Copyright © The Authors 2007. Published by Oxford University Press.
Why develop open-source software? The role of non-pecuniary benefits, monetary rewards, and open-source licence type
* University of Southampton, e-mail: r.m.sauer{at}soton.ac.uk
 |
Abstract |
|---|
A review of the basic theory of optimal open-source software contributions points to three key factors affecting the decision to contribute to the open-source development process: non-pecuniary benefits, future expected monetary returns, and open-source licence type. This paper argues that existing large-scale software developer surveys are inadequate for measuring the relative importance of these three factors. Previous econometric studies that collect their own unique datasets also fall short because they generally measure the importance of only one supply factor in isolation. To fill the gap, I specify an estimable dynamic programming model of joint labour supply and open-source participation decisions that can provide empirical estimates of relative importance within a single unified framework of optimal decision-making.
Key Words: software • open source • labour supply • dynamic programming
I am grateful to Julian Morris, Eric Raymond, Corinne Sauer, and Margaret Stevens for providing insightful comments on previous drafts.
1 Governments may also prefer open-source software solutions for ‘non-pecuniary’ reasons. For example, open-source software can in some cases be more easily adapted to meet linguistic and cultural preferences.
2 The following example is adapted from Raymond (1999c).
3 There are also developers that receive wages from commercial firms for working on open-source projects. The reasons why commercial firms might want to pay developers to work on open-source projects are mentioned below.
4 Raymond argues that approximately 95 per cent of all software development activities are for intermediate goods in the production process, such as accounting packages.
5 Raymond is essentially applying ideas developed in Hayek (1945) to the case of open-source software production. That is, open-source developers have different ‘local’ knowledge that can be effectively tapped to the firm's benefit through the coordinating institution of open-source collaboration.
6 The BSD licence is more conducive to commercialization of the resulting code.
7 Firms operating in a low-transactions-costs environment may be able to mitigate free-rider costs of opening up the code by forming a consortium (see, for example, West and Gallagher, 2004).
8 Nearly identical percentages of students are found in the developer surveys analysed by Hertel et al. (2003) and Lakhani and Wolf (2005).
9 There are five ranks (levels of recognition) in ASF. They are: developer, committer, project management committee member, ASF member, and ASF board member. Promotion to a higher rank is awarded upon positive peer review.
10 Indeed, median lines of code within ASF rank, which is used as an instrument for rank in 2SLS versions of the regressions, is strongly correlated with ASF rank in first-stage regressions. Putting the likely endogeneity of the instrument aside, the first-stage results suggest that rank is another proxy for open-source experience.
11 A non-negligible number of open-source projects have multiple licences attached to them (see Lerner and Tirole, 2005a).
12 A non-zero choice probability for each of the three employment states can be generated with only two error terms. Therefore, it is not strictly necessary to specify the returns to unemployment to be stochastic.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. S. Siegel and M. Wright Intellectual property: the assessment Oxf. Rev. Econ. Policy, December 1, 2007; 23(4): 529 - 540. [Abstract] [Full Text] [PDF]
|
|