On Negative Feedback and Simulacra — LessWrong