Open Source Software, a Brief Survey of the Economics of
The open source model is a form of software development in which the source code is made available, free of charge, to all interested parties; further users have the right to modify and extend the program. Open source software (OSS) methods rely on developers who reveal the source code under an open source licence. Under certain types of open source licence, any further development using the source code must also be publicly disclosed. In this brief survey, we will focus on several key aspects of open source software.
KeywordsDigital content Intrinsic motivation Licences Open source software R&D
The open source model is a form of software development in which the source code is made available, free of charge, to all interested parties; further users have the right to modify and extend the program. Open source software (OSS) methods rely on developers who reveal the source code under an open source licence. Under certain types of open source licence, any further development using the source code must also be publicly disclosed.
The open source model has become quite popular and is often referred to as a movement with an ideology and enthusiastic supporters – see for example Stallman (1999) and Raymond (2000). At the core of this process are two interesting phenomena: unpaid volunteers do a non-trivial portion of the development of open source programs and, unlike commercial software, open source software is not sold or licensed for a fee.
Having unpaid volunteers develop ‘free’ software is a puzzling phenomenon for economists. (Boldrin and Levine (2009) argue that from a historical perspective, the ‘open source’ model of development is the norm for many industries. In this entry, we will focus on the open source phenomenon in software. See the final section for extensions of open source methodology to other applications.) What are the incentives that drive contributors to invest time and effort in developing these open source programs, which are not sold or licensed for a fee? Intrinsic motivation may provide a partial explanation and suggests an analogy between academia and the open source movement. While publication plays an important role in academia, the analogy in the OSS world is being included in the ‘list of contributors’ of different projects. Being listed as a contributor may enhance the reputation of a programmer and can be instrumental in the job market. Additional incentives to develop open source software come from ‘self-use’ benefits and the enhancement of other (potentially proprietary) products in the market.
In this brief survey, we will focus on several key aspects of open source software. Much of the empirical work we review in this survey paper comes from high-quality data on open source software projects which are publicly available. Since most open source development takes place in the public domain (by which we mean publicly available ‘via the Internet’), data on many aspects of open source development are often available at various forges or platforms. These forges typically host many independent software projects. SourceForge, the largest forge, had more than 240,000 projects and 2.6 million registered users as of August 2010. Analysing the open source data available at SourceForge.net has already provided insight on worker motivation, the tradeoffs between intrinsic and monetary motivation, and the effect of the form of licensing on the contributions of developers (see Lerner and Tirole (2005b) and Fershtman and Gandal (2007), which are discussed below).
In the next section we examine motivation of programmers, while the following section examines the types of licensing employed in open source projects. The next two consider changes in the open source model, beginning with firm participation in the open source process and then reviewing some changes in the institutional structure of open source.
Open source development leads to very different incentives for R&D than the traditional proprietary development model – see Maurer and Scotchmer (2006) for a detailed analysis. Hence examining open source successes and failures may shed some light on the R&D process itself. We briefly examine this issue in the penultimate section. In the final section we briefly discuss the extensions of open source software model to digital content.
Finally, this is a short review; hence we focus on the topics we consider to be most important. Several books provide detailed reviews of open source software: see Dibona et al. (1999, 2006) and Lerner and Schankerman (2010). Excellent early survey articles include Lerner and Tirole (2005a) and von Krogh and von Hippel (2006).
Motivation of Programmers
Theoretical Research on the Motivation of Programmers
Early research on the open source phenomenon was primarily theoretical and focused on the motivation of unpaid programmers to work on open source projects. Several explanations regarding motivation have been offered in the literature: Lerner and Tirole (2002) argue that developers of open source programs acquire a reputation that is eventually rewarded in the job market, while Harhoff et al. (2003) argue that end users of open source benefit by sharing their innovations. Ghosh et al. (2002) argue that open source development is more like a hobby than a (paying) job. Johnson (2002) develops a model of open source software as voluntary provision of a public good – but for such a model one needs to assume that the primary motivation of developers is the ‘consumption’ or use of the final program. (Johnson (2006) presents a model in which the OSS organisation structure is superior to that of proprietary development as it minimises transaction costs and avoids agency problems.)
Empirical Research on the Motivation of Programmers
Using survey methods, Hars and Ou (2001) and Hertel et al. (2003) find that peer recognition and identification with the goals of the project are the main motivations for developers who contribute to open source software projects. In particular, Hars and Ou’s (2001) survey conducted among OSS programmers revealed that peer recognition was an important motivating factor for 43% of the respondents, while community identification was a key factor for 28% of the respondents. Similarly, Hertel et al.’s (2003) survey of 141 contributors to the Linux kernel project found that a prime motivating factor is ‘identification with Linux kernel’.
