Doctorat ParisTech THÈSE. l École nationale supérieure des mines de Paris. Tapping Into the Source : Corporate involvement in open source software - PDF

Description
École doctorale n O 396 : Economie, organisation, société Doctorat ParisTech THÈSE pour obtenir le grade de docteur délivré par l École nationale supérieure des mines de Paris Spécialité «Economie et finance»

Please download to get full document.

View again

of 140
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Arts & Culture

Publish on:

Views: 13 | Pages: 140

Extension: PDF | Download: 0

Share
Transcript
École doctorale n O 396 : Economie, organisation, société Doctorat ParisTech THÈSE pour obtenir le grade de docteur délivré par l École nationale supérieure des mines de Paris Spécialité «Economie et finance» présentée et soutenue publiquement par Jan EILHARD le 14 mai 2010 Tapping Into the Source : Corporate involvement in open source software Directeur de thèse : François LEVEQUE Co-encadrement de la thèse : Yann MENIERE Jury M. Marc BOURREAU, Professeur, SES, Paris Telecom - Paristech Rapporteur M. Eric BROUSSEAU, Professeur, Economix, Université de Paris X Examinateur M. Eric STROBL, Professeur, CECO, Ecole Polytechnique Examinateur M. Mikko VALIMAKI, Professeur, Helsinki University of Technology Rapporteur M. François LEVEQUE, Professeur, CERNA, Ecole des Mines - Paristech Directeur de thèse MINES ParisTech CERNA 60 boulevard Saint Michel PARIS cedex 06 L implication des entreprises aux logiciels libres Résumé : La participation des entreprises aux logiciels libres touche des domaines différents aux sciences économiques et sciences sociales. Elle est parmi d autres une expérience naturelle pour la production des biens publics, pour l innovation collective, pour les technologies disruptives, pour l externalisation des technologies ou pour les organisation décentralisée. Cette thèse se concentre sur la production des biens publics, l innovation collective et l externalisation des technologies. Dans notre analyse, nous utilisons une base de données de 10,000 logiciels libres trouvable sur SourceForge et lions l information des développeurs aux profiles académiques, salariés et bénévoles. Mots clés : Logiciel libre, innovation ouverte, économétrie Tapping Into the Source: Corporate involvement in open source software Abstract: Corporate participation in open source software touches upon a number of different issues in economics and social sciences. It is, among others, a natural experiment for the private provision of a public good, for gift giving, for collective invention, for disruptive technologies, for technology outsourcing or for decentralized organizations. Little does it surprise then that academia strives to understand corporate involvement in open source software. This thesis focuses on three aspects within this vast field: the private provision of a public good, collective invention as well as outsourcing of technology. For our analysis, we use a dataset of 10,000 open source applications from Source- Forge and relate information about developers to three categories as academic, corporate or private ones. Keywords: Open source software, open innovation, econometrics Acknowledgements I am grateful to Professor Lévêque for his support and guidance through this intellectual adventure. I would also like to thank Yann Ménière without whom this work would not have been possible. Working with the two of them has been a great pleasure. I thank Professor Strobl for his comments on my dissertation and earlier versions of my working papers. I am grateful to Professor Bourreau, Professor Brousseau and Professor Välimäki for their remarks on the thesis. I would like to thank Professor Glachant for organizing the PhD seminars and providing invaluable feedback on my papers. I thank all PhD students at CERNA for making work fun, in particular Antoine Dechezlepretre, Timothée Olivier, Henry Delcamp and Benjamin Bureau. I would especially like to thank Stéphanie Fen Chong and Yéhoudit Cohen for sharing the office and their friendship. I would like to thank Sésaria Ferreira for organizing my PhD defense and for helping me tackle all administrative hurdles. I thank the staff of the Berkman Center of Internet and Society for their hospitality. On a personal note, I thank my parents and my family for everything. Even if it sounds trite, it s true that I owe everything to them. Thank you. i ii La nécessité est, d ailleurs, de tous les maîtres, celui qu on écoute le plus et qui enseigne le mieux. Jules Verne (L île mystérieuse) iii iv Contents Preface xi Introduction 1 1 Corporate Open Source - A Literature Review Introduction Open Source as a Public Good Corporate Use of Open Source Software Open Source Business Models Open Source and Proprietary Software Contribution Corporate Strategies Protection of Intellectual Property Types of Corporate Contribution Conclusion A Look Inside the Forge: Developer productivity and spillovers in SourceForge projects Introduction Background Programmers Organization Firms involvement in open source projects Spillovers Econometric Model Data Data Source