Scott Garrabrant comments on Alternative to Bayesian Score