Dominated by editors: Wikipedia and traditional publishing

Tim O’Reilly recently posted his thoughts on a 2006 article by Aaron Swartz, Who Writes Wikipedia?. Swartz argues that Wikipedia founder Jimmy Wales’s contention that the encyclopedia is not “some emergent phenomenon—the wisdom of mobs, swarm intelligence, that sort of thing—thousands and thousands of individual users each adding a little bit of content and out of this emerges a coherent body of work” but rather “a dedicated group of a few hundred volunteers” is flawed. Swartz argues that Wikipedia needs to depend on that faceless mob, rather than the group of editors described by Wales for its content.

O’Reilly disagrees with Swartz, however, suggesting that the Wikipedia model—where numerous contributors supply the raw material which dedicated editors transform into a usable product—is quite similar to the traditional publishing model.

Take O’Reilly’s book publishing operations: we have far more outside authors than we have employees. Many of them are passionate experts rather than professional writers or editors, just like Wikipedia authors. Their work is improved by an editing team and brought to market in the context of brands that we’ve created, but we couldn’t do what we do without them. This is just as true of any publishing company. Did Bloomsbury’s editors invent Harry Potter? No, it was a welfare mom who dreamed up the idea while riding on the train.

I recently presented some research at the SCMLA that I believe offers some support to O’Reilly’s claim. I did a study of the revision histories of high- and low-quality Wikipedia articles, and what I found was that while the high-quality articles’ revision histories were very similar to those of high-quality articles in other contexts, the low-quality articles were quite different from those in offline contexts. (The “offline contexts” here are previously published studies of revision in academic writing.)

While low-quality offline writing in previous studies was characterized by excessive editing with very little content development, the low-quality Wikipedia writing in my study was characterized by very little editing of vast amounts of relevant content.

In other words, I found that while the process for creating good writing in Wikipedia looks very much like the process for creating good writing elsewhere, bad writing in Wikipedia is bad because it is not effectively edited. While my study was small and needs some further investigation, it seems like this result would support O’Reilly’s claim that the Wikipedia publishing model is quite similar to that of traditional publishing.

Update: In a reply to a comment on his post by John H, O’Reilly makes a point similar to mine:

Aaron’s point was that MOST of the articles are written by outsiders. They are then edited and improved by the insiders. The “long tail of articles”, as you put it, aren’t written by a different demographic, but they haven’t benefited from Wikipedia’s *editing* community.