Are Interesting Problems Useful? — LessWrong