Hann et al. (2004) empirically examined the Apache HTTP Server Project and found that contributions were not correlated with higher wages, but that a higher ranking within the Apache Project was indeed positively correlated with higher wages. Using a Web-based survey, Lakhani and Wolf (2005) found that intrinsic motivations help induce developers to contribute to OSS. Chakravarty et al. (2007) found that the motivation of OSS programmers depends both on private motivations (like future monetary payoffs or ego) and social motivations (like altruism).
Licensing of Open Source Software
Like other products based on intellectual property, the intellectual property in software is typically ‘licensed’ for use, not sold outright. This is the case regardless of whether the software is proprietary or open source. Even though open source software is distributed freely without payment, the programs are distributed under licensing agreements. There are several different types of open source licence. The main difference is the degree of restrictions they entail.
Reciprocal (or viral) licences require that modifications to the program also be licensed under the same licence as the original work. Examples of reciprocal licences are the GNU General Public License (GPL) and the GNU Lesser General Public License (LGPL). The most popular open source licence is the GPL. If a software program is distributed under a GPL, the source code must be made available to users. Further, programs that incorporate code from a software project employing a GPL also must ensure that the source code is available. The GPL is, hence, a very restrictive licence and it is difficult to develop commercial products under a GPL licence (the LGPL is also quite restrictive, but less so than the GPL).
More permissive (non-viral) licences enable redistribution under a small set of rules. Under these licences, the software can be modified without making the new source code available publicly as long as the proper attribution is given. Examples of such licences include the Berkeley Software Development (BSD) license, the Apache License and the Mozilla Public License. Commercial products can be developed using software licensed under a BSD-type licence as long as credit for the underlying code is given to the copyright holder(s).
Many open source programs employ restrictive licences that would seem to hinder commercial development, since these licences require that all ‘future’ software using the relevant code must also be in the public domain.
Several papers in the literature have empirically examined the effect of different licences. Bonaccorsi and Rossi (2002) surveyed Italian firms that use open source software and found that, on average, firms that employ software with restrictive licences supply fewer proprietary products than firms that employ software with less restrictive licences.
The remaining papers we survey in this section come from the very detailed data that are publicly available at SourceForge. Project-level data include the ‘names’ of contributors, their role in the project, who contributed each part of the code, when the development took place, the stage of development, communications among project members, how bugs were fixed, how many times the project was downloaded the intended audience, type of licence, operating system etc.
Lerner and Tirole (2005b) examine the choice of licences using the database of open source projects from the SourceForge web site. They find that open source projects that run on commercial operating systems and projects that are designed for developers tend to use less restrictive licences, while projects that are targeted for end users tend to use more restrictive licences.
Fershtman and Gandal (2007) find that output per developer is much higher in OSS projects with less restrictive licences. This is striking, since the type of licence does not technically affect the writing of the code. This result is consistent with the hypothesis that the main motivation of programmers to contribute to restrictive OSS projects is to be included in the ‘list of contributors’: programmers have a strong motivation to contribute until the threshold level, and weak motivation to contribute above that level. Comino et al. (2007) find that the more restrictive the licence, the lower the probability that the project will reach an advanced development stage.
Changes in the Open Source Model: Firm Participation
Increased Firm Participation in Open Source Projects
The degree of reliance on unpaid programmers has changed over time. Today, more of the work on open source projects is done by contributors who work for firms. Using a sample of 100 open source projects hosted at Sourceforge.com, Lerner et al. (2006) find that the share of corporate contributors is higher in larger open source projects, where large means more lines of code.
Open Source and Proprietary Software in Same Market
Web browsers: according to W3Counter, in April 2011, Firefox had 29.5% of the web browser market. (The market data are from W3Counter; see http://www.w3counter.com/globalstats.php, accessed 19 May 2011.)
Web servers: Apache has been the dominant system in this market for many years. As of May 2011, Apache served approximately 63% of all websites (see https://en.wikipedia.org/wiki/Apache_HTTP_Server, accessed 19 May 2011).
Server operating system market: according to IDC (2008), as cited by Llanes and De Elejalde (2009), Linux had 13.7% of the server operating system market. (According to W3Counter, in the overall operating system market, Linux held a 1.41% share in April 2011; see http://www.w3counter.com/globalstats.php.)
According to Trefis (see https://www.trefis.com/company?article=12891#, accessed 31 January 2011), MySQL, a database management system, had approximately a 20% market share in database installations worldwide in 2010.