Descriptives Discussion Developer Productivities Scale effects Spillovers v 2.6 Conclusion Loose Contracts, Tight Control: Corporate contributions in Source- Forge projects Introduction Background Incomplete Contracts Data Quantitative Data Exploratory Survey Discussion Concluding Remarks Conclusion 88 A Source Code of Tracer Application 94 B Survey Questions 105 C Cross-section Regressions for Corporate Participation Rates 112 vi List of Figures 1.1 Importance of Interoperability Reasons to Contribute Corporate Policies towards Revealing Protective Measures Types of Corporate Contribution Projects per Topic Maximal Number of Developers per Projects Number of Releases Simulations Simulations Survey Question Survey Question Survey Question Marginal Effects Marginal Effects B.1 Survey Question vii viii List of Tables 1.1 Open Source Business Models Summary Statistics Development Stages Likelihood Ratio Test Computed Marginal Products Main Results Impact of Average Input Factors Results for Spillovers Results for Spillovers Summary Statistics (N=74,004) Predictions Regression on Corporate Participation Control Regressions on Corporate Participation C.1 Cross-section Results for Corporate Participation Rate ix x Preface Chapter 1 : Corporate Open Source: A Literature Review is an extract of the report Open Source Incorporated (January 2009) available on SSRN: Early drafts of Chapter 2 : A Look Inside the Forge: Developer productivity and spillovers in Sourceforge projects were presented at the Economics and Econometrics of Innovation seminar in Paris in December 2008, the MERIT-UNU seminar in Maastricht 2009, the International Industrial Organization Conference in Boston, 2009, the FLOSSworkshop 2009 in Padova, Drafts of Chapter 3 : Loose Contracts, Tight Control: Corporate contributions in SourceForge projects were presented at the International Industrial Organization Conference in Washington, 2008, the Berkman Center in Cambridge, MA, 2008 xi xii Introduction Open source software was born out of necessity. As the anecdote goes, Richard Stallman conceived the idea for free/open source software because he did not want to walk to a shared printer for nothing. When he wanted to add a small computer program to the printer that would notify him when a print was finished, he was barred access to the software of the printer. Frustated by the state of commercial, proprietary software, he set out to do things differently. Richard Stallman drafted the first version of the General Purpose License, the first open source license, in Its principles were simple. Anyone can modify and redistribute software licensed under the GPL as long as they credit the original developer and allow others to do the same. This premise turns software into a public good, non-excludable and non-rival. Necessity and intellectual curiosity are still the main drivers of open source development today. There are countless software projects that address a wide variety of needs, from printing utilities to computer games, from scientific tools to commercial applications. Thousands of volunteers from around the world contribute in these software projects, communicating problems and sharing ideas. Besides volunteers, one can observe more and more professional programmers who work with open source software. It appears that companies realize the power of tapping into the curiosity of the many. To state that open source software has become important to firms and expect astonishment in 2010 is moot. Even the most ardent critic acknowledges by now the commercial potential of open source software. Examples such as the close collaboration between Microsoft and Novell or Oracle s acquisition of Sun, whose main products are all open 1 Introduction source, are just the proverbial tip of the iceberg. Technology firms use and produce open source software with increasing ease and many to do so successfully. A recent study suggests that 85% of surveyed firms use open source software (Gartner, 2008). The same survey asserts that usage is evenly divided between mission critical and non-critical processes. Industry experts estimate that companies are heavily involve in its production (Ghosh et al., 2002; Lakhani and Wolf, 2005). The estimated figures range between 38% to 54% of developers who are paid for their work. In light of this situation, the question at hand is not anymore whether firms use open source software and contribute in its development. The predominant issues are rather how can they create profitable business models, why do firms contribute in open source software and what are the effects of corporate contributions on open source development. These are the questions we want to investigate in this thesis. Our endeavor is worthwhile for another reason than the commercial importance of open source software. Corporate participation in open source software touches upon a number of different issues in economics and social sciences. It is, among others, a natural experiment for the private provision of a public good, for gift giving, for collective invention, for disruptive technologies, for technology