Steve [*] Schneier (security expert and tee-shirt provider to the cognoscenti) has a post in which he pokes at the massive costs of counter-terrorism, apparently out of all proportion to the threat. However, he has a bizarrely wrong calculation: I quote:
The death toll of all these is… sixteen deaths in the U.S. to terrorism in the past ten years. Given the credible estimate that we’ve spent $1 trillion on anti-terrorism security (this does not include our many foreign wars), that’s $62.5 billion per life saved. Is there any other risk that we are even remotely as crazy about?
I’m happy with the last sentence, but not the one before it: no Steve, that is $62.5 billion per person killed not per person saved. We’ve got no idea form the info you quote how many people were saved, so we can’t do the calculation you want to.
This obvious point is made in the comments, and SS even answers there, apparently failing to get the point. Certainly, he has updated the post several times, but still hasn’t corrected this rather basic blunder.
And just so I’m not being purely negative in this post, here is a nice picture from mt:
(I don’t agree with the exact shape, but the concept is correct)
[*] Ahem. See comments.
[Update: in the comments, MV manages to find a way to make BS's numbers work. I think it is somewhat smoke-n-mirrors, but the logic seems fine:
We assume that spending more resources leads to fewer deaths, but not to none. And we adopt the simplest possible model of this, viz:
R = k / D
Therefore, the cost of preventing one more death is (approximately) 1.dR/d(-D), which is
k / D^2 = R / D