I would think it's not only feasible, but also more simple than what the Great Firewall already does.

You could probably do it on on 0.1% of the NSA's yearly budget. (Which, after some Googling, turns out to be 10M USD. That's even more than I expected.)