Chances are

Floyd Rudmin, a Norwegian professor, calculates the odds of effectively finding terrorists using the NSA's big-net approach as being practically nil. Making the mesh fine enough to catch any reasonable percentage of the tiny number of terrorists who might actually be in this country, by including enough suspicious words or contacts, will catch such a huge number of non-terrorists that they'll still be searching for needles in haystacks. And tens of thousands of us will be the haystacks, hauled off in the middle of the night to be interrogated and held indefinitely (Halliburton has been given a contract to build new detention centers here in the US).

Rudmin explains the probabilities in some detail. As a simple example, suppose the NSA used the 1993 bombing of the World Trade Center and the 1995 bombing in Oklahoma City as their model for terrorists. The bombers bought fertilizers, went to hardware stores, and rented trucks. So did thousands of other people. The 9/11 hijackers, if you buy the official story, flew on planes, stayed in motels, and frequented strip clubs. They were hardly alone in any of that.

And can you imagine the NSA trying to make a list of words to identify terrorists? Bomb, Osama, kill... Who knows? How many non-terrorists use those words at least occasionally? Just about all of them, I'd guess.

Anyway, the point is that the big-net approach is useless--for catching terrorists. However, as a means of repression, and for quickly targeting political opponents, it's got a lot going for it.