Stewart, Ammeter and Maruping 2006: Impacts of License Choice and Organizational Sponsorship on User Interest and Development Activity in Open Source Software Projects
Stewart, K., Ammeter, T., and Maruping, L."Impacts of License Choice and Organizational Sponsorship on User Interest and Development Activity in Open Source Software Projects," Information Systems Research (17:2), 2006, pp. 126-144.
I read this paper for a theory-based perspective on the open source software (OSS) phenomenon.
- Rationale: The main motivation of this paper is an attempt to understand why "some OSS projects succeed and others fail." Among many studies of the motivations of OSS developers, two important factors explored here are the nature of the OSS license and the nature of (or absence of) organizational sponsorship for the project.
- Objectives: This study investigates the effect of restrictive (copyleft plus viral/share-alike provisions) vs. non-restrictive licenses, and the effect of no organizational sponsorship, commercial (market) sponsorship, and non-profit (non-market) sponsorship, on user interest and development activity.
- Key questions: What are the specific research questions addressed? What are the effects of restrictive vs. non-restrictive OSS licenses and no sponsor vs. commercial sponsor vs. non-profit sponsor on user interest (their proxy for usage–though nonetheless poorly operationalized, as I mention below) and development activity (also with operationalization problems).
- Theoretical background: Theoretical frameworks or perspectives (whether theory building or theory testing) used to accomplish the objectives. OSS success is measured in many different ways in the literature (e.g. project website traffic, code downloads, subscribers to news updates, developer activity, etc.). Considering OSS as a technology product, they draw from consumer behaviour and technology acceptance theories to theorize the effects of license choice and organizational sponsorship. In particular, they note that although OSS typically has zero purchase cost, other costs are relevant: "installation, maintenance and ongoing use." Quality of software is also a user concern in determing their product behaviour. For license choice, they draw from past theoretical distinctions on restrictive vs. non-restrictive licenses, as opposed to merely GPL or non-GPL. For the effects of organizational sponsorship, they partially draw from branding theory in marketing.
- Methodology: Adopting a positivist quantitative approach, they used ordinary least squares regression (the SPSS general linear model procedure) to test the hypotheses, involving linear, ordinal, and categorical data.
- Data sources if any: They obtained quantitative data from 138 projects listed on Freshmeat.net in the utilities, games, and software development categories, to get a broad range of different kinds of projects; they used the project information as listed, with sponsorship information obtained from project webpages. This study operationalizes success as user interest, measured by number of subscribers to e-mail updates. I think this is a very poor measure, since I would think it is mainly developers and at most IT administrators who would subscribe to such e-mails, not the bulk of end users. Another aspect of success is developer activity. Although they suggest measures such as frequency of bug fixes and responses to support requests, they forewent these and picked a third measure, the number of new releases produced. I consider this the poorest of the three suggested measures, since each project might have very different policies on what is eligible for a "release." In summary, I think the dependent variables are quite poorly operationalized, and thus the conclusions are questionable.
- Key findings: Because of the poor operationalizations of dependent variables, I question the validity of their findings. However, as reported, the key findings are: Nonrestrictive licenses attract greater user interest; sponsored projects attract greater user interest, and more so for nonmarket sponsors; nonmarket sponsors attract greater development activity; however, restrictive licenses did not attract greater development activity.
- Key contribution to knowledge: The theoretical background is quite valuable. Sponsorship of projects, especially by nonmarket sponsors, is very pertinent for generating user interest, especially for software with broad uses (mainly utilities in their study); this is more important than the nature of the OSS license. For development activity, nonmarket sponsorship is pertinent for encouraging developers to contribute. However, the study did not distinguish between paid and unpaid developers, which would probably have a very significant effect. However, as I mentioned, because the dependent variables are quite poorly operationalized, the conclusions overall are questionable.
- Key implications: For theory, studies of OSS need to consider the nature of organizational sponsorship and the nature of the license as influential factors for various outcomes. For practice, rather than automatically going with the GPL, project owners should carefully consider the implications of the restricteness and nonrestrictiveness of the chosen license. They should also try to obtain some organizational sponsorship, noting that nonmarket sponsorship could be more attractive to users and developers.