Stuart_Armstrong comments on Naive TDT, Bayes nets, and counterfactual mugging