Any-Benefit Mindset and Any-Reason Reasoning — LessWrong