THE OPEN SOURCE SOFTWARE DEVELOPMENT PHENOMENON: AN ANALYSIS BASED ON SOCIAL NETWORK THEORY Greg Madey Vincent Freeh Computer Science & Engineering Computer Science & Engineering University of Notre Dame University of Notre Dame [email protected] [email protected] Renee Tynan Department of Management University of Notre Dame [email protected] Abstract The OSS movement is a phenomenon that challenges many traditional theories in economics, software engineering, business strategy, and IT management. Thousands of software programmers are spending tremendous amounts of time and effort writing and debugging software, most often with no direct monetary compensation. The programs, some of which are extremely large and complex, are written without the benefit of traditional project management, change tracking, or error checking techniques. Since the programmers are working outside of a traditional organizational reward structure, accountability is an issue as well. A significant portion of internet e-commerce runs on OSS, and thus many firms have little choice but to trust mission-critical e-commerce systems to run on such software, requiring IT management to deal with new types of socio-technical problems. A better understanding of how the OSS community functions may help IT planners make more informed decisions and develop more effective strategies for using OSS software. We hypothesize that open source software development can be modeled as self-organizing, collaboration, social networks. We analyze structural data on over 39,000 open source projects hosted at SourceForge.net involving over 33,000 developers. We define two software developers to be connected — part of a collaboration social network — if they are members of the same project, or are connected by a chain of connected developers. Project sizes, developer project participation, and clusters of connected developers are analyzed. We find evidence to support our hypothesis, primarily in the presence of power-law relationships on project sizes (number of developers per project), project membership (number of projects joined by a developer), and cluster sizes. Potential implications for IT researchers, IT managers, and governmental policy makers are discussed. Keywords: Open source software, social networks Introduction The global E-Commerce infrastructure relies heavily on open source software (OSS). Examples include web-servers (Apache, iPlanet/Netscape), e-mail servers (Sendmail), languages (Perl, Java, Python, GCC, Tk/TCL), and operating systems (Linux, BSD Unix). Corporate IT managers and decision makers have traditionally depended on either internally developed systems or commercially purchased systems that have been mostly closed source and proprietary. Open source software is written and supported by programmers, many coming from the “hacker culture”. This development culture includes hundreds of thousands of distributed programmers voluntarily producing, sharing, and supporting their software with no monetary compensation for their efforts. This presents those corporate IT managers and decision makers with a new development process and culture to learn and 1806 2002 — Eighth Americas Conference on Information Systems Madey et al./ Open Source Software Development Phenomenon interact with. Although the open source development model has been used for over 25 years, its significance to mainstream business IT grew with the Internet and the emergence of E-Commerce. These OSS developers collaborate from around the world, self-organize, and rarely meet face-to-face. Questions about their motivations, about the quality and ethics of this OSS phenomenon, new intellectual property legal questions, and whether there is a role for new government policy in this area can be asked. For example, should governments widely adopt open source software, should governments take an active role in promoting standards with the OSS process, and what role does OSS have in promoting or diminishing innovation? A better understanding of the process of open source development is needed. In this paper, we report the results of an empirical investigation of the social and collaborative networks present in the open source movement. Open Source Software Development Open source software is by definition software for which users have access to the source code. This distinguishes it from the recent common practice by commercial software publishers of only releasing the binary executable versions of the software. Most open source software is also distributed at no cost with limited restrictions on how it can be used; hence the term “free” when used to describe open source carries two meanings: 1) free of cost and 2) free to do with the software as you wish (i.e., most importantly — free to read the code). Case studies documenting the open source software development model, albeit often sympathetic to that model, point to potential lessons and benefits that may be of value to corporate IT (O'Reilly, 1999; Wu, 2001). It is claimed that open source development produces more bug-free code, faster, than closed proprietary developed code, although this has yet to be conclusively demonstrated (Jorgensen, 2001; Koch, 2002; O'Reilly, 1999; Raymond, 1999; Sharma, 2002; Stamelos, 2002; Wang, 2001; Wu, 2001). Open source software development teams, are generally comprised of volunteers working not for monetary return, but for the enjoyment and pride of being part of a successful virtual software development project. Team members often come from around the world and rarely meet one another face-to-face. The open source projects are self-organized, employ extremely rapid code evolution, massive peer code review, and rapid releases of prototype code. Many of these practices are counter intuitive and the opposite of what conventional software engineering holds as the correct processes for the production of high quality code (Bollinger, 1999; Charles, 1998; Edwards, 1998; Fielding, 1999; Hecker, 1999; Lawrence, 1998; O'Reilly, 1999; Ousterhout, 1999; Payne, 2002; Raymond, 1999; Sanders, 1998; Torvalds, 1999) The Open Source Software movement is a prototypical example of a decentralized self-organizing process. There is no central control or central planning. It challenges conventional economic assumptions, it turns conventional software engineering and project management principles inside out, it threatens traditional proprietary software business strategies, and it presents new legal and government policy questions regarding software licensing and intellectual property. Moreover, OSS is a major component of the IT infrastructure enabling global e-Commerce. Open source software including BIND, sendmail, Apache, Linux, INN, GNU utilities, MySQL, PostgreSQL, and Perl are critical components of the Internet. They enable major services hosted on the Internet, e.g., e-mail, WWW, e-Commerce, domain name lookup. The Netcraft.com survey of 36.6 million web servers worldwide reports an over 60% market share for the open-source web-server Apache (Netcraft, 2002). Open source software is free and open; this generates questions about its impact on domestic digital divides, the third-world digital-disadvantaged, the OSS influence on innovation, and national and corporate security from cyber-terrorism. For example, while free (i.e., no cost) software may help reduce the digital divide, will it do so by weakening the software industry’s business- model? Does OSS promote innovation, or does it ride on the coattails of closed source software development? Several important studies on the OSS phenomenon have been conducted and provide important foundations for this research. Feller and Fitzgerald (Feller, 2000) developed a research framework and analyzed the OSS phenomenon. Hars and Ou (Hars, 2001) surveyed OSS developers and reported on their motivations for participation in OSS projects. Scacchi (Scacchi, 2002) has an extensive ongoing study of the socio-technical processes associated with OSS development work practices. Wolf, et al recently released the results of a survey of 526 OSS developers from SourceForge, and 134 participants on the Linux kernel mailing list, reporting on developer motivations and attitudes (Wolf, 2002). Social and Collaborative Network Theory Several research streams converge to provide us with a number of tools and models for analyzing the open source software movement: social network theory, small world phenomenon, power-laws, self-organization, and graph theory. Social network 2002 — Eighth Americas Conference on Information Systems 1807 Social Theory in Information Systems Research theory models persons as nodes of a graph and their relationships as edges of the graph (Jin, 2001; Wasserman, 1999; Watts, 1999; Watts, 1998). Thus two persons are directly connected if they have a relationship (e.g., friendship) with each other; they then are one link away from one another. More distant relationships are modeled as paths through the graph; a “friend of a friend” is two links away. Several studies reveal an interesting phenomenon present in many of these social networks; most persons are very few links from any other person – the Small World Phenomenon (Watts, 1999; Watts, 1998). This idea was popularized in the play (and movie) Six Degrees of Separation (Guare, 1990) which claims that all persons in the world are at most six friendship links away. Collaborative networks are variations of social networks, where the relationships are collaborations, e.g., actors in movies (Tjaden, 1996; Watts, 1999), or co-authors on research