To those people writing in snarkily that on the precise definition of the

statistical term “expected value,” Posner is correct, I say

(1) Posner does not use the term “expected value.”

(2) His book is plainly addressed to laymen, few of whom could be expected

to be acquainted with statistical terms of art.

(3) Feugh!

I do not believe, in fact, that Posner himself is so acquainted. On the

evidence of that sentence, and some other remarks in the book, and long

experience of trying to explain this sort of stuff to well-educated but

un-mathematical people, it seems highly probable to me that Posner believes

that if the probability of something happening on a trial is one in a

thousand, and if you conduct a thousand trials, you could “expect” — in the

colloquial, not the statistically-precise sense — that the thing will

happen once. Which is not so. Your rational expectation would be that it

would happen either once, or not at all, with about equal probability,

though you could nurse some modest minor expectation that it would happen

twice or more.