Thanks for taking the time to articulate this :) I just responded to lun4r about why the proximity-based model doesn't work in the East.

We're not sure if there is evidence that "big data" style network graph analysis of people's social circles works. But what we do know is that a lot of relationships materialize between friends of friends offline.

Having said that, we used that as the germ of an idea for Krush. We have some way to go before the algorithms are satisfactorily refined, but hopefully Krush will be "the evidence" that you're talking about.

Stay tuned!! :)