Using the universal prior for logical uncertainty — LessWrong