The dangers of dialectic — LessWrong