The man and the tool — LessWrong