In 37 U.S. states from 1979-1999, white incarceration reduced crime while black incarceration did not. I use state sentencing laws as instruments for both black and white incarceration rates, selecting which laws to include via Lasso (Belloni et al., 2012). The legislative changes that generate the exogenous variation are plausibly independent of crime rates. Changes in sentencing policy, particularly the introduction of determinate and structured sentencing, significantly affected both black and white incarceration rates, and in general white rates responded to a greater variety of policy changes. Using this identifying variation, I find that white incarceration significantly decreased robbery and burglary, while black incarceration had no effect on crime, or caused it to increase. I provide a rational account of how increases in punishment can lead to increases in crime, which matches some stylized facts of the period I study.
This paper argues that free will is a purely theological issue, and offers an error theory for the free will debates in analytic philosophy in terms of evolutionary naturalism. I introduce 'protean free will' (PFW) as the ability to play mixed strategies effectively in noncooperative interactions. Thence, I argue that traditional worries about divine foreknowledge, Frankfurt controllers, moral responsibility, and determinism are side effects of selective pressures for unpredictability in our evolutionary past. Finally, I interpret the Libet experiments as showing an adaptive response to such pressures. I conclude that PFW does most things most philosophers want free will to do, conditional on the nonexistence of God.