Exactly. Gather a could sources - as is already happening - and with a graph DB and not even grad level algorithms you could get a pretty accurate picture of enough people, given enough data.