Chess will never be solved, here's why

Sort:
Elroch
Optimissed wrote:
Elroch wrote:

 

Apparently, you and @tygxc  now "know" that chess is a draw. In your case this is inconsistent with your earlier position, but there is no necessity for this type of knowing to be consistent.

It may be inconsistent with my earlier position as you understood it but I've always maintained that I know that chess is a draw.

Quote from a book on philosophy I happen to be reading:

<<How do you know that you know the stuff you think you know? Take away the option of answering, “I just do!” and what’s left is epistemology.>>

Yes of course. Where does this knowledge come from and why do we classify it as knowledge? Earlier I pointed out that we can subjectively classify something as knowledge but that the common concept is that knowledge is something that's commonly shared. Obviously it may not be shared among a majority of people. It may be among a small minority or, obviously, in a minority of one when a person sees something physical, unseen by others: for instance, as a witness to a murder.

This business about chess isn't physical. It's cognitive and it's a possible interpretation of previous chess results. It seems a very reasonable interpretation. It has never been refuted. Those of us who believe it to be knowledge share the conviction that it will never be refuted. One or the other is correct and we think that our alternative is the one that is real. If so, then that would mean that btickler's opinion is the one that's deluded. Yet that does not excuse his use of the word "delusion", when referring to others. It's merely an effect of his weakness at debate. He ought not do it. It's a childish, Facebook kind of thing.

That doesn't mean that every belief that is based on something (rather than nothing) is certain. It is extremely common for people not to recognise that their basis for a belief means it is uncertain.

Yes, absolutely right. However, you and I are agreed that there's no likelihood of a full solution for chess coming soon. Personally I think it's impossible. That means that if we wish to try to answer the question of whether chess is a draw, we have no deductive argument to fall back upon: nor any likelihood of there ever being one. Therefore we have to go with evidence, such as we have. That means inductively, doesn't it?

 

The answer is really very simple: we have to accept there is uncertainty.

I would have thought you would agree on the semantics that where there is uncertainty, the correct word is "believe" rather than "know".

For example, a very rational person holds a lottery tick and says "I believe I will not win the superdraw tonight". She may be aware that there is a theoretical 1 in 500,000,000 chance of winning the roll-over prize. She does not say "I know I will not win the superdraw tonight" because she considers the distinction based on a very unlikely 2 in a billion chance important and it may have been why she bought a ticket.

Now, here is an example of a less rational person, apparently incapable of understanding this point:

tygxc wrote:

#3831

"It is widely hypothesized that classical chess is theoretically drawn"
++ That is the cautionous way to say "Chess is a draw" like Fischer said.

Firstly, no it isn't.  Being "widely hypothesised" is so different to being a certainty that it is surprising that anyone would make the claim that the two are the same.

Given this degree of sloppiness it is no surprise that I was unable to justify the claim that Fischer ever said "chess is a draw" (not that if he had it would carry any more weight than other things he said that were false).  Rather I find that Fischer thought that it's almost definite that the game is a draw theoretically, for which there are three references.  Note the appropriate uncertainty.

lfPatriotGames
Optimissed wrote:
lfPatriotGames wrote:
 

Any guess about the ultimate outcome of chess is hypothesizing. The guess "chess is a forced win for white" is just as valid as the guess "chess is a draw". Both are equal because both have good reasons to exist. 



Hi, equally valid doesn't mean equally correct. "Valid" sort of means "on subject". An invalid guess means that the answer to "is chess a draw?" is "a piano".

You're confusing guessing with scientifically based hypothesis. We have every reason to believe that the scientifically based hypothesis that chess is a draw is correct. There's no reason to believe "chess is a win" to be correct, so you're comparing mushrooms with octopusses. Superficially similar but not alike in reality.

No. We are just as close to proving chess is a forced win for white as we are to proving chess is a draw. So far nobody has even come close to proving either one. Not even close.  Lots and lots of guesses, lots of assumptions, lots of hypothesizing and grandstanding. Lots of faith, lots of belief, lots of wishful thinking. But no proof.

So as long as the two choices are equal, and they are, then there is no harm in choosing either one. Whatever your personal preference is on the topic, it's perfectly valid.  Which is why for those who say it MUST be one way, and can't be the other, prove it. Obviously it can't be proven, so until it can, it's probably best to keep an open mind on all the possibilities. 

lfPatriotGames
tygxc wrote:

#3846
"you and I are agreed that there's no likelihood of a full solution for chess coming soon."
++ On no basis at all. I agree with Sveshnikov it can be done in 5 years.

