Are there examples of chess games where the exact losing move has been found only with the help of engines?

Better still, games where the exact losing move cannot be found even by the best current engines?

  • 3
    Good question. I'm sure I remember, back before the days of engines, seeing annotated games of top positional players like Karpov, or Ulf Anderson, where the opponent's position just slowly rotted until it fell apart with no apart single move being the cause.
    – Brian Towers
    Feb 20, 2020 at 14:20
  • 1
    Most games between strong players, it really is not one move, unless it is a blunder...it is an accumulation of small errors, and they lead to a final error. Feb 20, 2020 at 15:45
  • @PhishMaster: I think you will agree that in every non-drawn game there is one, and only one, move that is the losing move, assuming best play in the sequel. This holds in theory, as chess is a finite game, but in practice it may seem like an accumulation of small errors, as the exact losing move can be extremely difficult to find.
    – exp8j
    Feb 20, 2020 at 17:57
  • 1
    @exp8j I do not think that at all. Feb 20, 2020 at 18:08
  • @exp8j if I blunder, then you blunder back, and then I blunder again, I've made two losing moves.
    – Allure
    Feb 21, 2020 at 20:50

2 Answers 2


Sure, examples exist. E.g. Game 6 of Carlsen-Caruana, World Chess Championship 2018:

[FEN "5k2/8/5pK1/3B1P1P/3n4/8/3b4/8 w - - 6 67"]
  1. Kg6? loses, but no human would have found the refutation.

As for games where the exact losing move cannot be found even by the best current engines - I spoke to a correspondence chess expert recently, and I'm confident they don't exist. Given a decisive game, an expert correspondence chess player will always be able to find the losing move, but it will not be trivial, and a lot of analysis is required.

  • Nice example. But just to check I'm getting you right, how could a correspondence chess expert find that 67.Kg6 loses if, as Kasparov famously said, "no human would have found the refutation"?
    – exp8j
    Feb 21, 2020 at 17:19
  • @exp8j they use engine assistance.
    – Allure
    Feb 21, 2020 at 20:48

Yes, but... Nobody will agree on all of them.

If two PERFECT computers played you could not tell the losing move or else it would be the first move. But I suspect a perfect game would be a draw.

The reason why moving loses is a myth is that the game is played by imperfect humans.

THE losing move is the next to last mistake. But what if a mistake confuses the opponent who resigns. What if the so called losing move wins this time but loses against another player in another game.

Except for the grossest of blunders it will be next to impossible to declare ONE move as THE losing move.

  • I think that in a given position, a given move belongs in one, and only one, of the 3 cases: winning, losing, or drawing; of course assuming best play from both sides, and also assuming that chess is a finite game (by the 50 moves rule). So in every non-drawn game, the exact losing move is very well-defined, although possibly very difficult to find.
    – exp8j
    Feb 20, 2020 at 17:41
  • I think that since we are not perfect nor are the computers that moves belong in a probability range just like quantum particles because we cannot know exactly what where they are nor how fast. Feb 20, 2020 at 18:10
  • I think that future computers will completely solve chess out. But this will not diminish the fascination it exerts on us! Time-limits aside, computers cannot interpret the unfolding of moves and plans like humans do, who value the aesthetics of chess and can embed its patterns in a complex of innumerable psychological associations.
    – exp8j
    Feb 20, 2020 at 18:23
  • Maybe. I doubt it will be in our lifetime. There are just too many possible moves from start to finish if you look at every variation. And if you dont look at every one you could get it wrong. Feb 20, 2020 at 19:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.