Scott Alexander comments on Bayesian Cabaret