Tuesday, July 18, 2006

The Probability of Terrorists

[The Politics of Paranoia and Intimidation][1] article by Floyd Rudmin tries to show how futile the NSA domestic surveillance effort is by throwing around a bunch of math. To the casual reader, it looks impressive, but if you take a closer look at the math, you'll realize that Floyd demonstrates the program can actually be pretty effective.

[1]: http://www.lewrockwell.com/orig7/rudmin1.html

I'll leave out the drama, and just quote some of the numbers from Floyd's article.

Suppose that there are 1,000 terrorists ...
NSA surveillance has an accuracy rate
of .40, which means that 40% of real terrorists
in the USA will be identified ...
NSA's misidentification rate is .0001, which
means that .01% of innocent people will be
misidentified as terrorists

First of all, I'll point out that innocent people are investigated all the time. Wouldn't you think that most investigations involve more innocent people than guilty people? Floyd seems outraged by that but I don't think it's surprising.

300 million people, 1000 terrorists, 40% accuracy and 0.01% misidentification translates to a false positive rate of 98.68%. That means the NSA will find 1 real terrorist after investigating less than 100 suspects. I'd say that's a pretty productive use of resources.

Floyd assumes that all 30,400 suspects will be investigated. That seems unlikely. He also assumes that investigating somebody would "damage" their life. In practice, I'd bet the NSA could investigate you without you even being aware of it.

Floyd goes on...

Suppose that NSA's system is really, really, really good,
really, really good, with an accuracy rate of .90, and a
misidentification rate of .00001

Wow, that means only 4 innocent people need to be investigated to find 1 terrorist. That would be great! What is Floyd Rudmin complaining about?!

No comments:

Post a Comment