"That means that if we wish to try to answer the question of whether chess is a draw, we have no deductive argument to fall back upon: nor any likelihood of there ever being one."
++ There are deductive arguments. White is 1 tempo up. 3 tempi equals 1 pawn. It needs 1 pawn to win. 1 tempo is not enough to win.

You believe a full solution to chess can be found within 5 years? Well, you also seem to believe chess is a draw, so there's that. 

Of course both those things are possible. I just don't think either one is very likely. You might believe chess will be proven a draw within 5 years, I believe it will be proven a forced white win within 200 years. And there is always the possibility we are both half right. Maybe chess will be proven a forced white win in 5 years. Or proven a draw within 200 years. 

tygxc

#3852

"We are just as close to proving chess is a forced win for white as we are to proving chess is a draw." ++ And as chess is a forced win for black too?
There is massive evidence for chess being a draw, none for a white or black win.

"Lots and lots of guesses, lots of assumptions, lots of hypothesizing and grandstanding. Lots of faith, lots of belief, lots of wishful thinking."
++ Lots of facts and figures. AlphaZero, TCEC, ICCF, human GM.

"So as long as the two choices are equal" ++ You mean the 3 choices?

"and they are"
++ They are not. It is not because there are 3 possibilities that they are equally likely.

"Whatever your personal preference is on the topic, it's perfectly valid."
++ No, that is not true.
Look at the position after 1 e4 e5 2 Ba6. There are 3 possibilities: a draw, a white win, a black win. They are not equally likely. I know black wins this position, but I do not have a full game tree ending in checkmate for all possibilities. I know it is a forced checkmate, but I do not know in how many moves.

"Which is why for those who say it MUST be one way, and can't be the other, prove it."
++ I have listed the empirical (AlphaZero, TCEC, ICCF, human GM) and theoretical (1 tempo = 1/3 pawn < 1 pawn) evidence.
Even more: each move dilutes the initial tempo up, so if there were a white win, then it would be a short win, not a long one. However, a short win would have been found long ago.
Even more: the simpler game Checkers (32 squares, 24 men, 2 kinds of men) took 10^14 positions to weakly solve, more than 10^9 for Losing Chess (64 squares, 32 men, 6 kinds of men, just like chess) because Losing Chess is a white win with 1 e3 and Checkers is a draw. A draw is harder to prove than a win. If chess were a forced win, then it would have been weakly solved before Checkers.

"Obviously it can't be proven"
++ Chess is a finite game so obviously it can be proven.
Based on facts and figures I agree with Sveshnikov that chess can be weakly solved in 5 years.

lfPatriotGames
Optimissed wrote:
lfPatriotGames wrote:
Optimissed wrote:
lfPatriotGames wrote:
 

Any guess about the ultimate outcome of chess is hypothesizing. The guess "chess is a forced win for white" is just as valid as the guess "chess is a draw". Both are equal because both have good reasons to exist. 



Hi, equally valid doesn't mean equally correct. "Valid" sort of means "on subject". An invalid guess means that the answer to "is chess a draw?" is "a piano".

You're confusing guessing with scientifically based hypothesis. We have every reason to believe that the scientifically based hypothesis that chess is a draw is correct. There's no reason to believe "chess is a win" to be correct, so you're comparing mushrooms with octopusses. Superficially similar but not alike in reality.

No. We are just as close to proving chess is a forced win for white as we are to proving chess is a draw. So far nobody has even come close to proving either one. Not even close.  Lots and lots of guesses, lots of assumptions, lots of hypothesizing and grandstanding. Lots of faith, lots of belief, lots of wishful thinking. But no proof.

So as long as the two choices are equal, and they are, then there is no harm in choosing either one. Whatever your personal preference is on the topic, it's perfectly valid.  Which is why for those who say it MUST be one way, and can't be the other, prove it. Obviously it can't be proven, so until it can, it's probably best to keep an open mind on all the possibilities. 


You're definitely wrong there, because in no way is it as close to proving it's a win as a draw. There isn't even a hint that it's a win, so you're falling back on the pseudo-theory of agnosticism being the only possible result of a failure to deduce.

I could be definitely wrong. All you have to do is prove chess is a draw. From what I've seen nobody has even come close to that. I know lots of people have strong opinions and strong beliefs, but that's not proof. 

It would be like you can "prove" you can't win the lottery. As proof you buy a lottery ticket, using the best skills available. You don't win. You you do it again, but don't win. Pretty soon 99% of the time you don't win. Pretty soon virtually every time you play the lottery, you don't win. It may even be a fact that you don't win 99.99% of the time. So the assumption is that it's impossible to win the lottery. It's a fact. It can't be done. Because over 99.99% of lottery plays are not wins. 

