Forums

Live Chess Doesn't Recognize Dead Positions

Sort:
EndgameEnthusiast2357

Comment Deleted. Please check my most recent thread on this topic if you're interested, thank you!

WSama

That's just chess. You may have mate in 7, but the servers aren't going to automatically end the game for you, that would be weird.

Speed chess is as much about time management as it is anything else.

WSama

Sorry about the diagrams, didn't look into them much.

Let's take the first one for example:

The fifty move rule and the three fold rule should eventually kick in. It would be a burden on the servers to have to analyse every live ongoing blitz games for dead positions as the one above.

WSama

Fair enough. You are indeed a student of the endgame. I'd like to extend this invitation to join my personal club Z-Gardens whose name was in part inspired by the endgame.

WSama

Interesting tastes, EndgameStudier. Join right up and we can start sharing some diagrams with everybody.

mercatorproject

If you are down on the clock, you are lost here, Sad fact, where I have been the loser far more often than not. Sometimes I save myself by 3 fold reps, but many times that is not possible in the remaining half minute if the guy ahead on time has a lot of places to put a piece like a bishop. 

JamesColeman

You’re talking about having the server simultaneously constantly analyse every single game - thousands at any given time - in real time to pick up the comparatively rare occasions where despite there being mating material, there’s no possible sequence of moves to win, even if both players cooperate.

 

That server strain would most likely create an insane amount of lag. You’re technically right of course, but I just don’t think it’s practical to implement.

kingandqueen2017

It only recognizes a draw when there is insufficient material left on the board, such as a king vs king, knight, bishop, etc. That would be the case in over the board too. If you flagged when your opponent still had pawns, you'd lose, since you can still technically "mate" with those pieces.

GrandMasterNoob21

There's the 50 move rule.

WSama

Like everyone already said, it's just too much lag. And I don't like the idea of sites running more and more JS scripts on my browser. Next thing you know you'll need a quad core phone just to open a single site. It's a bad trend.

ArtemKozirev

At @WSama, you are confusing the hardware/software problems with the correct reasoning from Endgame Studier who is absolutely right about the dead positions. The first step to solve this issue by looking for a creative solution is to acknowledge that the current system is simply unable to detect the dead positions, it is the elephant in the room and your solution is that everyone keeps his mouth shut and close the eyes to what is clearly a problem.

mercatorproject
ArtemKozirev wrote:

At @WSama, you are confusing the hardware/software problems with the correct reasoning from Endgame Studier who is absolutely right about the dead positions. The first step to solve this issue by looking for a creative solution is to acknowledge that the current system is simply unable to detect the dead positions, it is the elephant in the room and your solution is that everyone keeps his mouth shut and close the eyes to what is clearly a problem.

Don't play low time controls if you are not quick enough, or cannot handle being flagged, is my advice.. They do not happen all that often anyhow, 

ArtemKozirev

@mercatorproject, do not go off the point, this is not about low time controls, or quick enough, or being able to handle being flagged. This is not about that and you know it. Please do not give advice when nobody is asking you for advice. Do not justify or try to minimize the subject of the problem here by saying "They do not happen all that often anyhow". The system has a problem which is it does not recognize dead positions, full stop.

ArtemKozirev

Fully agree with you Endgame studier. It is about making a favor to chess.com to help them improve and make it better. I charge nothing for this to anyone, also not being paid by anyone, of course if they have their trolls who desperately want to close people's eyes and most of all shut everybody's mouth before something which is simply wrong. Perhaps they think of improving by not letting the word spread ...

mercatorproject

I must say that the Study by OP makes him a Latter-Day A .A. Troitski, Averbach would have loved it. That Knight is not merely on the dim rim but in the very corner. As for the Black Bishop, he is beautifully bad, and the course into b8 must have been quite something. Both the Knight and the Bishop and the surrounding pawns had to tread softly. Black's contempt for Centre should have been punished long before the position was reached. 

Ask White's Play, it was at least open to the criticism of being rather timid, I feel.

The position is not dead in every respect, is it? It certainly has a living History. 

  

Williamfwm
ArtemKozirev wrote:

@mercatorproject, do not go off the point, this is not about low time controls, or quick enough, or being able to handle being flagged. This is not about that and you know it. Please do not give advice when nobody is asking you for advice. Do not justify or try to minimize the subject of the problem here by saying "They do not happen all that often anyhow". The system has a problem which is it does not recognize dead positions, full stop.

 

Don't be rude to others when they express ideas you don't like. It's not important that the system recognize these positions, "full stop". They're perfectly within their rights to express that it doesn't matter. They don't have to speak only in the manner you prescribe.

