Did anyone say it wasn’t simply a problem of the last state?
Actually, I think if one would really want to stick to the proposed algorithmic solution then your proposal is not working: There are no “clues” since there is no guessing involved. There is a fixed set of outcomes of the tree. Whatever is not in the tree can’t be answered.
I will certainly admit that probably no one would do it that strict. However, when you guessed wrong once your next answer must be right. So it’s probably better to go the reset route instead of trying to guess a thing not in your tree from clues.
