[SEQ RERUN] Recognizing Intelligence — LessWrong