and Schnitzer) worry about the implication of open source competition drawing resources away from commercial enterprises, but experience seems to bear out that open source acts as a positive competitive force rather than a drain. This is true at least as long as open source coexists with commercial enterprises and does not push them out of business. The degree of innovation may yet be a factor to consider, in term s of social welfare. Imperfect Information, Market Failure So if consumers can be better off by realizing tha t open source exists or knowing its true characteristics, who will inform them of i ts existence? On the whole, copyleft programmers do not derive benefit from free riders using the software without contributing (code development, bug fixes, support forums, etc.), meaning that copyleft programmers are indifferent as to whether consumers use their product or not (Mustonen 2003, 117). In a less simplified expression, the on ly user base that programmers care about is other technically skilled users who may im prove the collective product or introduce new OSS to the commons. Commercial compet itors obviously have nothing to gain from informing customers of cheaper alternativ es, and they will have to hire more programmers and increase quality in order to staunc h losses due to migration to the alternative (117). In some cases, the emergence of OSS in the public consciousness can drive a commercial firm out of the market altogethe r. Only when an open source product is the de facto standard can commercial company’s a dvertisements informing consumers about the OSS product’s true properties be benefici al for the firm. Mustonen’s model creates a dichotomy between open source and commerc ial that does not take the hybrid models (above) into account, yet his work is nevert heless illustrative of broader

PAGE 102

97 competition analysis in the more traditional (or re volutionary) conception of open source versus proprietary. By far, the least intrusive policy that government s can implement, then, is “providing information about the existence and char acteristics of OSS to uninformed consumers” (Comino and Maneti 2005, 225). A univers ally recognized basis for government involvement is correcting market failure s, readily conceded even by critics of government OSS such as Evans and Reddy (2003, 368-9 ). Thus, Comino and Maneti couch their recommendations for government involvem ent in the logic of a market failure of imperfect information. Building from foundations like Mikko Mustonen, they observe that open source communities have fewer resources a nd incentives to advertise than closed source firms and realize that governments ca n correct this failure (Comino and Maneti 2005, 218-9). Strangely, Comino and Maneti ( 2005) cite Mustonen in their bibliography, yet they assert that previous literat ure ignores this market failure. Lack of information about open source is an explicit compon ent of Mustonen’s model, and one of his major implications was that neither open source communities nor commercial firms would serve information needs adequately (Mustonen 2003, 117). However, Comino and Maneti improve the discourse by giving government t his potential role, and thus are not entirely disingenuous in their presentation. Informational campaign policies are, then, legitim ate governmental policy, but they can have different results based on the types of software and their network effects. Where strong standards are enforced, interoperabili ty between programs will be high and consequently low network effects; for instance, ema il protocols make it easy to communicate between several mail programs like Micr osoft Outlook or Mozilla

PAGE 103

98 Thunderbird so that one program’s adoption does not significantly affect another program’s users. However, as network effects become stronger, there will initially be a decrease in welfare as the OSS adopters diminish st andardization before an increase as the market tips towards open source products (Comin o and Maneti 2005, 232). One policy conclusion resulting from this analysis is t hat promotional campaigns for OSS software with large network effects should be bold and pervasive, rather than timid or localized. In this context, campaigns to adopt Linu x, an operating system with high network effects, should reach as many as possible. Mandates, Strong-arming Adoption Promoting open source for the good of the consumer can also happen with government mandates, although the conditions for su ccessfully raising welfare are more tenuous. Mandates force some or all public agencies universities, and schools to adopt OSS, notably China’s push to adopt Linux and other OSS (Lerner and Tirole 2005b; Mustonen 2003). This policy ignores consumer prefer ences and does not distinguished between informed and uninformed users (Comino and M aneti 2005, 225). Splitting the two groups of consumers into camps of informed and uninformed users: if there is a greater ratio of uninformed users who would benefit from adopting the OSS, then the OSS mandates would increase welfare (226). To the e xtent that non-professional users are generally ill-informed of the characteristics o f open source, Comino and Maneti suggest that there is room to believe that mandates may sometimes have positive effects. They present this in contrast to Schmidt and Schnit zer (2003) by suggesting that those economists do not consider the search costs for uni nformed and unsophisticated users prohibiting them from discovering the OSS alternati ve (Comino and Maneti 2005, 227).