OP, Stockfish is open source so you're welcome to implement the dead position checker (C++). Or you can write a custom one in PHP which apparently they use on the server here. I think you're right in saying that it isn't a complex problem to check; the engine doesn't have to go out to arbitrary depth, it just needs to discover that nothing is capturable.

It would be done cheaply in terms of CPU time but IMO it just. doesn't. matter.

And still that's (small amount of server CPU time) * (a ton of games per day). It adds up for a feature that really. does. not. matter.

Not sure why you're talking about 1,000 moves. Obv 50-move is the rule you want here. Just premove the max premoves, you can't mess up

This almost never, ever happens, but in an OTB tournament there is a rule about this ("neither player could win through any series of moves...." or some such phrasing)

GrandMasterNoob21
EndgameStudier wrote:
mercatorproject wrote:

I must say that the Study by OP makes him a Latter-Day A .A. Troitski, Averbach would have loved it. That Knight is not merely on the dim rim but in the very corner. As for the Black Bishop, he is beautifully bad, and the course into b8 must have been quite something. Both the Knight and the Bishop and the surrounding pawns had to tread softly. Black's contempt for Centre should have been punished long before the position was reached. 

Ask White's Play, it was at least open to the criticism of being rather timid, I feel.

The position is not dead in every respect, is it? It certainly has a living History. 

  

Here's a mate in 15 puzzle, and this ain't no dead position either:

Now which player do you think is more stupid here? Black for getting into that cramped position, or white for losing all his pieces?

Was this a puzzle from chesstalk you copied?

WSama

Yeah. I don't get why @ArtemKozirev has to resort to disrespect. What's this business of closing eyes and silver coins?

In any case, the solution is not a difficult one to fashion or implement. The problem is it will cause lag, an issue we're already facing as players here on this site. If the analysis runs on the server, it all adds up with millions of bullet games taking place every minute. It's going to effect everyone for something that doesn't really matter, and is already handled by the 50 move rule.

If the solution runs on the client's browser, then sure this does away with lag issues, but I think we have enough scripts running in our browsers as it is. May as well download the app.

B1ZMARK
EndgameStudier wrote:
WSama wrote:

Sorry about the diagrams, didn't look into them much.

Let's take the first one for example:

The fifty move rule and the three fold rule should eventually kick in. It would be a burden on the servers to have to analyse every live ongoing blitz games for dead positions as the one above.

I highly doubt that as there is a built in analyzer for every position, every move, in every game on this site. I recently played a game where I forked my opponent's Queen and King with a Knight and he abandoned the game. I got a message saying that "your opponent has abandoned the game in a lost position, he will be restricted...etc" so it knew that it was a lost position based on 1 fork, so it could definitely recognize that these locked pawns positions with bishops in the same color as the pawns are theoretical draw. 

wait. it says that message for all opponents who disconnect and don't reconnect, it says they've been warned and perhaps be restricted. The site isn't detecting that one side is losing, it only is detecting that one side has disconnected.

B1ZMARK
EndgameStudier wrote:
Williamfwm wrote:
ArtemKozirev wrote:

@mercatorproject, do not go off the point, this is not about low time controls, or quick enough, or being able to handle being flagged. This is not about that and you know it. Please do not give advice when nobody is asking you for advice. Do not justify or try to minimize the subject of the problem here by saying "They do not happen all that often anyhow". The system has a problem which is it does not recognize dead positions, full stop.

 

Don't be rude to others when they express ideas you don't like. It's not important that the system recognize these positions, "full stop". They're perfectly within their rights to express that it doesn't matter. They don't have to speak only in the manner you prescribe.

OP, Stockfish is open source so you're welcome to implement the dead position checker (C++). Or you can write a custom one in PHP which apparently they use on the server here. I think you're right in saying that it isn't a complex problem to check; the engine doesn't have to go out to arbitrary depth, it just needs to discover that nothing is capturable.

It would be done cheaply in terms of CPU time but IMO it just. doesn't. matter.

And still that's (small amount of server CPU time) * (a ton of games per day). It adds up for a feature that really. does. not. matter.

Not sure why you're talking about 1,000 moves. Obv 50-move is the rule you want here. Just premove the max premoves, you can't mess up

This almost never, ever happens, but in an OTB tournament there is a rule about this ("neither player could win through any series of moves...." or some such phrasing)

And if each player only has 4 seconds left, premoves take .1 seconds, so you could only make 40 moves in that time, 50 move rule is not sufficient. You shouldn't have to continue making moves in ANY drawn positions (not just drawish, drawn). This is no different then forcing the players to keep going in a position where one side has a king and knight vs a lone king.

I mean, you could offer a draw. If neither side is low on time it will be a draw agreement.