There's an inherent exploit/explore tradeoff in "social". Or, more to the point, are you going to document social relationships or improve them (in spite of the fact that there are powerful people who don't want improvements to occur)? There's a lot more money and less risk in the former, but I'm more a fan of the latter, because that's where the great companies come from. Which is why I like Meetup a lot (with its unusual focus on growing the social graph) and feel more tepid about LinkedIn. They're a great company and they've executed well, but their model is still stuck in the old, broken, credential-and-resume way of doing things.
Since my colorful Google history has (against my will) been exposed to public digestion, I'll admit that what started that fracas was a document-vs.-improve debate regarding Google+ that I, unwisely and half-unintentionally, got myself caught in. (I was on the "improve" side.)
What on earth does this have to do with the Ivy League and college admissions? Well, this talk about "disrupting" education through online courses (e.g. edX, Coursera) is really about improving social and knowledge graphs, and that's really exciting. Admissions officers just document; they take notes on who looked impressive at a young age. I see admissions as a less interesting problem with time. If our generation does its job, no one will give a shit about college admissions in 30 years.