x
Another toy model of the control problem — LessWrong