Relative likelihood - History — LessWrong