Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Comment author: Stephen_R._Diamond 12 March 2009 05:59:00PM 0 points [-]

That the probability assigned to flipping a coin depends on what the assigner knows doesn't prove probability's subjectivity, only that probability isn't an objective property of the coin . Rather, if the probability is objective, it must be a property of a system, including the throwing mechanism. Two other problems with Eliezer's argument. 1) Rejecting objective interpretations of probability in empirical science because, in everyday usage, probability is relative to what's known, is to provide an a priori refutation of indeterminism, reasoning which doesn't square with Eliezer's empiricist analysis of knowledge. (Although, perhaps, objective probability is incoherent, but Eliezer hasn't shown this.) 2) A purely subjective interpretation of probability leaves probability in the same position as religion (or, maybe exactly, in the same position as Kant's a prior forms of understanding). A purely subjective interpretation doesn't explain the adaptive utility of taking the first Bayesian step of forming an a priori probability.