Expanding the domain of discourse reveals structure already there but hidden — LessWrong