But I don't consider that proof. I consider that a belief. 

If you can prove that 100% of all chess games, played with the "best moves" whatever that might be at the current moment, are draws, then that would be a start. But so far nobody has ever done that. 

So until then we are not any closer to proving either side. But it would be interesting to see if at the highest level if whites win percentage is going up, going down, or staying the same. 

DiogenesDue
tygxc wrote:

++ On no basis at all. I agree with Sveshnikov it can be done in 5 years.

You put a period after that first sentence when it should be a comma wink.png.

haiaku

@Optimissed

I have read your essay carefully. I have already stated in a previous post that since logic is based on postulates, in fact deductions are a form of faith. Therefore I do not agree with your distinction between cultists, scientists and theorists; we are all cultists to start with. But this discussion about epistemology and philosophy of science may lead too much off topic. The core point is that without an exhaustive proof, no scientific theory can be guaranteed to hold true in any possible case. Galilean relativity, Newton's law of gravitation and classical mechanics are golden examples. They were thought to be always true, "unbeatable" so to speak, and they were consistent with centuries of experiments. There was no evidence that they might produce quite inaccurate predictions. We all know how it went: they fail miserably under some circumstances.

In game theory, "optimal" is not a casual attribute. For chess, it means that an optimal player would be unbeatable in a match with an even number of games: if the game value is a draw, the optimal player would at least draw every game; if it's a win for either colour, the optimal player would always force the win with that colour. Therefore, the optimal player cannot score less than 50% of the points. Without a mathematical, exhaustive proof, a player cannot be guaranteed to be optimal, exactly like a scientific theory, without an exhaustive proof, cannot be guaranteed to always hold true.

This is not what people expect, when they read something like "the game xyz has been solved". They think about a definitive solution, that nobody will be able to disprove, ever.

DiogenesDue
stancco wrote:

Goos one 🤣🤣🤣

we are witnesses of a slack jaw dimness here

🤣🤣🤣

Did you know that your username means "tired out" in Italian?

DiogenesDue
Optimissed wrote:

Having thought about it for a few minutes while toasting some crumpets, I've arrived at the following conclusion.

No-one here has any just nor reasonable grounds to contradict the simple assertion, by tygxc and myself, that we know that chess is a draw by best play for both sides.

That is because you would be attempting to interdict our use of the verb "to know" and our correct use of it here. If you veto philosophically based discussion, which can only be because you don't understand it, then you cannot discuss what is meant by "knowing". If you have no understanding of the psychology of epistemology, or in plain English, of understanding knowledge and belief, you have no business on this thread in anything but a learning capacity.

That definitely applies to haiaku and to all others who have fixed  but highly simplistic opinions on this. If you felt able to comment on my little discourse on superstition, science and hypothesis, you would have attempted to do so by now. It's absolutely necessary to engage, if you don't wish to look like children. That's because you are trying to use philosophy and psychology to claim that tygxc and I cannot know something we claim to know, whilst having no knowledge of what you're talking about.

In particular, haiaku's hiding behind his "understanding" of science is ridiculous.

If only the topic of this post were "tell us what you know about chess being a forced draw"...oh wait, it's about solving chess, which is defined as proving it past personal belief.

All this diatribe accomplishes is to say "we have an opinion and we have stated it".  Bravo.

lfPatriotGames

From what I've heard not every game at the highest level is a draw. Sometimes one side wins. So if almost all chess games are draws that means chess is a draw, that also means since almost all lottery plays are not wins, winning the lottery is not possible. 

I understand that many chess games are draws. But I also understand that most lottery plays are not wins. For me, that doesn't prove anything. It's a pattern. But not proof.

If chess has been proven to be a draw, why are there wins? If it's been proven to be a draw, why do people disagree? There must be some reason not everyone agrees. 

I could just as easily say because white has a first move advantage, AND the percentage of white wins at the highest level goes up, that's proof chess is a forced win for white. But until white wins every single time it's only evidence, not proof. And until all games are draws it's evidence, not proof. 

Elroch

Even if every one of a billion games between the top players were draws, this would not prove chess to be a draw. The reason is that it is at least conceivable that there is a very narrow winning strategy that requires dozens of unique moves in every line and the players are just too weak to find them all (like current top engines are too weak to accurately play complex tablebase positions without help).

DiogenesDue
tygxc wrote:

#3838

"That figure is meaningless, as it is completely relative to the various engines' levels of play."
++ More time = higher level = more draws

"chess players (with no other applicable qualifications) are not qualified to make a judgment"
++ That is a ridiculous argument with no point at all. Should we ask a hockey player then?

