6

I will clarify my question with an example:

A grandmaster plays a game against stockfish, with a material handicap for stockfish, and the advantage is not clear for either of the players. Throughout the game, the position is a dead draw. However, during the endgame, stockfish discovers a loss 50 moves later (100 ply). For the game, stockfish was set to resign the game if the evaluation of the opponent's position exceeds 900 centipawns for 5 moves (2 ply). Consequently, stockfish resigns the game.

Was this a wise decision? What if the human blundered? Did stockfish resign a game which probably could've been drawn?

  • 1
    "Was this a wise decision?" As you write the decision was made (by humans) before the game, so it is king of pointless to argue about it. – user1583209 Apr 15 '18 at 15:49
  • "Noone as ever won a game by resigning" - Savielly Tartacover. (However, much later, Ivanchuk has drawn a game against Yusupov by resigning.) – Evargalo Apr 16 '18 at 9:13
  • @Evargalo, may you please give the link for this game? I am too eager to watch it. – Wais Kamal Apr 22 '18 at 9:26
  • @WaisKamal : There it is. Check the first comment. To sum up : Ivanchuk needed a win, when the game peterred out in a drawn endgame Yusupov offered a draw but Ivanchuk said "I resign" and left. Yusupov nevertheless insisted he had not won and the game was registered as a draw. chessgames.com/perl/chessgame?gid=1060181 – Evargalo Apr 24 '18 at 14:39
3

For beginner players, who tend to blunder a piece or mate in one regularly, there is no point to resign early. Any game, even in (objectively) clearly lost position will help them to gain experience.

For more advanced players, the decision to resign will depend on various factors and cannot be generalized.

Most importantly, an objectively clearly won/lost position can be of very different character. If it is a quiet, easy to calculate (e.g. endgame) position with few pieces on the board, and a clear easy to see winning plan, there is not much point to continue the game. On the other hand, if it is a wild position with lots of pieces and tactics, there are many chances for your opponent to go wrong and you would not resign immediately.

Some factors to take into account before resigning:

  • Do I have any chances for active counterplay?
  • Do I have any chances for drawing (building a fortress, stalemate,...)?
  • Is the position easily won for my opponent or is it rather a "theoretical" win, but difficult to do in "practice"? (e.g.
  • How much time does my opponent have left?
  • Over the last couple of moves, did my opponent show that he understands how to play the position for a win?

Similar things hold for the question: "Should I offer a draw?".

4

The time to resign is when think your opponent will win. Against a beginner, I would never resign, because they will most probably blunder and lose. Against a decent club player, I could resign after being a rook down in a lost position. Against a GM, I would resign after the 12th move if I thought I had a bad position (it's a good learning possibility to see the technique).

I find it insulting to not resign when your opponent has a won game. (An exception can be made for cases like checkmating with a knight and bishop).

In the case of a computer, it can only do as it's programmed to do. (Although humans now stop the game when an endgame database position has been reached to prevent the human from making a mistake the the computer looks up in a book. This is cheating if a human did it, why not the computer?) I don't think that the computer should resign this early, but he late should the computer resign?

  • 1
    @kmartin I think that's already factored in. Fred says he'd resign " Against a decent club player, [...] after being a rook down in a lost position." In particular, that requires that Fred has enough skill to recognize that the position is lost, and the belief that the opponent has enough skill to make the win. – David Richerby Apr 15 '18 at 11:31
  • 3
    "when you think your opponent will win" is too soon. If I had done that in my last tournament, I would have scored 0.5/4 instead of 2/4. Better to wait until you're sure your opponent will win. – D M Apr 15 '18 at 17:56
  • 4
    "Against a GM, I would resign after the 12th move if I thought I had a bad position" You're kidding, right? By that logic, if I'm paired against a GM, I should resign after the first move, because I know how it's going to end. Actually, when I have the rare opportunity to sit down across the board from a GM, I keep playing until it reaches the point where evey I know how to beat him with his position. – bof Apr 15 '18 at 21:59
  • 4
    You "find it insulting" when your opponent does "not resign when" you have "a won game"? Have I understood you correctly? If so, then may I suggest that you should be less easily insulted? Unless the opponent deliberately runs out his clock, so long as the opponent plays more or less at normal speed, why not just politely checkmate him? Resignation short of checkmate is the free prerogative of the losing player; for the winner to demand it is bad sport. – thb Apr 15 '18 at 23:08
  • 2
    @FredKnight it’s important for players to learn how to play from behind, even in lost positions where they have no belief that they’ll win. When they’re a Queen down it’s probably hopeless, but lost positions can often be turned into draws through clever play and getting the feel for careful calculation when all seems lost is a skill you have to develop at some point. It doesn’t have to be about them expecting you to make a mistake, and if you feel it’s a waste of your time then read a good book while they practice calculating under pressure. – Dennis Apr 16 '18 at 11:56
1

