The parable of the underdog — LessWrong