PAGE 104

99 As with information campaigns, network effects can have a large impact on the success of government mandates. High network effect s tip the market towards standardization of either a proprietary or an open source product, and disturbing a market equilibrium with mandates can have adverse effects. Therefore, if a government mandates only a moderate amount of uninformed users to adopt OSS, drawing customers away from the commercial product can hurt welfare; mandating many uninformed users can tip the balance towards open source and improve or maintain welfare (Comino and Maneti 2005, 231). In order for this to be the case the open source and commercial products must be of equal quality or the OSS must b e of higher quality, but such 1:1 comparisons are questionable outside of a theoretic al model (219). The inevitable counterargument is that mandate policies would alwa ys need to be informed of the relative quality of such programs, and competitive firms and markets are better suited than governments to choose winners and losers (Evan s and Reddy 2003, 387). Comino and Maneti respond preemptively that instead of enc ouraging government policies designed to pick winners and losers, the aim should be to guaranteeing a level playing eld. Providing unbiased information is the best ch oice, but mandates can be useful for overcoming switching costs that may lock software u sers into a particular product when they might otherwise switch (Comino and Maneti 2005 235). Government Funds, a Third Policy Arm A decent rebuttal to government mandates is that s uch policy may force nonfriendly software on general users who may find the OSS harder to use than commercial alternatives (Evans and Reddy 2003, 357). If open s ource user-innovators are truly unresponsive to consumers’ wants and needs and also lack entrepreneurial ambition for

PAGE 105

100 profits, then the market forces that drive a capita list market system to care for the mass market may disappear. Instead of concluding that a mandate will harmfully eliminate friendly user interfaces, another option is that co mplementary services can arise around the foundation per a value-added service enabling m odel. To gain market share, commercial developers have powerful incentives to p erform consumer research into usability and mass-market needs, but this observati on does not preclude such research and implementation from ever occurring in open sour ce, especially given how dynamic the open source sector is. Returning to the Chinese example of mandates, part of the plan is to develop its own Linux version that better accounts for the gove rnment’s needs (Evans and Reddy 2003, 377). It is difficult to imagine that the gov ernment would be wholly unsuccessful, disregarding human rights concerns that the governm ent may have plans to customize the state distribution for more than usability. While h ard pressed to argue that the government should ever provision all software as op en source, I would argue that government funds can be used successfully as capita l for improving code as long as it remains subject to the rules of the open source com munity: namely, governments have the right and proper incentives to participate in d evelopment but not to dictate design decisions/directions at the expense of democracy an d autonomy in the community. The nature of open code, free to modify and reuse, allows for different versions of a program to suit different audiences, e.g. there a re distributions of Linux specifically for lighter systems, like Fluxbox, and others for speci al purposes like education, such as Sugar seen on One Laptop Per Child computers. Typic ally each project is headed by a programmer with skill enough to integrate all of th e contributions into a coherent product,

PAGE 106

101 and nothing stops this from being someone appointed by the government to fashion the OSS in the interests of the state. (Indeed, when th e city of Munich switched to Linux from Windows, it customized its own distribution, k nown as LiMux.) These heads can wield almost despotic control over their groups wit h the power to be selective about what contributed code is accepted into the official code base distributed to users. Sustained and unpopular stances, however, may prompt discontents to start up their own groups which take the collective code in a new development direc tion. Rival projects based on the source code are called forks, normally starting wit h almost identical code bases but different philosophies to guide development that ca n cause significant variations in the long run. Hence, in order to maintain control, OSS development heads must be “benevolent dictators” who take their co-developers ’ opinions into account (Raymond 1999, 45). Linux development, incidentally, is a wonderful exa mple of this dynamic; even though Linus Torvalds enjoys an uninterrupted perio d as the official and functional head of kernel development with the almost despotic powe r to direct its evolution, his continued rule is ultimately subject to review by h is peers. The community of contributors trust Torvalds because his intuition f or kernel development has played out well over the years, but there was an important dis agreement in the late 1990’s that even threatened to fork the Linux kernel. The crux of th e issue was that a communication-only approach to development, say through email, works f or smaller projects but is inefficient and therefore cannot scale to larger and more compl ex ones (Mockus et al. 2002, 342). By 1998, Torvalds became tardy in committing import ant code. Meanwhile, another

