The harnessing of complexity — LessWrong