No.  You shouldn't ask *any* players, of any game or sport, a question like that.  Do you ask a person riding a roller coaster to design roller coasters?

"Neither has written any kind of attempted proof." ++ But they expressed their expert opinions.

Read the topic again.  We're not debating opinions.  We're debating the possibility of proof.

"Solving chess cannot be achieved using Stockfish.  Not now, not 5 years from now.  Not 500 years from now." ++ That is your misguided opinion, without anything to back it up.

"Not one of them.  They would not even say that chess is a forced draw without hedging their statements, never mind how to get to that conclusion." ++ They did say that.

Nope, time after time, decade after decade, GMs have made statements about draws, and when pressed they always hedge their bets.  Because they are smart enough to know that they don't know the answer.

"So you agree with me then, he was pandering to an audience."
++ He was giving an interview to leave a legacy, as he was terminally ill with cancer.

What a great time to make a claim nobody will ever be able to call you out on...

"Your "50% accuracy" is a number that has no meaning in terms of solving chess."
++ You misunderstand. 50% accuracy is an arbitrary threshold to distinguish the few sensible positions from many non-sensible positions. If a position results from a game with 99% or 100% accuracy, then that does not mean the position results from perfect play. If a position results from a game with 49% accuracy, then that means the position does not result from perfect play.

Any number you stick in front of "accuracy" here is garbage, because the calculations that derived them are flawed.

"Those evaluations change in a matter of weeks and sometimes days, with each new release."
++ Yes, but that does not matter. 99% today may be 100% or 96% tomorrow, and 49% today may be 42% or 62% tomorrow, but never 100% or 99%.
Besides the > 50% criterium is only used to define sensible positions to estimate the time needed to weakly solve chess, not in the actual solving.

If you eliminate a single position from evaluation based on your fuzzy criteria, your solution fails on the spot.

"These arguments are no better now" ++ Those are solid arguments based on facts and figures, unlike your own ridiculous 'million years' based on erroneous toilet paper scribbling.

Lol.  No, my calculations on current supercomputer capabilities and what it would take using current technology to solve chess are dozens of orders of magnitude more accurate than your assessment.

"nobody is going to put up the money" ++ Maybe. Humans have walked on the Moon. Unmanned vehicles have driven on Mars. Humans can walk on Mars. The only limit is money.

A pointless argument.  Do you feel that FTL travel is also therefore just a matter of money?  Because that leap is closer to what your premise is.  There's no "walking on the moon" steps in terms in terms of solving chess.

 

stancco
btickler wrote:
stancco wrote:

Goos one 🤣🤣🤣

we are witnesses of a slack jaw dimness here

🤣🤣🤣

Did you know that your username means "tired out" in Italian?

Yes, I'm aware of that, I mean tired.

I'm just across Italian coast 👋

stancco
Optimissed wrote:
stancco wrote:
btickler wrote:
stancco wrote:

Goos one 🤣🤣🤣

we are witnesses of a slack jaw dimness here

🤣🤣🤣

Did you know that your username means "tired out" in Italian?

Yes, I'm aware of that, I mean tired.

I'm just across Italian coast 👋

Whereabouts are you?

Split, Croatia

Io sono Stanco, to not be confused, I mean my name is Stanko.

DiogenesDue
stancco wrote:

Split, Croatia

Io sono Stanco, to not be confused, I mean my name is Stanko.

Piacere.  Perhaps hold off on assumptions of dimwittedness.  I know it might be difficult with the example behavior displayed by some.

DiogenesDue
Elroch wrote:

Even if every one of a billion games between the top players were draws, this would not prove chess to be a draw. The reason is that it is at least conceivable that there is a very narrow winning strategy that requires dozens of unique moves in every line and the players are just too weak to find them all (like current top engines are too weak to accurately play complex tablebase positions without help).

Not just conceivable, more and more likely as tablebase construction continues to find longer and longer forced mates.

If, at some point, tablebases run into a wall where there are no more forced mates or great length being found for a significant period (say from 9 piece tablebases all the way to 10 piece), *then* you could make an argument that chess is a forced draw.  It would not be proven, but it would be far more robust than the assertions of "knowledge" being made here by a couple of microbes standing on a dust mote whose field of vision is 10^40+ too short-sighted.

stancco

I consider you lucky exploring Yugoslavia in 70's!

Unfortunately I don't know mr. Stanković.

I was born in 81.

However, I do know many (G)Masters of chess from Croatia and Bosnia&Hercegovina personally (not to mention their names here).