PAGE 107

102 server--originally meant to mirror Torvalds’ own, n amed VGER13--instead hosted more up-to-date code that became functionally out of syn c with Torvalds' “official” version. “[T]he equivalent of a major boardroom power strugg le took place” that ultimately solved the issue by negotiating not to fork develop ment as long as Torvalds adopted a more efficient patch-acceptance system (SCM) than h is antiquated methods of manual code replacement (Moody 2001, 172-181). The VGER in cident’s successful resolution hinged on the ability to fork, even if the ultimate result was no split. Examples of actual forks abound, with some contentious examples such a s the LibreOffice fork in response to Oracle’s acquisition of Sun Microsystems and its OpenOffice intellectual property. From this, several implications about government be come apparent. Competitive forces are essential to OSS development, limiting t he usefulness of top-down authority in state involvement. Second The most disastrous OSS/g overnment mix is a policy simultaneously contributing to OSS and mandating it s use, because this destroys the democratic and autonomous nature of OSS development However, this is not to say that government cannot support developers financially to have the freedom to do what they do. Often, these contributors are the types willing to forgo some wages for the freedom to code, making it relatively inexpensive for governme nts to fund their production of a public good. Either policy alone avoids the perfect storm of unaccountability that results of mixing them, and mitigating circumstances may ev en make one or the other successful. 13 As with many open source projects, the name Vger h as an interesting but esoteric back story that makes remembering it easier. As explained by D ave Miller, “Remember Star Trek, where the spaceship named ‘Voyager’ flies past the screen, an d some of its letters had been scuffed off? The remaining letters spelled out VGER” (Moody 2001 172-3).

PAGE 108

103 Another point is that government should be careful in what they choose to split, for one reason the Linux fork did not occur is its large structural importance. Glyn Moody takes a rather strong stance against splittin g efforts on infrastructural projects. Whereas forking individual programs will not advers ely affect many people, split development tracks on a platform like GNU/Linux exp onentially increases difficulties with compatibility. Incompatibilities at a lower l evel affects all higher levels, in this case, programs on top of the operating system (Mood y 2001, 305-6). Therefore, the OpenOffice/LibreOffice fork mentioned above did not spark nearly the fuss that the VGER/Torvalds incident did, because the office suit e does not have as much dependent on its perfect stability. Overall, the user bases w ould be dependent on whether developers would fork to each version of the Linux kernel, and the more fragmented things get, the weaker and poorer all systems become as a result of weaker collaboration. Static versus Dynamic Measures (of Innovation) Supporting open source software through costless in formation campaigns automatically increases welfare in the short term, even if prices increase for the remaining commercial software users.14 Yet, in reducing the market of uninformed users, the price of commercial software can either increas e or decrease, but the firm will be affected by lower profits, smaller market share, or both (Comino and Maneti 2005, 2278). The fear is that this disincentivization of pro prietary software production will ultimately hurt the long-term success of the econom y. In terms of open source software, because users can fix bugs and add features to suit their needs, small innovations and 14 Comino and Maneti build a static model, hence my c omment about the short term.

PAGE 109

104 tweaks are not costly and are likely to continue ev en without commercial involvement; large or radical innovations requiring structural c hange, however, are likely to diminish if left entirely to user-innovator volunteers (Lerner and Tirole 2005b, 112). Commercial firms lose their ability to appropriate value from their innovations if intellectual property rights are eliminated, indicating that innovation c ould suffer as patent or copyright protections diminish. Critics of government support for OSS often challe nge open source software’s claim to being another innovation model by emphasiz ing its derivative nature. Integral even to Mustonen’s pro-adoption analysis is the ide a that copyleft/open source programming has a substitute nature in relation to commercial products (2003, 119). The GNU GPL is the dominant open source license, and th e majority of projects developed under the GPL are also imitative of commercial prod ucts. For example, OpenOffice is an office productivity suite similar to Microsoft Offi ce’s ubiquitous set of programs, based on StarOffice commercial code and released by Sun M icrosystems as a loss leader in the same way Mozilla’s Firefox browser is derivative of Netscape Navigator (Evans and Reddy 2003, 354). Whereas most scholars seem to und erstand this facts through the lens that open source eliminates commercial profit, I di sagree that OSS is inherently derivative and destructive to commercial prospects. Even critics like Evans and Reddy acknowledge that alternative licenses have stronger claims to being innovative, especially non-restrictive BSD-style licenses (384). Rather, I hope that governments fuel the coexistence and competition of both sectors through choice of sharing licenses. Non-restrictive licenses allow commercial use of o pen sourced code, indicating that the government may be able to structure its su pport to be beneficial to business

