solving the draw problem by allowing white to take back moves

[Uri Blass]

Suppose you change the rules of chess and decide that white is allowed to take back moves 1000 times during a game when after taking back moves we also take back times(meaning that if white started 1.e4 e5 2.Nf3 and find that you cannot win after 40 moves and you want to play 2.Nc3 then the times on the clocks are exactly the time that were after 1.e4 e5 2.Nf3 and the number of take back moves is 39 because white took back moves 2-40)

If white is going to fail to win even after 1000 take back it is a loss for white.

Suppose programmers change their engine for the new game.
What is going to be the expected result with top engines at different time controls?

Unfortunately today it is impossible to check it because engines are not designed to take decision of taking back moves

[BeyondCritics]

Pranksters amused themselves with this in the past, posting constructed engine losses, effectively pointing to engine weaknesses.
You could ask Opus 4.6 to construct a python script doing similar stuff on top of UCI. Could be kind of useful for debugging or constructing test positions.