True or False Chess is a Draw with Best Play from Both Sides

Sort:
ponz111

nirav it is true that when you give the starting position to a chess engine it will show White with a slight advantage. But the slight advantage is not near enough to win.  Also I have noticed the longer a strong chess engine looks at the starting position--this slight advantage becomes less.  

 

ponz111

Regarding poker, of course there is a lot of luck involved but the best players win money in the long run even though with any given tournament they are more likely to bust out with no money than to actually win money.

It is also true that with age--my poker abilities have become less but I won the tournament the very last day I played at the casino in Illinois!?

lendacerda
ponz111 wrote:

Also I have noticed the longer a strong chess engine looks at the starting position--this slight advantage becomes less.  

 

 

Yes i noticed that too. As Stockfish went from v8 to v10 to v11, the evaluation went slightly down. Also goes down if you put the starting position but with black to move xD

lfPatriotGames
nirav2611 wrote:

I do not think chess is a draw with best play.why? Because if you show the starting position to a computer it wil give white an advantage 

That's possible. It depends on when the advantage happens. If there is a whole knight advantage at the very beginning of the game, it's probably a forced win from the start. If at the very end of the game, it's KN vs K then that whole knight advantage means nothing. So the earlier in the game there is an advantage (say from the first move) the more likely that side is to win. I'll bet at the very highest levels of chess the side with the first move wins more than it loses. 

lendacerda
lfPatriotGames wrote:
nirav2611 wrote:

I do not think chess is a draw with best play.why? Because if you show the starting position to a computer it wil give white an advantage 

That's possible. It depends on when the advantage happens. If there is a whole knight advantage at the very beginning of the game, it's probably a forced win from the start. If at the very end of the game, it's KN vs K then that whole knight advantage means nothing. So the earlier in the game there is an advantage (say from the first move) the more likely that side is to win. I'll bet at the very highest levels of chess the side with the first move wins more than it loses. 

 

Couldn't agree more. As for "the side with the first move wins", white does win more, statistically, wether it be master games, online games and even engine games.

ponz111

As stated before a slight advantage does not equate to a win. Of course in games with errors it is better to have a slight advantage than if your opponent has a slight advantage. 

JimDiesel22

Hi ponz111,

This is a very interesting question, and I applaud you for continuing this conversation for seven years.

But I disagree with you, and I think I can convince you. In fact, I think there is no way to definitively know without brute force calculations. You've claimed that out of all 10^10 chess games played, all required a mistake to not be a draw. A) this is just as hard to prove as the original statement and B) even if it were true, 10^10 is an incredibly insignificant amount of games. In fact, there are more possible games played in 8 moves and then drawing. There are estimated to be more possible games than atoms in the observable universe (https://en.wikipedia.org/wiki/Shannon_number).

It is possible that in the completely expansive tree of every possible move, that the game could decisively end in a win/draw/lose for white because it isn't possible to calculate with modern machines (never mind the human mind). 1. e4 e5 could be a mistake. The only way to know is to evaluate every option to exhaustion.

Ziryab

It is a remote possibility that chess could be decisively in favor of White or Black, but more than a century of practical play (several decades with very strong computer assistance in postgame analysis) has supported Steinitz’s claim that chess is a draw with best play.

Every decisive game contains identifiable errors. Some errors are quite complex and subtle, but they are errors. Sometimes the critical error remains unclear and debatable.

But, I have yet to see a game that was decisive and contained no error.

Account_Suspended

The question of the OP predisposes some way of assessing the position and we can only do with human programming. Since it is humans making up assumptions in judging positions (compare various engines for the exact position and you will get varying numbers), we cannot have absolute certainty as to whether not a mistake was not done.

I'm of the opinion there's a chance white can force a win and the lines permitting this may be only in the thousands out of hundreds of billions. It has yet to be proven. Some will make the claim it has been proven without giving concrete evidence of it.

Thee_Ghostess_Lola

thats not the question. the question is why or why not is chess a draw. and all i can come up w/ is that its cuz theres not enuf pawns on the board. the minute initiative that white gets cant be converted into a full point (pawn) w/ only (8) pawns. given a larger board and more pawns ?...then probably yes. but then that wouldnt be the chess as we know it.

we can observe all this by mirroring whites moves w/ black. try the english symmetrical. black cant keep up the move mirror w/out giving up a signi-edge. just set alphaZ against itself and see wut happens.

lfPatriotGames
Ziryab wrote:

It is a remote possibility that chess could be decisively in favor of White or Black, but more than a century of practical play (several decades with very strong computer assistance in postgame analysis) has supported Steinitz’s claim that chess is a draw with best play.

Every decisive game contains identifiable errors. Some errors are quite complex and subtle, but they are errors. Sometimes the critical error remains unclear and debatable.

But, I have yet to see a game that was decisive and contained no error.

Hey, look at that. We agree. Because nobody has ever seen a decisive game with no mistakes, it's impossible (so far) to know if one side can force a win. I also agree with toom, that if there is a forced win somewhere, it's probably a very low number of possibilities that produce those wins. 

