Articles About Free Books

The Cathedral, Not the Bazaar

The bazaar model seems to have been almost a complete failure in the world of free books, although not for want of trying. Tellingly, The Cathedral and the Bazaar was itself written cathedral-style by Raymond. He has also started a bazaar-style book project, The Art of Unix Programming^[7], which appears to be languishing. I can only think of one high-quality, finished bazaar-model free book, a textbook in which each author wrote one chapter.^[4]

The failure of the bazaar model with free books might not seem surprising, since to most people it sounds like the silly party game where each person takes a turn adding more onto a story. We normally assume that an author has a unique voice, and that authorship can't be delegated. However, quite a high percentage of the world's free books seem to be software documentation, which in principle should be amenable to a decentralized approach. The GFDL copyleft license for books is clearly aimed at such projects, and requires, for example, that the book be maintained in a form that can be edited with free software, so that it will never become a pig in a poke if the original author loses interest or goes incommunicado. Group authorship, however, just doesn't seem to have caught on, even in software documentation. Maybe the explanation is that in software projects, the number of programmers interested in writing documentation averages to less than one. However, that wouldn't explain the failure of the Nupedia open-content encyclopedia project, which would seem to have been ideally suited for group authorship. My own experience attempting to contribute an article to Nupedia suggests a simpler theory: people make free information because it's fun, and group authorship is not always fun. After I wrote the original version of this article, several people pointed out the existence of Wikipedia, which was meant to be a minor ``fun'' project alongside the Nupedia online encyclopedia. Unlike Nupedia, Wikipedia has taken off, and has a lot of people contributing articles. This suggests that bazaar-style book projects can work, but whether they'll work may depend a lot on how they're organized, and on whether they're fun to participate in.

Note added December 2002: I dipped my toes in the water as a participant in Wikipedia, and had some fun at first working on some of the physics articles. However, it seems to me that Wikipedia has a problem with certain articles getting hijacked by cranks. If Nupedia was too strict and regimented, Wikipedia seems to have the opposite problem: a lack of serious quality control. Specifically, take a look at the articles on certain topics related to astrology, e.g., Horoscope, Zodiac, and Walter Mercado. If you click on "older versions," you can see the history of how these articles were edited. In my opinion, the true believers in astrology have continually rebuffed attempts by various people (including me) to make these articles neutral in tone, rather than credulous and one-sided. Skeptics add some text to try to balance the treatment a little, and then the true believers delete it again. To me, this is the antithesis of what ``free'' is about: silencing any voice with which one disagrees. In my opinion, the Nupedia and Wikipedia projects were both interesting social and technological experiments, but neither model can be successful in the long term without some big changes. Right now, I'd be reluctant to look anything up on Wikipedia and trust that I was getting the whole story. The good news, however, is that even if (as I hope) Wikipedia's social and technological structure mutates and improves in the future, that doesn't mean all the old text is lost. It's still free information, and it can be incorporated into a new and improved version of the encyclopedia.