PAGE 110

105 interests. Specifically, where the government would be foolish to undertake development of all of its own systems at the expense of elimina ting entrepreneurship, some funding of highly skilled and motivated individuals to contrib ute to public code might work, in the analog of government subsidization of science. Dasg upta and David (1987, 1994) outline incentives similar to OSS contributors in a framewo rk of “science” and “technology”: In ‘Science’, peer recognition and the resulting re putation lead to complementary benefits, such as grants, positions i n academic organizations or highly compensated future position s in firms. The combination of these is the incentive. Scientific r ecognition is achieved by making one’s contribution public to peer review as quickly as possible and acquiring priority to the new knowledge. In ‘Techno logy’, the incentive structure is the traditional one: maximization of p rofit by securing property rights. By definition, latter rights keep the new knowledge private (paraphrased by Mustonen 2003, 103). Many of the motivations listed for Science resurfac e from earlier chapters as nonmonetary incentives to produce software--or, more g enerally, information goods--while profit motives drive proprietary firms. Tellingly, Science has a vantage as long term as any profit-seeking firm. Some of the biggest names in open source started a t educational institutions, and students continue to be a significant contingent of the OSS sector. Richard Stallman worked in the AI lab at MIT where he started the GN U project that birthed the Free Software Foundation, and Linus Torvalds was a stude nt at the University of Helsinki in Finland when he started Linux. One might even posit that the impetus to start the projects

PAGE 111

106 was only possible with the freedom to pursue what t hey liked without having to justify themselves to an employer/supervisor. To reiterate, part of the power of open source is that user demand is met through direct action rathe r than commercial incentives, meaning that different incentives can lead to new approache s or types of innovation. Governments can be the patrons to ensure that open source proje cts continue to receive the optimal amount of attention or resources, similar to the wa y grants and research positions currently help scientific advancement. Even Evans a nd Reddy (2003, 392-3) agree that a non-restrictive licensing scheme may be appropriate to government-sponsored OSS as it funds or subsidizes other research and development. Challenging the Open Source/Commercial Binary By creating a more distinct identity as a third pa rty to the traditional open source community and business binary, government may find a complementary existence that does not sacrifice commercial interests or innovati on. The least invasive policy would be completely hands off with information provisionment requiring unbiased assessment of individual open source products’ true characteristi c and non-propagandist advertising. To an extent, this already exists as government employ ees like teachers and librarians become more familiar with open source and introduce it to others, but few such explicit policies exist today (North Carolina initiative? ca nnot find source). Mandates may be useful in some instances, such as a collection of c itizens largely ignorant of the OSS alternative, but forced use is equally likely to re sult in some inefficiencies if the government does not know everything about the OSS. Finally, there is some room for direct development of OSS by government actors, but the appropriate license type is nonrestrictive so that the commercial software industr y can make use of such advancements.

PAGE 112

107 Conclusion The business world is in a state of flux, on the on e hand consolidating into fewer, larger entities and on the other encouraging “virtu al companies” that monitor and manage the decentralized nature of complex, modern product ion. Ownership centralizes even as production decentralizes, involving more people and specialties. The nature of capitalism will drive the system to collect resources into a s tock that can be efficiently allocated to new ideas, but at the tipping point we are witnessi ng an inversion of the incentive structure of the grants of intellectual property ri ghts. Open source and the Creative Commons operate in the disavowal of strict intellec tual property control. In these systems, contributors create for multifaceted reaso ns not entirely reducible to dollar signs, and in the process they learn from and teach one another in ways that simultaneously embrace cooperation and competition. Reputation and mutual gain counterbalance the strictly profit-driven urge to p roprietize information and make it secret, accounting for a more nuanced paradigm of i ntellectual property rights management that fits the interdependence of a moder n information society and its various motivations. Our current guiding principle is that people requir e some way to capture the gains of their ideas, resulting in the patent and copyrig ht system that grants singular rights. Yet, the individual grants are becoming increasingly con centrated so that individual grants routinely enter larger pools. Individuals stand to gain less from appropriating their individual accomplishments than the large companies with stockpiles of individual grants, and competition becomes more oligopolistic while the efficiencies of the originally devised system start to erode. Copyright extensions assumed that more rights

