Conditional probability — LessWrong