I'm not sure I agree with his premise that, in order to change the world you must work on a problem with "big data." I can't see how the energy crisis is fundamentally data-driven, for example, and it's hard to say that you couldn't change the world by working on that.
Regardless, best of luck to him wherever he winds up (within Google or elsewhere)! And I hope he keeps writing!
I think he's implying that it requires the same basic computer science hacking that is required for all of the "cat pictures" projects he points out. The point isn't that we should choose between data-driven and not data-driven; it's that we should choose between cat pictures and the human genome.
Regardless, best of luck to him wherever he winds up (within Google or elsewhere)! And I hope he keeps writing!