Photo: Flickr user lifeontheedge

Saturday, September 16, 2006

Sanger Spearheads Wikipedia Fork

How's that for a headline? (Larry Sanger co-founded wikipedia then left it, disatisfied with its direction. It's good to see him working on something other than the silly Digital Universe project.)

Basically, the Citizendium looks really exciting -- a mirror-world wikipedia, run from the bottom up by regular people but guided by experts who are given authority over articles in their field (and who'll have to give their real name and CV).

Forks have been tried before, of course (Wikiinfo comes to mind (and that hilarious neo-nazi attempt)). But Sanger has plenty of ideas for attracting academics who aren't inclined to march through wikipedia clearing brush:

This will be "expertpedia," won't it? Experts only, right?

Not at all! As with Wikipedia, ordinary people will form the backbone of the Citizendium.

But experts will be involved and made into editors. Aren't you trying to turn the successful "bazaar"-style Wikipedia model into a failed "cathedral" style of project?

Again, no. Experts will be expected to work shoulder-to-shoulder with ordinary people in this project in more or less the same bottom-up fashion that Wikipedia uses. The difference is that, when content disputes arise, whatever editors are paying attention to the article will be empowered to articulate a resolution--if the article falls in their area of specialization. Furthermore, their decisions will be enforceable.

And the forking process looks well-thought-out:
A "progressive fork" works like this: we will begin with all of Wikipedia's articles, so that the Citizendium will begin as, simply, a mirror of Wikipedia. Then people start making changes to articles in the Citizendium. On a very regular basis, we will refresh our copies of Wikipedia articles. If an entry in the Citizendium has never changed since being copied from Wikipedia, but the Wikipedia version has, then we upload the most recent Wikipedia article. But if the Citizendium has changed an article, then it is not refreshed. Tools will no doubt be written that will allow users to compare the differences between the Wikipedia article and the Citizendium article side-by-side. In addition, of course, people will be able to start brand new articles on topics Wikipedia has not yet covered.

Don't get me wrong: it probably won't work. But if you look at the Citizendium as part of a series of wikipedia-style projects, it's promising -- it's the first one built on a real understanding of what makes wikipedia tick.
A slashdotter:
Citizendium's design does seem to address what I consider the main problems with Wikipedia: disorganized, low-quality edits by well-intentioned people. The design of Wikipedia basically wastes huge amounts of time. Most articles gradually rise to a certain level of quality, and then the pioneers lose interest in the topic because there's not much left to be done. After that, the article gradually decays in quality. You'll get hundreds of edits on an article, but the diff between the beginning and the ending version can be zero. The current system basically requires serious editors to have huge watch-lists, and check them vigilantly to keep entropy from having its way. That's no fun, and it's the reason why, after several years of heavy participation, I gave up on WP.

I'll be happy if the Citizendium's just big enough to incubate new wiki models, find new social dynamics that work, serve as a kick in the pants for Wikipedia. Sanger seems to have a keen sense of how the website interface modulates the social interaction (WP, on the other hand, hasn't even fixed the location of its "Edit This Section" links). There are tremendous design challenges here, but maybe he's equal to them.


No comments: