You're not following. Yes, Bayes says that following up on stupid Facebook posts is also a waste of time. But the cost of following up on Facebook posts is low; nobody is surprised that the FBI finds out about stupid threats you post to Facebook, because Facebook is essentially public. The cost of following up on a Google dragnet is very high; the whole artifice of the Internet surveillance program is that it's not used this way.
Note that I'm not saying that Internet surveillance data doesn't get used this way. I'm saying that when it does, the USG sure as shit doesn't advertise the fact. They don't burn sources and methods for no good reason.
Note that I'm not saying that Internet surveillance data doesn't get used this way. I'm saying that when it does, the USG sure as shit doesn't advertise the fact. They don't burn sources and methods for no good reason.
It's almost totally implausible.