x
Improving the modal UDT optimality result — LessWrong