MIRI/OP exchange about decision theory — LessWrong