PAGE 113

108 to control intellectual property would stimulate pr oduction and cause a net gain for society. However, only a few monolithic properties actually benefited from new legislation, postponing the second wind that certai n works gain when entering the public domain. The current intellectual property regime is also proving overly restrictive for the highly collaborative and dynamic nature of modern i nformation goods production. Entire teams create movies and computer applications, and mash-ups are a popular medium for re-purposing original works with new meanings. Open source and the open content movements it inspired, such as the Creative Commons exist as a reevaluation of this structure of IP ownership; individuals release thei r works freely as a common resource to all others, a major decentralization of ownership r ights to keep up with the decentralization of production. The key difference of open source from the traditio nal intellectual property regime rests in its licensing terms, shaping the co mmunity it engenders and fostering further growth. Licensors retain enough control to ensure that their contributions to the public good are not appropriated by others in ways that leave them uncomfortable, providing a range of options that allow each to dec ide how much to contribute without coercion. In particular, the GNU GPL constitutes a social contract based on reciprocity, where one cannot diminish the commons because of th e nonrival nature of information goods and the general lack of ability to appropriat e from it directly. Some companies like Red Hat thrive on services built around it, but the code is ultimately free so that others may conceivably compete for the service industry of that sector. Redistribution automatically expands the commons, either from stre ngthening network effects by increasing users or from additions being released t o the commons. Basically a pay-as-

PAGE 114

109 you-go system, developing OSS relies upon more intr insic motivations than most other productive activities today. Altruism is never the key to producing public goods, so it is important that intrinsic and extrinsic rewards matc h or exceed the costs of contribution. Licensing does not address intrinsic motivations di rectly, but it does staunch the traditional losses of giving away work freely by cr eating a middle ground between copyright and the public domain. The public domain releases content into a wilderness without many rules limiting its use, but the types of copyleft or sharing licenses permit owners limited control over how they are willing fo r their works to be used by others: whether that is as-is or modified, commercially or non-commercially, etc. These choices create a range of options that are relatively restr ictive or non-restrictive, but in all cases the sharing licenses are less restrictive than defa ult copyright terms. Standardizing a few choices, such as the GNU General Public License (GP L) and the Creative Commons licenses, reduces transactions costs and allows for much more experimentation within the confines of the licensing terms. When licensees wan t to do something they are not permitted by the licenses, there is always the pros pect of negotiating with the licensors to obtain those rights. Sharing licenses shall never s upplant traditional copyright ones, but the former complements the latter nicely. Calibrating licensing terms also gives open source the greatest possibility of growth, according to each type of party: ideologica l contributors, individuals driven by need or desire of recognition, companies, governmen ts, etc. All parties can use the open source process, and all can manufacture such a prod uct, and the commons is stronger for having each involved. Ironically, hardliners like R ichard Stallman of the Free Software Foundation share a closer vision to business firms for restrictive terms that force

PAGE 115

110 additions to the commons to remain in the commons. Commercial firms benefit from gaining market share and forcing competitors to com pete against free for most users, and the ideologically inclined open source contributors grow a direct competitor to business interests. There are plenty of convincing arguments against government involvement with this form of open source because doing so potential ly harms business interests by replacing a functioning market system with less qua ntifiable and stable production. If government actively supports open source, it should do so under terms that permit commercial entities to also operate and thrive, i.e ., non-restrictive licensing terms. Where other firms do not want to support competition, gov ernments by and large should, so terms of using governmentally supported code should not preclude commercial applications. Recognizing motivational differences between parties emphasizes the nuanced approach and illustrates the illusory quali ty of any “clear cut” division of open source versus commercial. Despite a history of anta gonism with open source, for the first time ever a white paper by the Linux Foundation lis ts Microsoft as a top-20 contributor to the Linux kernel, responsible for 688 code changes, or 0.9% (Corbet et al. 2012, 1 and 11).15 Users often require only one program, with network effects that may favor a single supplier, but the introduction of an open so urce competitor does not fundamentally create an all-volunteers or all-firms environment. Just as profits are high enough to attract multiple competitors to a market although second be st is often far less profitable or altogether unprofitable, history demonstrates that an OSS alternative serves as a reliable 15 I have no indication whether this contribution was voluntary or not. The number of changes suggests so, but Microsoft has been discovered, not once but multiple times using GPL’ed code and subsequently releasing code under copyleft term s to avoid legal trouble.

