Bayesian view of scientific virtues — LessWrong