outsourcing or for decentralized organizations. Little does it surprise then that academia strives to understand corporate involvement in open source software. This thesis focuses on three aspects within this vast field: the private provision of a public good, collective invention as well as outsourcing of technology. How can firms provide a public good? This is a question that lies at the core of the economic analysis of public goods. Public goods are non-excludable and non-rival. This means on one hand that, once they are made available to someone, anyone has access to them. Nobody can be excluded. On the other hand, it also means that there is no scarcity of the product. The consumption of the good does not affect its availability to others. The nature of public goods render their private provision difficult. Economic theory predicts that firms and private individuals will generally underinvest in the production of the good. 2 Introduction We look at corporate open source software to understand the possibilities to provide a public good privately. Open source licenses create software that is, to some extent, non-exclusive. Anyone who uses the software can also modify and redistribute it. This renders applications with an open source license difficult to market. The firm needs to find alternative ways to create value for the consumer and sell their product or service. Reviewing the existing literature on corporate open source software, we point out common business models and possible forms of investment in open source software. What are the effects of corporate interaction with open source communities? This question addresses collective invention and more generally open innovation. Open innovation proposes that the firm benefits from opening up research and development (Chesbrough, 2003). Involving external actors in research can lead to economies of scale or scope and knowledge spillovers. The notion that the firm benefits from collaborating in research and sharing information is not new. Muntz (1909), writing about the iron ore industry, already noted that Each individual has some cherished bit of knowledge, some trade secret which he hoards carefully. Perhaps by sharing it with others, he might impart useful information; but by an open discussion and interchange he would, almost for certain, learn a dozen things in exchange for the one given away. Corporate involvement in open source software provides us with useful information on the performance of one form of open innovation. To shed light on the effectiveness of involving corporate with voluntary developers and of knowledge spillovers, we estimate the coefficients of a production function of the average open source project. This estimation gives us an indication on how well open innovation works in open source software. Why do firms contribute in open source software? This question is set into the broader context of technology outsourcing. We contend that corporate use of open source software is a kind of technology contracting. The firm signs the open source license in return for certain functionalities of the application. Contracting external suppliers can entail risks. 3 Introduction Transactions are costly to enforce, contracts cannot encompass all possible contingencies and specific investments might be made before transactions take place. These risks can deter the firm from entering into deals in the first place. To overcome these obstacles, the firm devises additional mechanisms to secure its transactions, e.g. vertical integration or long-term contracts. The open source license is an incomplete contract. It does not encompass any specifications, deadlines or specific features neither does it allow for any future contigencies. We argue that these potential deficiencies in the contract between the firm and open source project lead the firm to dedicate programmers to work on the software. Corporate open source contribution is therefore a form of vertical integration. For our empirical analyses, we use data on all open source projects which are hosted on the SourceForge website (Gao et al., 2007). SourceForge is an online repository that provides a platform for open source developers to manage their software projects and for users to find open source applications more easily. It is the largest of its kind with around 230,000 software projects and 2 million registered users (SourceForge, 2010). For comparison, Savannah, a similar repository hosts around 3,200 projects. This database is extended with background information on registered developers. We relate the developers addresses with top-level domain names, download the corresponding website and check its content for several keywords. 1 Counting the frequency of occurence for these keywords, we can rank each website according to three categories: Academic, corporate or private. This method allows us to have a proxy of corporate involvement in SourceForge projects. We discuss the downsides of this method in Chapter 2. This thesis is divided into three parts. Chapter 1 presents the literature review on corporate open source software. Chapter 2 focuses on the production of open source software and Chapter 3 deals with corporate contributions due to incomplete contracting. 