PAGE 116

111 check to monopolistic control. This is even true wh en the alternative is offered by a nondominant commercial firm looking to gain market sha re by calling upon the OSS community instead of abandoning the market complete ly to the dominant firm. This substitute quality of open source to commercial pro ducts almost always takes some profit from the dominant firms, but in welfare terms consu mers should always benefit from the choice of a free program. What is perhaps most inte resting about the profit-loss described above is that it may represent specialization and n ew directions for capital rather than an overall price gouging or direct loss to GDP through loss of licensing revenue. As a product to an end user, the difference between open source put out by volunteer/non-profit or commercial efforts is large ly moot, anyways, but the processes and motivations can lead to different results. As u ser-innovators, volunteers and scholars have different insights into product development th an traditional research and development (R&D) at profit-driven firms, typically more directly related to functional needs than marketability. Open source is, itself, s imultaneously process and product : a mode of production that self-selects tasks by need and interest of participants as well as the information good of software itself. The power of open source is precisely that many different types of people can self-selectively impr ove the process and product, according to their strengths and motivations. Anyone with the means to contribute may do so, and presumably most of the best ideas thrive as all gai n from the ability to take what came before and reconstruct it into a new creative expre ssion if it may be improved. Two basic directions could correct for the technical bias of OSS products: either commercial companies could provide the friendly user interface s to make the software accessible to general users, or governments could invest in givin g its citizens the technical literacy to

PAGE 117

112 use them. Governments already have a large hand in education and human capital, and adding computer skills to a curriculum is a smaller step that could affect participation or utilization of OSS and thereby make it much more va luable. Because open source products are free, usefulness of the product is the continuous raison d’etre, but production in open source is simultaneously dependent on the r ecognition of skill and experience through process Governments, then, should be concerned with establi shing the best process that they can, which requires integrating the duality of OSS as product and process into the fabric of the society. Mirroring academic scholarsh ip, open source contributors exhibit many of the same intrinsic motivations and the repu tation rewards that substitute for standard monetary extrinsic motivators. Large resea rch universities support scientific research that helps commercial ventures, providing vision for what positive role public spending could have on OSS development. Government provisioning of software has too many command economics connotations to be useful, s o financially supporting some committed open source contributors like research pr ofessorships may be useful as a framework. Several top US non-profits related to op en source still pay respectable, CEOlevel salaries according to income tax forms submit ted to the IRS in 2010. Mitchell Baker and Brendan Eich of the Mozilla Foundation earned t he most in compensation just shy of $590,000, while Jim Zemlin of the Linux Foundation received a direct salary of $344,200 (Proffitt 2012). Public support of the non-profits constitutes 80-99% of revenue for about half of the organizations listed by Proffitt. With what can be gleaned from the IRS tax statements, paying big money for organizational sup port and business management oils the machine to be run efficiently.

PAGE 118

113 Market forces are not being obstructed, and many be nefits of public support of OSS may become apparent. For instance, increasingly complex information goods necessitate the incorporation of multiple copyright ed and patented materials into a single product, exponentially increasing transaction costs of negotiating the use of all relevant parts. The Patent Office continues to award unneces sary patents despite the fact that the monopoly grants are not organizing or particularly incentivizing any production as they were designed to do and instead permit patent troll s to threaten innovation. Open source volunteerism is ill-equipped to pay out-of-pocket t o apply for patents or to demonstrate prior art in lawsuits against patents, such as Linu s Torvalds’s involvement in the initial ruling against a Microsoft suit against Android OS, a Linux mobile distribution on smartphones (McMillan 2012). Recognizing these fail ings, however, can lead us to renovate the IP regime to empower creators and inno vators rather than speculators. Government has an enormous opportunity to reverse t he trend towards monopoly power by supporting non-restrictive open source, as outli ned above, to increase welfare. Because the commercial market for software, and tec hnology in general, is so strong in the United States and fulfills domestic n eeds so readily, the US government shows unsurprisingly less inclination to open sourc e than others around the globe. Technology exports are threatened by lower cost OSS alternatives, and software is more likely to be ported rather than adapted for differe nt audiences around the world. Emerging powerhouses like China are particularly in terested in using Linux to cut dependence on Windows, but even the European Union shows interest in using open source that it can participate in and influence dev elopment. Participation is beneficial because learning, reputation, and other gains canno t accrue to free riders, and influencing

PAGE 119

114 development can customize the product to its target audience to maximize usefulness. A powerful example of large-scale government change, the city of Munich, Germany switched wholesale from Windows to a custom Linux d istribution in 2006; by the end of 2011, the municipal government saved 4 million euro s, or a third of IT costs, after accounting for training costs and costs of migratio n and increasing IT jobs from 1500 to 9500 (Ude 2012). This has powerful implications for developing co untries for which American software can be relatively more expensive, ill-suited to the country’s specific needs, and with a perpetually dependent domestic so ftware sector. Developing countries especially benefit from strong institutional vision qualified personnel, and strong links to the user community for success that strongly point to appropriate government support (Camara and Fonseca 2007, 130). While the US alone benefits financially from postpo ning an open source revolution, the sector is growing undeniably and un stoppably. Creative destruction will claim some companies but not all, and too many vari ables are in play to accurately predict how open source efforts will evolve. Better to proactively engage the new communities sooner rather than later, instead of pr etending that the paradigm is not shifting, like a proverbial ostrich with its head i n the sand.

