Maps of Maps, and Empty Expectations — LessWrong