If the death penalty deters murder, how often must it be applied to have that deterrent effect?

Jac has 2 posts on this subject. In the first post, he catches Steven Levitt (of Freakonomics) in what I think is a real contradiction. And in the second post, he points to something striking about the statistics on deterrence and deals with whether he's contradicted himself about what he said about Levitt's supposed contradiction. I'm not going to try to summarize all that here, but it has to do with how a would-be murderer contemplates the prospect of execution — rationally? — and how quickly and often the state needs to execute convicts in order to maximize the deterrence of murder.