PAGE 120

115 Bibliography Benkler, Yochai. 2002. “Coase’s Penguin, or, Linux and “the Nature of the Firm”.” The Yale Law Journal 112, no. 3: 369-446. Bonaccorsi, Andrea, and Cristina Rossi. 2003. “Why Open Source Software Can Succeed.” Research Policy 32, no. 7: 1243-58. Bowles, Samuel, Richard Edwards, and Frank Roosevel t. 2005. Understanding Capitalism: Competition, Command, and Change 3e. New York: Oxford University Press. Camara, Gilberto and Frederico Fonseca. 2007. “Info rmation Policies and Open Source Software in Developing Countries.” Journal o f the American Society for Information Science and Technology 58, no. 1: 121-1 32. Comino, Stefano and Fabio M. Maneti. 2005. “Governm ent Policies Supporting Open Source Software for the Mass Market.” Review of Ind ustrial Organization 26: 217-240. DOI: 10.1007/s11151-004-7297-4. Corbet, Jonathan, Greg Kroah-Hartman, and Amanda Mc Pherson. March 2012. “Linux Kernel Development: How Fast it is Going, Wh o is Doing It, What They are Doing, and Who is Sponsoring It.” The Linux Fou ndation. Accessed April 9, 2012. x-2012 Bureau of Labor Statistics. “Computer and Mathemati cal Occupational Employment and Wages, May 2010.” Accessed February 1, 2012, oes150000.htm. Carroll, Michael. 2006. “Creative Commons and the N ew Intermediaries.” Michigan State Law Review 45: 45-65. Cornes, Richard, and Todd Sandler. 1986. The Theory of Externalities, Public Goods, and Club Goods New York City: Cambridge University Press. European Patent Organization. 2007. “EPO Revokes Am azon’s ‘Gift Ordering’ Patent after Opposition Hearing.” Accessed April 8, 2012. web/20090604043421/http:/ 07/20071207.html Evans, David S. and Bernard J. Reddy. 2003. “Govern ment Preferences for Promoting Open-Source Software: A Solution in Searc h of a Problem.” Michigan Telecommunications and Technology Law Review 9: 313 -394.

PAGE 121

116 Free Software Foundation. Updated September 20, 201 1. “(Formerly) Boycott Amazon!” Accessed April 8, 2012. Fitzgerald, Brian. 2006. “The Transformation of Ope n Source Software.” MIS Quarterly 30, no. 3: 587-598. Frey, Bruno S., and Reto Jegen. 2001. “Motivation C rowding Theory.” Journal of Economic Surveys 15, no. 5: 589-611. Geeknet. n.d. “About SourceForge.” Accessed Februar y 1, 2012. Gomulkiewicz, R.W. 1999. “How Copyleft Uses License Rights to Succeed in the Open Source Software Revolution and the Implication s for Article 2b.” Houstan Law Review 36: 179-94. Hartman, Peri, Jeffrey P Bezos, Shel Kaphan, and Jo el Spiegel. 1999. Method and system for placing a purchase order via a communica tions network. US Patent 5,960,411, filed September 12, 1997, and issued Sep tember 28, 1999. Heffan, Ira V. 1997. “Copyleft: Licensing Collabora tive Works in the Digital Age.” Stanford Law Review 49, no. 6: 1487-521. Heller, Michael A, and Rebecca S. Eisenberg. 1998. “Can Patents Deter Innovation? The Anticommons in Biomedical Research.” Science 28 0: 698-701. Hippel, Eric von, and Georg von Krogh. 2003. “Open Source Software and the “Private-Collective” Innovation Model: Issues for O rganization Science.” Organization Science 14, no. 2: 209-23. Koget, Bruce, and Anca Metiu. 2001. “Open-Source So ftware Development and Distributive Innovation.” Oxford Review of Economic Policy 17, no. 2: 248-64. Krugman, Paul and Robin Wells. 2005. microeconomics New York: Worth Publishers. Kumar, Sapna. 2006. “Enforcing the Gnu Gpl.” Journa l of Law, Technology, & Policy 1: 1-36. Lerner, Josh, and Jean Tirole. 2001. “The Open Sour ce Movement; Key Research Questions.” European Economic Review 45: 819-26.

