[kgwgk]

> running simulations to see if models give sensible predictions, but that is itself a pretty frequentist thing to do

Is looking at probability distributions “a pretty frequentist thing to do”? Even when those models and simulations include _prior_ probability distributions? Sure, one can (re)define frequentist to include Bayesian models - as Gelman seems to want to do in that post. I just don’t see how this helps to clarify anything.