Glad to hear you, a hail of mine to you. 👋

 

haiaku
Optimissed wrote:
haiaku wrote:

The core point is that without an exhaustive proof, there is no doubt that no scientific theory can be guaranteed to hold true in any possible case. Galilean relativity, Newton's law of gravitation and classical mechanics are golden examples. They were thought to be always true, "unbeatable" so to speak, and they were consistent with centuries of experiments. There was no evidence that they might produce quite inaccurate predictions. We all know how it went: they fail miserably under some circumstances.

Not at all. They produce perfectly accurate results that are suitable for the environment in which they were conceived.

😦? They were thought to be universal and they are not.

Optimissed wrote:

You seem to be wanting to introduce a fallacy ... that of assuming that there are mysterious circumstances that alter the characteristics of chess [ . . . ] The rules do not allow for weird, relativistic effects within chess, so you're wrong again.

Is this a straw man? wink What I and others are saying is that there might be yet unknown lines which could disprove the assumption that chess is a draw. While we could safely ignore that Newton's law of gravitation is very incorrect under some circumstances and happily live with that (GPS satellites would not work properly, but...), for a solution a single unexpected line can make all the difference.

Optimissed wrote:
haiaku wrote:

In game theory, "optimal" is not a casual attribute. For chess, it means that an optimal player would be unbeatable in a match with an even number of games: if the game value is a draw, the optimal player would at least draw every game; if it's a win for either colour, the optimal player would always force the win with that colour. Therefore, the optimal player cannot score less than 50% of the points. Without a mathematical, exhaustive proof, a player cannot be guaranteed to be optimal, exactly like a scientific theory, without an exhaustive proof, cannot be guaranteed to always hold true.

This is confused. The first part is unnecessary. At least we know what we mean by "best play" or perhaps, "optimal", because we've been talking about it, in various threads, for about four years now and we did reach a consensus. Off the top of my head, optimal play is that which doesn't alter the game result negatively, for the player who made that move. Some like to say "from the game-theoretic value" but that's unnecessary, because it doesn't add anything useful. Just an illusion of grandeur.

What about the second part?

Optimissed wrote:
haiaku wrote:

This is not what people expect, when they read something like "the game xyz has been solved". They think about a definitive solution, that nobody will be able to disprove, ever.

You could never be sure that there wasn't a mistake in the analysis, due to a glitch of some unexpected kind. So wrong again, I'm afraid.

That is a point already made by @MARattigan. In fact, I said that people do expect a definitive solution, not that they can be 100% sure that a computer-assisted proof is correct. But the point is: is a statement like "chess is a draw because... [unproven motivations]" as acceptable and reliable as a statement like "a computer-assisted proof by exhaustion established that chess is a [draw, win]"?

stancco

Great experiences. I admire you, indeed. Wish I was there!

When it comes to Russians btw, they are well known for ignoring their enemies 😁

tygxc

#3864

"From what I've heard not every game at the highest level is a draw."
++ That is right. It is human to err.
Even in ICCF correspondence they sometimes mix up the move order and blunder a piece.

"if almost all chess games are draws that means chess is a draw"
++ No, the higher the level, the more draws. For every win we can pinpoint the error, usually the last move. For the high level draws we cannot pinpoint any error.

"If chess has been proven to be a draw, why are there wins?"
++ Because ultra-weakly solved is not weakly solved. Chess is known to be a draw, but not yet known how. I know for sure 1 e4 e5 2 Ba6 is a black win. Black is up a bishop. If you believe 1 tempo is convertible to a win, then you surely accept 1 bishop is convertible to a win even more. However I do not have a complete game tree up to checkmate. I believe with enough time I can win that as black against Carlsen or against an engine, but I might err at some move.

"There must be some reason not everyone agrees."
++ All stong players agree. Not everyone agrees on scientifically proven facts either.

"I could just as easily say because white has a first move advantage, AND the percentage of white wins at the highest level goes up, that's proof chess is a forced win for white."
++ The win % goes down with level. Chess is a draw, but the path to the draw is wider for white and narrower for black. So black has more ways to err than white.
White can afford to lose a tempo, then white just becomes black. Even black can afford to lose a tempo and white can afford to lose 2 tempi. Black cannot afford to lose 2 tempi: then he gets 1 pawn equivalent behind and that is enough to win.

"Until all games are draws it's evidence, not proof."
++ Yes, it is evidence and not a formal proof. Checkers has been solved to be draw. Checkers is still played and not all Checkers games end in draws. There is still human error especially at the lower level. The proof tree for Checkers has 10^7 positions. Nobody has these memorised.