New Doctoral Thesis Explores the Effects of Software Patent Policy on the Motivation and Innovation of Free/Libre and Open Source Developers
Summary: The thesis of Marcus Manfred Dapp offers an explanation of why software patents are bad for Europe (and for any other continent/country for that matter)
This study offers a first empirical investigation into the effects of motivation and SWP presence on individual innovation behavior of FOSS developers. A new metric is proposed to measure individual innovation behavior based on code contribution types: in this scale, algorithm-based code contributions are rated more innovative than reuse-based contributions. In a separate analysis, the effect of motivation and SWP presence on reverse-engineering as a special contribution type is analyzed as well. Another new metric is proposed to measure SWP presence: instead of only considering the legal situation of a jurisdiction, the patent pressure within a software domain is also included. A survey was conducted to provide a new data-set for the empirical analysis.
Concerning the effects of motivation on innovation behavior, strong support can be reported for the following result: Above-average intrinsic motivation (joy and self-expression in code-writing) increases the odds for more innovative, algorithm-based code contributions, while above-average extrinsic (monetary and skills-related) motivation seems to decrease the odds. In connection with reuse-based contributions, the opposite relationship finds moderate support: Above-average extrinsic motivation increases the odds for reuse-based contributions, while above-average intrinsic motivation decreases the odds. The third result relates to reverse-engineering: None of the five motivational factors included in the analysis seem to explain why FOSS developers engage in reverse-engineering activities.
These results emphasize the role of motivation within the FOSS system. Particularly intrinsic motivation appears to not only keep this system alive and kicking, but more of it also seems to lead to more innovative contributions. Simply put: ‘Programming challenging new stuff is fun’. On the other side, it appears that reuse-based contributions with a lower innovation level – often needed for ‘the last mile’ before a program is end-user-ready – can be supported by offering extrinsic incentives. What still remains opaque from a theoretical point of view is the question why developers engage in reverse engineering. A broader analysis of motivational factors is needed here.
Concerning the effects of SWP presence on innovation behavior, the empirical results are less conclusive. Neither opponents nor proponents of SWP will find support for their positions that the presence of SWP decrease or increase respectively the odds for innovative, algorithm based contributions by FOSS developers. None of the three metrics used to capture SWP presence lends sufficient support to either side – be it positive or negative. Support, however, is found for a hypothesis related to reverse-engineering: stronger SWP presence attracts reverse-engineering based contributions by FOSS developers.
These results confirm several challenges for research as well as for policy-makers. Both continue to lack a broad, sound empirical foundation to discuss the effects of software patents on FOSS innovation.
For researchers, the challenges raised in this study are (a) to develop an easy-to use yet nontrivial metric to measure the presence of software patents empirically; (b) to quantify their effect on the FOSS system, helping policy-makers make better-informed decision. For future research, it would be useful to verify some of the links argued for in this study using other data sources. CVS logs have been used in the past for code contribution analysis. Maybe the innovation metric proposed here could be helpful in that regard.
For policy-makers in innovation and intellectual property policy fields the challenges are (a) to decide whether FOSS deserves a special case when debating software patents because of its unique way of producing software for the common good; (b) to continue treading carefully in the field of software patents before jumping to legislation. The FOSS market has reached a size where harm cannot be considered collateral damage as it may have in the past. Although the results have not shown systematic harm to the FOSS communities, there is still no empirical support that the traditional arguments in favor of patents do hold for the FOSS system – or software in general as some continue to argue.
Some limitations of the study deserve mentioning. First, taking the individual developer as unit of analysis ignores explanatory factors on project level that can also influence innovation behavior, such as project size and organizational structure. The larger a project is, the more elaborate its organization structure becomes, the more contributors tend to specialize in their contributions – up to a point where dedicated roles may emerge. Such a division of labor biases the measurement of individual innovation behavior. Second, it is impossible to investigate whether software patents caused projects to stop by only surveying ‘alive’ projects from SF as it has been done in this study. To obtain a complete picture, it is necessary to run a dedicated study on failed projects – even if the response rate will be very low.
No extraordinary claims are made, but it is clear that a correlation does seem to exist. If a legislator wishes to encourage software freedom and local production, then software patents would only act as a deterrent. They are detrimental. █