[SEQ RERUN] Marginally Zero-Sum Efforts — LessWrong