A note on misunderstanding the boundaries of models — LessWrong