Last month Philip Yam of the Scientific American blog also had something on science and the death penalty. Yam estimates that execution error (i.e. executing an innocent person) to lie between 1/30 to 1/12 - and this in the United States! He asks: Would you be in favor of the death penalty if one innocent person were executed for every 10 guilty ones? How about 1 in 100?
Hmmm. One way to answer this is: does the death penalty save lives? (Of course this works through deterring homicide). Gary Becker, guru of the economics of crime and the family, is rather sanguine about the deterrent effects of capital punishment. This is based on a priori analysis (using the common sense observation that the typical individual dislikes death more than life imprisonment), combined with some early empirical studies (e.g. by Isaac Ehrlich).
The empirical evidence is reviewed by a recent NBER paper. (Hat tip: Ben Muse.) According to this paper, the link between capital punishment and deterrence is inconclusive; earlier studies that appear to have established some connection suffer from serious flaws.
So, should we be killing to save lives? Clearly the criterion of deterrence and net reduction of the death rate provides no sound justification (as of yet) for capital punishment. Every polity that does institute capital punishment would have to justify it using other ethical imperatives, rather than a practical, measurable impact on public safety.