Even with the centuries of play, and recent very good and very advanced games it's still far too early to know for sure. I think computers are in their early, early, infancy. I think about how quickly (and oddly) really modern chess computers came up with new ways to play. Two hundred years from now we may still not know the answer but at least we'll be closer to making an educated guess. 

checkmator11111

If you mean optimally, then it is a draw: if you analyze it you get a draw

checkmator11111

If you mean no mistakes then I think it can be decisive

blueemu

As the skill level goes up (from Beginner to class players to Masters to Grandmasters to Super-GMs) the percentage of draws increases.

Extrapolate that to the limiting case of perfect play, and chess is a draw.

JimDiesel22

None of these arguments have any merit.

@blueemu That's a laughable argument. Chess data bases are on the order of million. Let's say a trillion (10^12) games have been played to be generous. There are roughly 10^120 possible games. An argument from experience is irrelevant. Even assuming the games are unique, less than a googolth of the game has been played.

@Ziryab Same as above. Experience doesn't matter. Just because some (yes "some") games were played and the only decisive results contained errors we could understand doesn't mean there doesn't exist an irrefutable line white can play to win.

@Optimissed A trend doesn't prove anything. Imagine if there is 1 line that white can play to win. That trend would exist unless the line was discovered.

 

The way I see it, I think an answer is actually possible without brute force. I can imagine a way you get to that solution by building a model based on games that are definitive at a certain depth but training the model at a shorter depth. Then estimate the longest possible game and build a graph of accuracy vs game length past depth to get an estimation for accuracy of a full game. The issue is there could be a very small number of forced wins/loses. Also, if this was a reasonable approach, someone probably would have done it.

JimDiesel22

@Thee_Ghostess_Lola You say very confidently "That's not the question" below the title: "True or False Chess is a Draw".

lfPatriotGames
Optimissed wrote:

<<if there is a forced win somewhere, it's probably a very low number of possibilities that produce those wins.>>

Honest, it's logically impossible. I once tried to explain why but the explanation wasn't understood.

It might be logically impossible. But that doesn't mean it's literally impossible, does it? Think of all the things that exist but defy logic, like honeydo lists. 

I have no idea if chess is a draw with "best play". Nobody does. I just think that white has a proven advantage. I dont think anybody disputes that. I agree that the advantage is very small, and because of how chess is played even an advantage of a bishop and pawn can still be a draw. But, I just think that it's far too early to say for sure one way or another. I think the very best computers in the world, that can easily routinely beat the world champion, still make huge mistakes. Imagine the computing power of chess machines 200 years from now. 

If 200 years from now a computer hasn't figured out a way to win every time, then I will agree it's a draw with "best play". But not a day before that. 

gullupakka

i dont think so, ponz111 there are thing like sacrifices and there are a million tactics!

Ziryab
JimDiesel22 wrote:

None of these arguments have any merit.

@blueemu That's a laughable argument. Chess data bases are on the order of million. Let's say a trillion (10^12) games have been played to be generous. There are roughly 10^120 possible games. An argument from experience is irrelevant. Even assuming the games are unique, less than a googolth of the game has been played.

@Ziryab Same as above. Experience doesn't matter. Just because some (yes "some") games were played and the only decisive results contained errors we could understand doesn't mean there doesn't exist an irrefutable line white can play to win.

@Optimissed A trend doesn't prove anything. Imagine if there is 1 line that white can play to win. That trend would exist unless the line was discovered.

 

The way I see it, I think an answer is actually possible without brute force. I can imagine a way you get to that solution by building a model based on games that are definitive at a certain depth but training the model at a shorter depth. Then estimate the longest possible game and build a graph of accuracy vs game length past depth to get an estimation for accuracy of a full game. The issue is there could be a very small number of forced wins/loses. Also, if this was a reasonable approach, someone probably would have done it.

 

Your sense of logic needs to be informed by a better sense of chess. Most of the unplayed variations will not change the conclusion because they are so bad that they lose at once.

It's funny how you attack an argument that @IfPatriotGames thought was agreeing with you, and partly it was. But, contra @IfPatriotGames, I was trying to steer between @Ponz111 (whose views expressed here I'm mostly in agreement with) and your point (which I did not wholly dismiss). Evidently the nuance in my comment was too poorly expressed for comprehension.

Welcome back, BTW.

JimDiesel22
Ziryab wrote:

Your sense of logic needs to be informed by a better sense of chess. Most of the unplayed variations will not change the conclusion because they are so bad that they lose at once.

 

Oh, most? I had no idea!? What does most mean? Are only 1% good variations? .1%? I'll consider the argument when it's around .0000000000 0000000000 0000000000 0000000000 0000000000 0000000000 0000000000 0000000000 0000000000 0000000001%. The conclusion of "Is chess a forced draw?" only requires one line to be false.

Your personal experience with chess doesn't matter. Everyone's personal experience combined doesn't matter. The only acceptable approach is AI.

Your sense of logic needs to be informed by a better sense of logic.