Recent theoretical work examines this phenomenon as well. See for example Casadesus-Masanell and Ghemawat (2006), Economides and Katsamakas (2006), Athey and Ellison (2009), and Llanes and De Elejalde (2009).
Towards Mixed-Source Strategies
A key change over time in the open source model is that many proprietary firms now initiate open source projects themselves, in addition to supplying programmers. Indeed, many proprietary firms now use a mixed source model, that is, a model in which some of their products are proprietary and are distributed under traditional licences, while others are open source and distributed under an open source licence. Such a mixed source strategy enables firms to benefit from the advantages of both open source and proprietary development. One key advantage to open source software development is that because the code is developed in the public domain, problems (bugs) can be found and solved quickly.
In a huge survey of more than 2300 companies in 15 countries, Lerner and Schankerman (2010) found that more than 25% of all firms surveyed develop both open source and proprietary software. Using data on 73 Finnish software companies, Koski (2005) empirically examined which factors affect whether the firm releases its product using an OSS or proprietary licences. She found that the more service oriented the firm is, the more likely it will be to offer products using OSS licences.
Institutional Changes in the Open Source Model
(This section draws heavily from Greenstein (2011) and comments and suggestions made by Greenstein.)
Rules for participation and governance in open source software projects have changed over time. Initially, open source projects were rather informal organisational processes. While some open source projects still allow unrestricted participation, many do not. In addition to rules regarding participation, open source projects typically have rules for deciding versions, and rules about reuse.
The institutional setting in which open source development takes place has also evolved over time. Sourceforge, which we discussed above, is clearly not the only setting in which open source development occurs. Indeed, Sourceforge is an ideal platform when an open source project lacks an institutional home. But, there are many important cases in which open source projects are hosted within an institutional setting. Linux operates within a consortium supported by many firms – and senior personnel receive salaries from the organisation. In other cases, firms sponsor open source projects – WebKit, which received financing from Apple, is an example (see West and O’Mahoney (2008) for work in this area).
Open source has also become a part of standard development by standard-setting organisations (SSOs.) The Internet Engineering Task Force (IETF) is essentially both an open source organisation and an SSO (Bradner, 1999).
Open Source Software and Incentives for R&D
(This section draws from Maurer and Scotchmer (2006).)
Incentives for engaging in R&D are quite different for open source software than in traditional proprietary software. Under the latter development method, products are often protected by patents and copyrights, which do not typically require disclosure of the source code. Hence intellectual property laws provide protection against imitation. Since open source software is typically put into the public domain, open source software would not provide innovation incentives when the goal is to prevent imitation.
However, as Bessen and Maskin (2006) note, imitation of a discovery can be desirable in a world of sequential/cumulative and complementary innovation because it helps the imitator develop further inventions. Since a non-trivial amount of software innovation is either sequential/cumulative or complementary (or both), this suggests that the open source development method may be socially preferable. Interestingly, Maurer and Scotchmer (2006) argue that open source development can also be privately preferable to traditional intellectual property protection when innovation is either sequential/cumulative or complementary. Open source development also has implications for the cost of R&D. Open source development can be thought of as ‘pooled’ R&D, which typically implies cost savings – see West and Gallagher (2006) Firms share code to test software, fix bugs and make improvements – see Rossi and Bonaccorsi (2005). Without open source, they would have to do this independently, which would imply duplicated costs.
Empirical research in this area is at a nascent stage. Using the data at SourceForge, Fershtman and Gandal (2011) find empirical support for the existence of knowledge spillovers among open source projects. The paper shows that the structure of the project network is associated with project success and that there is a positive association between project closeness centrality and project success. This suggests the existence of both direct and indirect project knowledge spillovers among open source software projects.
Open Source More Broadly Defined: Digital Content
(This section draws heavily from comments and suggestions made by Greenstein.)
