[SEQ RERUN] Optimization and the Singularity — LessWrong