Compare the Bill Clinton articles at Wikipedia and the new Congresspedia.

Congresspedia's is mediocre -- and will likely continue to be, because the site's allowed only logged-in users to edit pages. That might work once the site's past a certain level of quality, but it won't get there under this system becauase (and the circle is complete) nobody wants to register for a mediocre site. (Update, June 9th: I'm not so sure about this anymore: Clinton is not, after all, a congressman, so he's outside congresspedia's mission. The site might fill a useful niche. I still stand by the following paragraph, though.)

There needs to be more innovation in the wikipedia-clone area -- don't just start a parallel project: instead, code some greasemonkey, snatch a database mirror, and build a plugin that makes wikipedia itself more dynamic.