An optimality result for modal UDT — LessWrong