Jonathan’s link to Pete Leeson’s article on the Trial By Ordeal (the full article itself is fascinating and fun) raises a point that I stress when talking to students about crime and deterrence in law and economics–whether criminal punishment serves as an effective deterrent actually depends on whether it is perceived as accurate/truth-revealing and outcomes are not systematically biased around the mean, not whether it is actually accurate. Although the perception of accuracy is highly dependent on whether it is actually accurate, of course, what is most important is that it is not considered to be wholly random or systematically biased. This assumes that criminals are risk neutral. If they are risk preferring or risk averse then this changes the analysis at the margin.
And that’s the central point of Pete’s paper–that if the trial by ordeal was believed to be accurate then it would be effective. So the really cool and (to me) completely unexpected finding of the paper is that the overwhelming number of people who were tried by ordeal were actually acquitted. One would have thought that the trial was so goofy that the results would have been at best random and at worst a complete kangaroo court that produced convictions. I’ll let you read the paper as to why this turns out not to be the case.