Open source has spread well beyond the field of software development. Digital content is one area where open source has made major impacts. Creative Commons, which developed a way to help creators of content grant various degrees of copyright permissions to their work, is one of the most important outgrowths of the open source movement. Creative Commons licenses enable those who develop content to choose among a range of copyright protection, from ‘all rights reserved’ (full protection), via ‘some rights reserved’, to ‘no rights preserved’. Several key institutions use Creative Commons licenses. Wikipedia, the incredibly successful online encyclopedia, started with a variant of a GPL licence for text, and then adopted ‘Creative Commons’ methodology (The ‘wiki’ concept was developed in 1995 by a software engineer named Ward Cunningham. Wikis were developed in order to fix bugs in software development, but are now applied to many other applications – see Greenstein (2011). Wikipedia recently celebrated its tenth birthday. According to The Economist, it now has over 17 million articles (3.5 million in English). The content is created and edited by users. It was ranked as the Internet’s top research site in 2005, and consistently has been and continues to be one of the most popular websites. Currently it is used by a staggering 400 million users each month. (See ‘Wiki birthday to you – a celebration of an astonishing achievement, and a few worries’, The Economist, 13 January 2011.) Some YouTube and Flickr users share their content using Creative Commons licenses. The success of Wikipedia and other digital content providers using open source methodology shows that the open source model continues to evolve and will likely continue to be an important part of the digital economy.
We are especially grateful to Shane Greenstein for many comments and suggestions that significantly improved the manuscript. We are also grateful to Jacques Lawarree, and Nick Tsilas for very helpful comments and suggestions. An academic research grant from Microsoft is gratefully acknowledged. Any opinions expressed are those of the authors.
- Athey, S., and G. Ellison. 2009. Dynamics of open source movements. Mimeo.Google Scholar
- Bessen, J., and E. Maskin. 2006. Sequential innovation, patents, and innovation. http://econpapers.repec.org/paper/clanajeco/321307000000000021.htm
- Boldrin, M., and D.K. Levine. 2009. Market structure and property rights in open source. Washington University Journal of Law & Policy 30: 325.Google Scholar
- Bonaccorsi, A., and C. Rossi. 2002. Licensing schemes in the production and distribution of open source software: An empirical investigation. http://opensource.mit.edu/papers/bnaccorsirossilicense.pdf
- Bradner, S. 1999. The internet engineering task force. In Open sources: Voices from the revolution, ed. C. DiBona, S. Ockman, and M. Stone. Sebastopol: O’Reilly Media Inc.Google Scholar
- DiBona, C., S. Ockman, and M. Stone, eds. 1999. Open sources: Voices from the revolution. Sebastopol: O’Reilly Media Inc.Google Scholar
- DiBona, C., D. Cooper, and M. Stone. 2006. Open sources 2.0: The continuing revolution. Sebastopol: O’Reilly Media Inc.Google Scholar
- Ghosh, R., R. Glott, B. Krieger, and G. Robles. 2002. Free/libre and open source software: Survey and study, Part IV: Survey of developers. FLOSS Report, International Institute of Infonomics, University of Maastricht, The Netherlands. http://www.infonomics.nl/FLOSS/report/Final4.htm
- Greenstein, S. 2011. Innovative conduct in U.S. commercial computing and Internet markets. In Handbook on the economics of innovation, ed. B. Hall and N. Rosenberg, Forthcoming.Google Scholar
- Hann, I., J. Roberts, S. Slaughter, and R. Fielding. 2004. An empirical analysis of economic returns to open source participation. http://wwwrcf.usc.edu/hann/publications_files/economic_returns_to_open_source_participation.pdf
- Hars, A., and S. Ou. 2001. Working for free? Motivations for participating in open source projects. International Journal of Electronic Commerce 6: 25–39.Google Scholar
- Koski, H. 2005. OSS production and licensing strategies of software firms. Review of Economic Research on Copyright Issues 2: 111–125.Google Scholar
- Lakhani, K., and R. Wolf. 2005. Why hackers do what they do: Understanding motivation and efforts in free open source projects. In Perspectives on free and open source software, ed. J. Feller/Open, B. Fitzgerland, S. Hissam, and K. Lakhani. Cambridge, MA: MIT Press.Google Scholar
- Lerner, J., and J. Tirole. 2002. Some simple economics of open source. Journal of Industrial Economics 52: 197–234.Google Scholar
- Llanes, G., and R. De Elejalde. 2009. Industry equilibrium with open source and proprietary firms. Mimeo, http://www.hbs.edu/research/pdf/09-149.pdf
- Raymond, E. 2000. The cathedral and the bazaar. http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/
- Rossi, C., and A. Bonaccorsi. 2005. Intrinsic vs. extrinsic incentives in profit-oriented firms supplying open source products and services. First Monday, Issue 10, No. 5. http://pear.accc.uic.edu/htbin/cgiwrap/bin/ojs/index.php/fm/article/viewArticle/1242
- Stallman, R. 1999. The GNU operating system and the free software movement. In Open sources: Voices from the revolution, ed. C. DiBona, S. Ockman, and M. Stone. Sebastopol, CA: O’Reilly Media Inc.Google Scholar