J. and the Australian Gamer Podcasters have realized an important truth, namely that most user-made content is crap. Then again, most x is crap for all values of x. The question, I comment there, is whether you have tools to separate the wheat from the chaff (to jump metaphors). The goal is to set a million people loose, let it be 99% crap, and still get the work of 10,000 talented people (and remember that even talented people produce a lot of crap to get their good stuff).
Editing is hard. I will not even get into that here, except to note that many companies edit their own stuff too poorly to consider harvesting user-made content. If your internal content-production still gives mostly crap after filtering and editing, what hope do you have?