PAGE 122

117 2002. “Some Simple Economics of Open Source.” The Journal of Industrial Economics 50, no. 2: 197-234. 2005a. “The Scope of Open Source Licensing.” Jour nal of Law, Economics, and Organization 21, no. 1: 20-56. 2005b. “The Economics of Technology Sharing: Open Source and Beyond.” The Journal of Economic Perspectives 19, no. 2: 99-120. Linux Online. n.d. “Linux Online Linus Torvalds A nnounces Linux.” Accessed January 28, 2012. linus_post.html. Loren, Lydia Pallas. 2007. “Building a Reliable Sem icommons of Creative Works: Enforcement of Creative Commons Licenses and Limite d Abandonment of Copyright.” George Mason Law Review 14: 271-328. McConnell, Campbell R., Stanley L. Brue, and Sean M Flynn. 2012. Microeconomics: Principles, Problems, and Policies 19e. New York: McGrawHill/Irwin. McMillon, Robert. 2012. “How Linus Torvalds Helped Bust a Microsoft Patent.” WIRED Magazine. e/2012/03/ms-patent/ McNutt, Paddy. 1996. “Economic Analysis of Clubs.” In The Economics of Public Choice 178-212. Brookfield: Edward Elgar Publishing Limi ted. Mockus, Audris, Roy T. Fielding, and James D. Herbs leb. 2002. “Two Case Studies of Open Source Software Development: Apache and Moz illa.” ACM Transactions on Software Engineering and Methodology 11, no. 3: 309-46. Moody, Glyn. Rebel Code: The Inside Story of Linux and the Open Source Revolution New York City: Perseus Publishing, 2001. Mustonen, Mikko. 2003. “Copyleft--the Economics of Linux and Other Open Source Software.” Information Economics and Policy 15, no. 1: 99-121. O’Hara, Sean, Gary Responte, Lauren Utigard, and Jo e Tate. 2005. Method for Controlling a handheld computer by entering command s onto a displayed feature of the handheld computer. US Patent 6,956,562, file d May 18, 2000, and issued October 18, 2005.

PAGE 123

118 O’Reilly, Tim. 2000. “My Conversation with Jeff Be zos.” O’Reilly. Accessed April 9, 2012. 000/bezos_0300.html Olson, Michael. 2008. “Chapter 5: Dual Licensing.” In Open Sources 2.0: The Continuing Evolution, ed. Chris DiBona, Danese Coop er, and Mark Stone. O’Reilly Media. Osterloh, Margit, and Sandra Rota. 2007. “Open Sour ce Software Development--Just Another Case of Collective Invention?” Research Pol icy 36: 157-71. Person, Carl E. n.d. “Differences between Federal a nd State Courts.” Accessed April 9, 2012. Public Library of Science. n.d. “PLoS | Leading a T ransformation in Research Communication.” Accessed February 1, 2012. http://w Proffitt, Brian. March 22, 2012. “Nonprofit open so urce organizations booming.” Accessed April 9, 2012. http://www.itw 260688/nonprofit-open-source-or ganizations-booming Raymond, Eric. 1999. “The Cathedral and the Bazaar. ” Knowledge, Technology, & Policy 12, no. 3: 23-49. Roberts, Jeff John. 2011. “Kindle Fire hit with Pat ent Suit.” Accessed April 8, 2012. /10/10/419-kindle-fire-hitwith-patent-suit/ Shah, Sonali K. 2006. “Motivation, Governance, and the Viability of Hybrid Forms in Open Source Software Development.” Management Scien ce 52, no. 7: 10001014. EISSN: 1526-5501|06|5207|1000. Ude, Christian. 2012. “Mnchen spart mit Limux Geld und hat weniger Strungen (Munich LiMux saves money and has less interference ).” Translated by Google Translate. e-muenchen-spart-mitlimuxgeld-und-hat-weniger-stoerungen-1203-90821.h tml. University of Michigan Library. n.d. “Orphan Works Project -FAQ.” Accessed February 1, 2012. orks/faq. Varian, Hal R. 2006. Intermediate Microeconomics: A Modern Approach 7e. New York: W.W. Norton & Company.