When Intuitions Are Useful — LessWrong