Good luck, Mr. Rationalist — LessWrong