The Open Source software industry is not well understood either theoretically or empirically. A broad and diversified research program is necessary in order to provide sound basis for public policy. The aim of these two pages is to present some of the lines of research that we feel are more important in economics. Our reflection on these topics has been enriched by the discussions that took place at the conference “Open Source: Economics, Law and Policy” held in Toulouse in June 2002 (see http://www.idei.asso.fr)
At the fundamental theoretical level, we need to understand better the properties of an industry based on voluntary contributions to a public good. The literature on this topic usually assumes that the agents contribute because they attach some value to the public good itself. However, in the case of Open Source Software, developers are presumably also concerned by the utility they derive from the fact of programming, by the increase in their reputation because of quality of their contributions, and by their own financial welfare. The consequences for the evolution of the industry need to be explored in more detail. Furthermore, little work has been done on competition between an Open Source software product and commercial products, and we need to understand better the dynamics of this interaction. Our colleague Gilles Saint-Paul has begun thinking about competition between Open Source Software and commercial software in a dynamic setting. This is a promising line of work which should be expanded both in the “endogenous growth theory” framework, in which Saint-Paul works, and in a modern game theory based industrial organization framework.
At a more applied level, the main priorities are to understand better both the determinants and the consequences of the choice of different licences; the consequences of the involvement of profit maximizing firms in the development process; the motivations of programmers who contribute to Open Source software. In all of these areas, the effort should be both theoretical and applied. We should stress that applied work in this area, as in other areas, can only be fruitful if it is based on solid econometric methodology. This requires considerable efforts in data collection, for which public funding is of great importance..
Nearly all the reflection on Open Source software has been oriented towards a few specific products. Although these products are important and deserve the attention they have received, this implies that there is little variety in the underlying data on which our theories are based. It would be useful to study other experiences. For instance, part of a thesis at the University of Toulouse will be devoted to a study of TeX as Open Source software; it would be useful if more such projects could be identified and studied.
Although this is a very difficult task, some effort should be made to compare the quality of Open Source and commercial software. This is a topic on which strong opinions are expressed, but for which there is very little data. For instance, Open Source advocates often stress the fact that Open Source software is, in some sense, more innovative than commercial software, in particular that it leads to more “user led innovation”. On the other hand, it has been argued that commercial software has historically been responsible for most of the truly innovative breakthroughs. These statements should be analyzed by using a more definitions of innovation, and measuring the relative performance of the two modes of production along these lines. Similarly, a more systematic analysis of the relative security of the two modes of software should be conducted.
Public policy towards Open Source software needs to be better understood. Economists who discuss these issues often rely on a implicit general principle that public policy should be neutral towards the mode of development. Although we believe this principle to be basically sound, it should be studied more thoroughly and refined. In particular, the costs and benefits of publishing the results of publicly funded research under the GPL licence need to be studied formally. A special effort should also be made to evaluate the claims that Open Source software is especially valuable for developing countries; this requires thinking not only about the total cost of using one or the other form of software, but also on the consequences for the dynamics of the local software industry.
Although progress is being made, relatively little is understood about this most fascinating phenomenon. As a consequence, any research program on Open Source software should use a bottom up approach; calls for proposals should not limit the topics that can be addressed. On the other hand, important public policy issues arise, and more focussed funding aimed at research that can provide sound footing for government intervention on these issues is certainly appropriate. Let us finish by stressing that both fundamental and more applied research should be based on the most rigorous and up to date methods.
 This note was prepared for the workshop on the Open Source Agenda, organized by the EC/IST programme and the US Natl. Science Foundation, Brussels, 14 October 2002.