1 See Appendix A for the Python source code. 4 Chapter 1 Corporate Open Source - A Literature Review 1.1 Introduction The improbable success of open source software stirs the interest of economists. For the last twenty years, open source software has gained more and more momentum in technology sectors. It is used for diverse applications such as servers, embedded systems or consumer products. Its success is improbable because open source software is provided voluntarily by a large number of private individuals, academics and, increasingly, by companies. What makes this phenomenon even more remarkable is the fact that open source software is a public good. The private provision is a centerpiece in the economic analysis of public goods ever since the seminal works of Olson (1971) and Coase (1974). Traditional economic theory contends that public goods create market failures. A market failure arises because the private benefit derived from the good does not coincide with the social benefit. Consequently, the private provision of a public good is riddled with free-riding, coordination issues and conflicts of interests. Despite these obstacles, open source applications have managed to enter and, in some cases, even to dominate former monopolistic markets. 5 Chapter 1. Corporate Open Source - A Literature Review In this sense, the research on the corporate provision of open source software is an extension of the public goods literature. It sheds light on a particular case of a public good: Software with an open source license. Researchers illustrate various aspects of its provision. They uncover the incentives for individuals to contribute in its provision. Other scholars examine the mode of open source production and community governance. A growing body of economic literature explores the reasons for corporate use and provision of open source and its implications for the software industry. Our objective in this chapter is to present a comprehensive survey of the economic literature on corporate open source software. Our approach is twofold. First, we look at why firms use open source software, in the course of which we shed light on the different open source business models. Then, we consider the reasons for corporate provision of open source software. Most of the papers in this field are empirical studies and there has been little theoretical work on corporate open source software. A notable exception is the literature on the strategic interactions between open source and proprietary software (Bitzer, 2004; Mustonen, 2005; Economides and Katsamakas, 2006). The chapter begins with a discussion of the properties that make open source software a public good. We then turn to the reasons for corporate use of open source software. Next, we look at the different business models that have arisen around open source software and then consider the interaction between proprietary and open source software. The second part of the chapter deals with corporate contribution in open source software. We talk about the reasons for firms to contribute. Finally, we end the chapter with a discussion of the different corporate strategies for contribution which include the adoption of business processes, the protection of intellectual property and the types of corporate contribution. 6 Chapter 1. Corporate Open Source - A Literature Review 1.2 Open Source as a Public Good Open source licenses are the building blocks for open source applications. They lay out the rules for modification and redistribution. The aim of open source licenses is ultimately to open up computer applications, encourage collective development and promote redistribution (Rosen, 2004). To study and improve the software, developers need to access an application s source code, the programming instructions in human-readable form. To this end, the source code needs to be open, hence the term open source. These licenses create a public good. Open source applications are non-rival and to some extent excludable. Because it is an information good, the open source application is by its very nature non-rival: One s own consumption does not diminish the availibility to others. A potential user of an open source application has to agree to the conditions of the open source license upon installing the software. Once he accepts the terms, he is a member of the community of this application, he benefits from the application, but at the same time has to abide by the rules of the license agreement. All such licenses delineate the way modifications to the software are treated and how one can redistribute the software. These restrictions make open source software in fact a specific kind of public good: A club good. The open source license acts as a gatekeeper who ensures that the rules of the open source community apply to every user of the software. In contrast to the textbook example, open source software does not suffer from congestion
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks