Draw by 3 fold repetition...

[Uri Blass]

here an example... but it's one example amongst many : https://lichess.org/Mm6BnWjv/white#92

Obviously you must be a quite weak player yourself?

1. Those are quite likely SF vs. SF games
2. White has simply no other moves and your opponent (=Black) forced the draw
3. Of course White (your SF) doesn't want to lose, so it has to take the draw

So much for creating a hypothesis about a big general problem, which does not exist

A quite annoying thread and your attitude in this thread makes me doubt I would like replying to you again.

Lol nobody forced you to reply. You are quite agressive. If you have problems in your life others are not responsible for it.
Let me think that an engine should avoid 3 fold repetition draws (especially those forced by king's checking) in its search for playable positions, and this could be a nice improvement if these type of draws could be avoided in top engines matches.

If you think that some move should be different then you should give a position and say what move stockfish played and what it should play based on your opinion.

[Pi4Chess]

If you think that some move should be different then you should give a position and say what move stockfish played and what it should play based on your opinion.

I am just asking those who know how engines are coded why a top engine like stockfish "accepts" to get into a position where the opponent can draw by 3 fold repetition with forced King's check. Is it unavoidable ? Is this took into account enough to discard certain moves from others within its evaluation ?
I am just asking.

[Uri Blass]

If you think that some move should be different then you should give a position and say what move stockfish played and what it should play based on your opinion.

I am just asking those who know how engines are coded why a top engine like stockfish "accepts" to get into a position where the opponent can draw by 3 fold repetition with forced King's check. Is it unavoidable ? Is this took into account enough to discard certain moves from others within its evaluation ?
I am just asking.

Simply because it found nothing better.
It does not have to be because it is unavoidable and it is possible that a different line is going to lead to a different equal position.

[Pi4Chess]

What an engine finds better depends on the rules the authors give to it to find it better (and the improved results). So your answer is like the snake that bites its tail.
For instance would it be relevant to "learn" to an engine that a draw by insuficient material is better than a draw by repetition ?
May be it's already done?

[Ras]

What an engine finds better depends on the rules the authors give to it to find it better (and the improved results).

If you think that Stockfish would have had better alternatives, why don't you show them? Asking why Stockfish draws amounts to asking why it doesn't win, but in chess, you don't win if the opponent doesn't make losing mistakes.

There is already the contempt approach in Stockfish, and that means it will only accept a draw through whatever means if the alternatives would be worse. So either it has no better moves, in which case it will want the draw to happen. Or it sees that the opponent would have something even better than repetition, but the opponent fails to see that, too, in which case Stockfish will be happy to agree to the draw.

For instance would it be relevant to "learn" to an engine that a draw by insuficient material is better than a draw by repetition ?

A draw is a draw, and it doesn't matter how you achieve it.

[RubiChess]

here an example... but it's one example amongst many : https://lichess.org/Mm6BnWjv/white#92

Obviously you must be a quite weak player yourself?

1. Those are quite likely SF vs. SF games
2. White has simply no other moves and your opponent (=Black) forced the draw
3. Of course White (your SF) doesn't want to lose, so it has to take the draw

So much for creating a hypothesis about a big general problem, which does not exist

A quite annoying thread and your attitude in this thread makes me doubt I would like replying to you again.

Lol nobody forced you to reply. You are quite agressive. If you have problems in your life others are not responsible for it.
Let me think that an engine should avoid 3 fold repetition draws (especially those forced by king's checking) in its search for playable positions, and this could be a nice improvement if these type of draws could be avoided in top engines matches.

No. It is YOU that is quite cheeky ignoring the advises of Guenther for three times and then calling him aggressive and assuming "problems in his life" when he gets impatient by your ignorance.
Believe us: Good engines know about the 3 fold rule and if they agree to a 3 fold they just can't find anything better even at depth 99 or whatever they searched. If you keep believing that this could be improved: Create a pull request.

Regards, Andreas

[RubiChess]

What an engine finds better depends on the rules the authors give to it to find it better (and the improved results). So your answer is like the snake that bites its tail.
For instance would it be relevant to "learn" to an engine that a draw by insuficient material is better than a draw by repetition ?
May be it's already done?

But a loss by inferior endgame facilities is worse than a draw by 3-fold. What now? I'm happy about every 3 fold my engine can force SF into. Sadly it doesn't happen very often. Guess why.

[Pi4Chess]

@ras : when you Say a draw is a draw, are you saying that when SF evaluates a position as a draw he is always right ?

I was thinking about Sting Black hole generator that can sometimes get out of 0 evaluation and fortress positions (but with much more time).

A forced king's check 3 fold repetition is a SURE draw while an evaluated 0 draw might be not as sure as it seems if you assume you can play against a stronger engine (assuming evaluation is not perfect).
I don't know How and if i could write a pull request out of that, but if someone understands what I am trying to Say it might be useful.

Of course since i am not a programmer and what i am saying is possibly already adressed by comtempt or other evaluation funcions and cannot be improved, i am just looking like an ignorant. But sometimes ideas come from new untrained and exterior eyes.

[RubiChess]

A forced king's check 3 fold repetition is a SURE draw while an evaluated 0 draw might be not as sure as it seems if you assume you can play against a stronger engine (assuming evaluation is not perfect).

You got it! The problem of "wrong" 3 fold can be solved by better (perfect) evaluation and search. This is exactly what engine developers are doing (or at least trying) all the time.

Regards, Andreas

[Pi4Chess]

A forced king's check 3 fold repetition is a SURE draw while an evaluated 0 draw might be not as sure as it seems if you assume you can play against a stronger engine (assuming evaluation is not perfect).

You got it! The problem of "wrong" 3 fold can be solved by better (perfect) evaluation and search. This is exactly what engine developers are doing (or at least trying) all the time.

Regards, Andreas

. Ok what you Say seems obvious,but until we achieve a perfect evaluation May be we can "tell" to the engine that 3 repetition king's check is 100% draw while 0 evaluation is less than that.