Lame Bayes Theorem Analysis

Bruce Schneier says that you can use Bayes theorem to show that NSA data mining systems are useless.  I think he didn't apply any economic analysis to the results he came up with.  Using the worst case example he gives, the NSA has a accuracy rate of 0.4 and a false positive rate of 0.01.  That means that roughly 31,000 people will be identified as terrorists but only 400 of them will really be terrorists.  If it costs $30,000 to investigate the 31,000 people to identify the 400 real terrorist, then it costs roughly a billion dollars.  However, if those 400 terrorists were in 20 cells each which could have created a billion dollars of economic damage, you saved $20 billion in damage.

Oppose NSA monitoring for civil liberities concerns or on constitutional grounds, but don't oppose it on some lame Bayes Theorem analysis which doesn't take into account economic costs.

Bruce is a smart guy, but I disagree with his conclusion.