Don't get me wrong, wiki is a pretty cool idea. And in many ways, it's better than almost anything to which it might be reasonably compared.
But let's face it, wiki pretty much blows.
The problem? Scalability.
One of the other guys where I work, the chairman of a technical working group, decided to try using wiki as a collaboration tool as the group hammered out some revisions to a honking old document.
With only as few as three people working on the document -- some 100 different pages -- we had a problem with people overwriting each other's edits.
One part of the cause was a kind of social phenomenon: although there were around 100 pages in the wiki, at any given time only about 2 or 3 were actually of any interest to anyone. You know how it happens: one guy modifies a page; another guy checks to see what pages have been recently modified, notices that one, goes to check out what changed, and immediately wants to weigh in with his own changes. Get two guys doing that, and whammo! -- big headaches and discontent.
I'm now quite convinced that the wiki paradigm of editing whole web pages cannot reliably scale above exactly one author .
The real problem, of course, is locking. The lack thereof, that is.