For a computer, unlike for a human, there's no fatigue associated with playing on, so there's no reason for the computer to benefit from resigning. But, if the organizers of a computer-human match or tournament want to attract human players, they may want to make some concessions like computer resignations.

Did stockfish resign a game which probably could've been drawn?

It's impossible to say. Was the position a known theoretical win that every GM knows how to convert, or was it a position that you pretty much need a tablebase to navigate? It is certainly at least possible for a computer programmed to resign in such situations to resign too soon. Perhaps the parameters need to be tweaked so that a 900 centipawn for 5 ply position must be reached within the first 10 ply to trigger the resignation.

1

People have been known to resign for a variety of reasons, but it is chiefly a question of good manners. You should not insult your opponent by wasting their time and you will make yourself look a bit of a fool. I think that most questions about the timeliness of resignation among human players come down to defining what is and what is not an insult. Suppose that in a moment of blindness I put my Queen where she will be lost to a Knight fork. I probably don't resign until I see that my opponent has noticed, but as soon as they make the fork I probably should. However, there is no foolproof algorithm to decide the question. Gawain Jones won a game in the British Championship after hanging his Queen for almost nothing.

Since computers cannot insult or be insulted they need an algorithm, and it is very hard to come up with one. I think that quite often the operator/programmer resigns on the machines behalf.

  • It's okay. Most players resign lost positions, but various opponents behave differently for various reasons. If an opponent wishes to play out a lost position for whatever reason, then, so long as the opponent plays at more or less normal speed, why, that is a slightly unusual choice but hardly a provocative one. As far as looking "a bit of fool" is concerned, some players might feel that it is even worse to look like a loser/quitter. Grace demands that the winner let the loser play out his game as he likes. Scrappy play at more or less normal speed in a lost position never hurt anyone. – thb Apr 17 '18 at 14:15
  • At higher levels of play, continuing a lost position is increasingly pointless, but it is also rare. I worry more about winners with bad manners. That is what brings the game into disrepute in my view. – thb Apr 17 '18 at 14:22
  • About twice a year I encounter somebody who plays until mate after being down by stupendous amounts of material. I let them do it because the alternative would be to create a scene, but I do feel insulted. You are quite right about the obnoxiousness of winnerswith bad manners(but the question was about resigning) – Philip Roe Apr 17 '18 at 19:48
1

It's never wise to resign a game if the only way you can lose is from some long, forced combination. There's a significant chance your opponent could miss it. In fact, in a typical chess position even Stockfish would not be able to calculate 50 moves / 100 ply ahead (unless the position was a simple endgame with few pieces).

It's only wise to resign if there's not a realistic chance of you not losing. In a situation where your opponent has to calculate a long forced variation, there's definitely a realistic chance he'll miss it.

Comparatively, if you're down a Queen in a quiet position against an opponent equal or greater than your strength, there isn't any realistic chance you won't lose (unless your opponent is really low on time).

0

You're asking the wrong question.

Stockfish chess engine should never resign in any scenario. There is absolutely no possibility that Stockfish should resign a game. There's no resign command in the Stockfish engine protocol. Please don't even think about that.

Instead, you're talking about graphical user interface (GUI). Most GUI don't resign; if you think you've won the game just stop the application. Forcing a GUI to resign for "... opponent's position exceeds 900 centipawns for 5 moves..." is also reasonable.

  • 4
    It seems heavy-handed to berate somebody for "asking the wrong question" (your emphasis) just because of the actually rather pedantic point that it's the GUI that decides to resign, rather than the engine. Correcting that kind of misapprehension seems more like a comment than an answer, to me. – David Richerby Apr 15 '18 at 14:02

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.