Null-boxing Newcomb’s Problem — LessWrong