Open Scholarship

On Motherboard, Daniel Oberhaus recently write “Nearly All of Wikipedia is Written by Just 1 Percent of Its Editors“. The post describes results from the book Structural Differentiation in Social Media by Sorin Matei and Brian Britt.

Oberhaus notes that Wikipedia has 43M articles in 285 languages and is the 5th most popular website in the world, and that “all of the articles are written and edited by an autonomous group of international volunteers”.

I suspect many (even most) people use Wikipedia but don’t really care how the content is created, just as many people use free & open source software (FOSS) but don’t care how it is created. I also suspect that many of the people who do care don’t really understand the people and processes involved. This is a problem – when we use information and tools, we should understand their origins.

FOSS, Wikipedia, and other “open” projects are often portrayed as the products of egalitarian, meritocratic communities. In my experience, the “FOSS community” shares some norms, but is really a wildly diverse set of project communities, each with their own norms, policies, and traditions. (The same is true of the “higher education community”.) Larger FOSS project communities (e.g. Apache, Gnome, Mozilla, Drupal, Moodle, … ) tend to be more explicit and thoughtful about such things, and smaller projects tend to be more, shall we say, idiosyncratic. (Not unlike small colleges…)

Matei and Britt used software to analyze 10 years of Wikipedia history, totaling 250 million edits. (Like most wikis and nearly all FOSS projects, Wikipedia keeps a complete edit history – the author, date, and content of every change.) Oberhaus reports their finding that “only about 1 percent of Wikipedia’s editors have generated 77 percent of the site’s content”. Furthermore, “roughly 40 percent of the top 1 percent of editors bow out about every five weeks”.

Is 1 percent too concentrated? I’m not sure. With 32M accounts and 130K active editors, “1 percent” is 1300 people. With 43M articles, that comes to 30K articles per active editor, but presumably many non-active editors monitor articles and edit as necessary. Demographically, Wikipedia editors have a range of ages and educational levels, but are predominately male. The number of edits by Wikipedians seems to follow Zipf’s Law – an inverse rank-frequency relationship. Note that some people use software to create 10,000 articles in a day.

Is 40% turnover every month or so too high? Again, I’m not sure. There is a core group of long-term Wikipedians who develop and maintain processes and policies, and a larger group who actively edit articles over a period of months or years. In these groups, high turnover could be a problem. Other people don’t set out to be long-term participants – they edit articles in their area of interest or expertise, or as part of an assignment or project.  An even larger group of people (like me) make occasional minor edits.

FOSS projects have similar processes, which have also been studied. For example, in The Emergence of Governance in an Open Source Community, O’Mahony and Ferraro described how concepts of leadership and authority developed in the Debian Linux community.

In Role Migration and Advancement Processes in OSSD Projects: A Comparative Case Study, Jensen and Scacchi studied 3 large FOSS projects (Apache, Mozilla, and NetBeans) and found similar structures of project participants. In most FOSS projects, the largest group are passive users and observers who make few if any contributions to the project. Some of them become active users who ask and answer questions on discussion forums, request new features, or report defects. Some of the active users become developers who contribute code, documentation, or other content to the project. Some developers become managers who lead specific parts of the project and/or community, and some of them become core developers or project leaders.  Almost all FOSS contributors follow this progression, and only a tiny fraction become core developers or leaders.

This progression matches my experience in a variety of FOSS projects. I have passively used or observed many projects, made minor contributions to some of them (e.g. Moodle, MoinMoin, RedMine, Trac, and TikiWiki), and made larger contributions to a few (e.g. Drupal and TWiki).

In Magnet or Sticky – An OSS Project-by-Project Typology, Yamashita, McIntosh, Kamei, and Ubayashi analyze the migration of participants in FOSS projects and identify 4 project types – attractive, stagnant, fluctuating, and terminal. In Software Population Pyramids: The Current and the Future of OSS Development Communities, Onoue, Kata, and Matsumoto match these types to distinction patterns in population pyramids, which show the longevity of participants in project roles.

It might be interesting to apply some of these techniques to other forms of open production and explore the similarities and differences.

Laws, like sausages, cease to inspire respect in proportion as we know how they are made. — John Godfrey Saxe

Perhaps the same is true of FOSS, Wikipedia, and other open processes.