Understanding strategic deception and deceptive alignment — LessWrong