academia agriculture art books cities commons strategies conferences copyright law digital commons economics education enclosure enclosures environment finance food free culture free software Germany government Great Britain history India international Internet land law localism market culture nature ontology open source software peer production politics videos water
Why Do Some Software Commons Succeed and Others Fail?
Wed, 07/25/2012 - 10:16
The tech world frequently talks about open source software as a collaborative endeavor, but it is less apt to use the word “commons,” let alone engage in rigorous empirical analysis for understanding how software commons actually work. The arrival of Internet Success: A Study of Open-Source Software Commons (MIT Press) is therefore a welcome event. This book is the first large-scale empirical study to look at the social, technical and institutional aspects of free, libre and open source software (often known as “FLOSS”). It uses extensive firsthand survey research, statistical analysis and commons frameworks for studying this under-theorized realm.
While most people may associate open source software with Linux, there are in fact tens of thousands of open source projects in existence. Many consist of no more than two or three participants, and may have only an irregular existence. However, many thousands of others attract a small but spirited team, and still others are huge, robust social ecosystems in their own right.
The authors of Internet Success, UMass Professor Charles M. Schweik and consultant Robert C. English, looked at the large universe of FLOSS projects hosted on SourceForge.net, a website that functions as a kind of clearinghouse for over 260,000 FLOSS projects (as of February 2011) and 2.7 registered software developers. The site provides most of the tools that developers need to find colleagues and build a new FLOSS program – a Web repository of code, bug-tracking utilities, online forums, email mailing lists, a wiki, file downloading services, etc.
While SourceForge is not the only such site for FLOSS projects, it is the largest and arguably representative of the universe of such projects. With support from the National Science Foundation, Schweik and English set out to study the pool of software development projects on SourceForge to try to determine why some succeed, why others fail and why others simply languish. They explain in excruciating technical, social science detail how they assembled and analyzed their datasets, which originate in a vast collection of SourceForge data on more than 130,000 projects as well as their own survey questionnaire of programmers.
Schweik and English conclude that FLOSS commons will be more successful when:
- The project has a relatively clearly defined vision and a mechanism to communicate the vision early in the project’s life, before getting to the growth stage;
- The software produced has a higher utility as perceived by the end user community;
- The project has one or more leaders who generate excitement and demonstrate “through doing.”
- The project has well-articulated and clear goals established by the project leader(s).
Schweik and English found no support for the hypotheses that projects succeed because they add functionality to core FLOSS technologies; because they are components of a computer operating system or other foundational infrastructure; or because they are built on certain programming languages that are popular in developer circles.
What’s particularly impressive about Internet Success is the painstaking rigor of the data collection and its use of sophisticated statistical and social science methodologies. This book is no anecdotal, conjectural account of why and how software commons function. It bores deeply into the social phenomena of software commons by studying a large, representative sample of projects. (Full disclosure: Schweik is a friend and I provided a blurb for the book jacket.)
Schweik and English assess some 41 possible variables that make for a successful software commons. Does payment to participants help or hurt? Do programmers participate because they have more leisure time, or do they have other motivations? How important is it to have highly skilled programmers on a team? Does a successful team need to have a mix of different people, or “sociocultural heterogeneity”? Are regular face-to-face meetings important? How important is it that there be levels of trust among team members? Do General Public License (GPL)-compatible projects outperform ones that are not compatible with the GPL?
The authors even go so far as to make nuanced judgments about how these factors might be relevant in different ways at different stages of a project’s evolution. It turns out that early-stage projects benefit from having a “benevolent dictator” style of leadership whereas in the growth stage, it is precisely the opposite, leadership and executive rights need to be shared among key participants.
One reason that this book is so interesting is that Charles Schweik once studied with Elinor Ostrom at Indiana University. He is therefore fully conversant with the Institutional Analysis and Development (IAD) framework and other standard tools for assessing commons. Schweik and English are thus able to situate their analysis of software commons within the larger literature studying the management of common-pool resources.
The authors make the interesting observation, for example, that software commons differ significantly from natural resource commons in how authority is structured. In successful and sustainable natural resource commons, participants tend to have a say in the direction or management of the commons. In FLOSS settings, however, “[I]t appears that more of the decision making is placed in people of accepted authority following meritocracy principles. Other participating developers may not have as much of a say, at least when it comes to major project decisions.”
Internet Success sets a new benchmark for the study of software as a commons. First, it validates such collaborative activity as a commons, which itself is a major advance. Second, it applies some of the most rigorous tools imaginable to data-collection and analysis. And third, it gives us a more solid footing – beyond hacker folklore – for understanding how to help software commons succeed. These lessons, I suspect, will hold many insights into other sorts of digital commons as well.