One of my best strokes of luck in recent times has been the chance to work on Atlassian's wiki product, Confluence. It's a technology I'm intensely interested in, mostly because it represents a new way of producing content. Distributed collaboration is not only permitted by the wiki, it's pretty much the only way it will work.
Which brings me to the point. Wikis aren't a method to force people into collaborating. Forced collaboration isn't.
Quite a few people seem to have forgotten that, especially now that enterprise wikis are starting to take off. Andrew Roberts has rightly pointed out how the WYSIWYG editors in most wikis suck, and while I agree with his sentiment, his argument is that you need a WYSIWYG editor to force people into using the wiki:
Having an easy-to-use and powerful editor is very, very important in large scale publishing system in order to drive user adoption. … Pushing adoption of web publishing (WCM, blogs, wikis, etc.) into all corners of an organization is hard work.
Drive user adoption? That doesn't sound like collaborating. It sounds more like a pointy-haired boss foisting something on his unsuspecting employees.
This is not the way I see a wiki being implemented successfully. Rather than a top-down push from a central team to 'drive user adoption' of their latest fad, successful collaboration develops from the bottom up.
The best wikis are dead simple to get running, prove useful to the small bunch of initial users, and as they grow it's easy to involve others and work with large amounts of data. It's these features that let the software spread like a noxious weed once it has taken root. (Well, more like something good that spreads fast.)
Enterprise software has a habit of being underutilised and despised by those who are forced to use it, regardless of its quality. In contrast, collaboration can develop by itself if the management takes care of setting up the required infrastructure then getting out of the way.