How to Frame Negative Feedback as Forward-Facing Guidance — LessWrong