Here's my view: by submitting your article to wikipedia, you keep the copyright. Bomis does not own the copyright to any article (except maybe the ones written by Bomis employees), just like Linus Torvalds does not own the copyright to parts of the Linux kernel he did not write.
What you do if you hit the submit button on wikipedia is to release your material under GFDL, without invariant sections.
Bomis owns a copyright on the collection, just like RedHat owns a copyright on the whole Linux distribution they sell. So if somebody copies the website, simply changes all "Wikipedia" to "Yahoopedia" and publishes it, keeping individual articles under GFDL,, Bomis could sue; individual authors could not because none of their rights were violated. If Yahoo wants to publish a Yahoopedia, they would have to suck the individual GFDL articles individually, and then build a website out of it. Nobody could sue them for it, just like nobody can sue Suse for starting a new Linux distribution out of available GPL material. Of course, Yahoo, when publishing their Yahoopedia, would have to follow the prescriptions of the GFDL under which the articles were released; in particular, they have to release their modifications under GFDL and have to allow others to suck the articles from their website in easily modifiable form. If they don't, individual authors could sue them. Bomis could not sue, since none of their right were violated. Analogy: if somebody publishes a proprietary version of emacs, RMS could sue but Redhat could not.
There is however one issue: if I release my article to wikipedia under GFDL, Bomis, *per the GFDL, section 4B*, has to maintain information about at least five of the principal authors. I think the easiest way to do this would be to maintain unlimited page histories, maybe downloadable by FTP somewhere if the material gets to voluminous for the web server.
Axel
wikipedia-l@lists.wikimedia.org