I've worked in varying degrees of depth within Knowledge Management, and what I've seen over the years is that decisions like this aren't based on any real-world data. Guys want to roll out giant systems with the assumption that "everyone is going to use it," without defining the problem they're trying to resolve, and without gathering requirements. Further, there is zero serious thought given to the maintenance nor curation of such a resource once it is implemented.
So, either a) lots of people use it, and it gets completely polluted by useless, outdated, and possibly confusing information (e.g. dump all of their crap in Box.net); or b) nobody uses it, because there was no reason to use it (e.g. Yammer is so cute, isn't it?).
Talk to somebody who works at a company that has treated Knowledge Management as an actual project and you'll get a completely different perspective on $social_platform/$knowledge_platform. Unfortunately, it's the exception rather than the norm.