Chess will never be solved, here's why

#3867

"Do you ask a person riding a roller coaster to design roller coasters?"
++ Weakly solving chess is not designing, it is using existing software on existing hardware.
The best person in the world to ask about that was GM Sveshnikov. Others are Kasparov, Kramnik, Carlsen, Karjakin, Caruana, Nepo, Dokhoian, Kazimdzhanov, or any ICCF grandmaster.

"GMs have made statements about draws, and when pressed they always hedge their bets."
++ Nope, see #3854 and #3856.

"Any number you stick in front of "accuracy" here is garbage, because the calculations that derived them are flawed." ++ That is besides the point. The accuracy cannot determine if play is perfect, but it can tell play is not perfect. It is like a modulo 3 primality test: it can determine that a number is not a prime, but it cannot determine that said number is a prime.

"If you eliminate a single position from evaluation based on your fuzzy criteria, your solution fails on the spot."
++ You still do not get it. The accuracy is not used in weakly solving chess. It is only used in defining sensible positions so as to assess the number of sensible positions that intervene.

"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"
++ You still do not understand the difference between weakly solving and strongly solving.

#3883

"I'd much rather ask a Digital Intelligence expert." ++ No not at all. Top grandmasters, their seconds, and ICCF grandmasters know most about chess and chess analysis.

"a weak solution is an overall verdict on what has been called the game- theoretical value."
++ No. You still do not get it.
Ultra-weakly solved means that the game-theoretic value of the initial position has been determined.  In layman's terms: it means a formal proof that chess is a draw.

Weakly solved means that for the initial position a strategy has been determined to achieve the game-theoretic value against any opposition.
In layman's terms: it means that a way to draw for black has been found against all reasonable white moves. That would need to visit 10^17 positions, can be done in 5 years.

"strongly solved is being used for a game for which such a strategy has been determined for all legal positions." In layman's terms: a 32-piece table base. That is all 10^44 legal positions, beyond the capability of present engines.

"That can't be obtained without a full solution of all possible and relevant games, each explored to the point where it's obvious what the result will be." ++ A solution tree of 10^17 positions would lead to a proof tree of about a billion positions, i.e. about 10 million perfect games.

"It would probably be impossible to store all these results" ++ No, 10 million perfect games are not that much more than existing data bases holding millions of games.

"btickler's calculations will be the more accurate" ++ No, he has no clue. He still does not understand the difference between weakly solving and strongly solving.

I apparently understand the definitions better than you do.  Weakly solving means solving against all moves from the initial position, not just "reasonable" moves.  Trying to change the definition of weakly solved to fudge your numbers doesn't help your case, it just makes you look desperate enough to mislead people willfully...

Ultra-weak

Prove whether the first player will win, lose or draw from the initial position, given perfect play on both sides. This can be a non-constructive proof (possibly involving a strategy-stealing argument) that need not actually determine any moves of the perfect play.

Weak

Provide an algorithm that secures a win for one player, or a draw for either, against any possible moves by the opponent, from the beginning of the game.

Strong

Provide an algorithm that can produce perfect moves from any position, even if mistakes have already been made on one or both sides.

Your premise is a bastardized dilution of weakly solved.

#3886

"I apparently understand the definitions better than you do."
++ Your toilet paper calculation shows otherwise.

"Weakly solving means solving against all moves from the initial position"
++ Cutting out unreasonable moves based on knowledge is allowed per van den Herik.
I know 1 e4 e5 2 Ba6 loses for white, so I do not need to calculate to checkmate.
I know 1 a4 cannot be better than 1 e4 or 1 d4, so I do not need to calculate 1 a4 to a draw.

"Trying to change the definition"
++ I do not change the definition. I replicate the definition verbatim.
I just explain in layman's terms as some people complain they do not understand jargon.

"Weak Provide an algorithm that secures a win for one player, or a draw for either,
against any possible moves by the opponent, from the beginning of the game."
++ It is not an algorithm, but a strategy.
Such a strategy can entail a proof tree, but also a set of rules, or a combination.
Allen has weakly solved Connect Four by brute force
and Allis has independently weakly solved it by a set of 7 rules. 

"Strong Provide an algorithm that can produce perfect moves from any position,
even if mistakes have already been made on one or both sides."
++ Not an algorithm, but a strategy for all legal positions.
Not only after one or more mistakes have been made, but also alternative drawing paths after one drawing strategy has been found. If 1 e4 e5 is proven a draw, then for weakly solving it does not matter if 1 e4 c5 draws as well or not, but for strongly solving that is needed too.
The essence is that weakly solving needs to visit far less positions than strongly solving.
Weakly solving Losing Chess required 900 million positions, not 10^44.

#3895

"I completely agree with btickler that tygxc has drastically underestimated the number"
++ So you do not understand the difference between weakly and strongly solving either.
10^44 is the number of legal positions for solving strongly.
Weakly solving requires far less, about 10^17.
Losing Chess has been weakly solved using 10^9 positions, not 10^44.

"produce a perfect chess engine which makes no mistakes"
++ No, that is not weakly solving.
Weakly solved means that for the initial position a strategy has been determined to achieve the game-theoretic value against any opposition.
'a strategy' ++ can be a proof tree or a set of rules or a combination of both
"the game-theoretic value" ++ a draw
"against any opposition" ++ white tries to win, black tries to draw, white fails, black succeeds

"points of imbalance and recrystallisation" ++ mumbo jumbo

#3989

"I think that the terminology is complete nonsense" ++ I am sorry, that is the scientific terminology in the game theory field. I explained in layman's terms for your convenience.

"we disagree, therefore you do not understand" ++ People disagree because they do not understand. They should read and think before disagreeing. People are better at slinging insults and accusations than at reading and understanding.

"reading yours and btickler's correspondence with each other, it's apparent that you're talking past each other"  ++ btickler does not understand the difference between weakly and strongly solving, that is why about weakly solving he erroneously uses the number for strongly solving.

"the terminology, which is not fit for purpose. Its purpose should be to convey meaning."
++ The terminology is fit for purpose, it applies to any game, not just chess, but it is intended for scientific readers. That is why I have added an explanation in layman's terms and specific to chess for your convenience. 

You've already written a criticism of the terms long ago.  It got the attention it deserved then, as well.  Take the hint.  Everyone gets that you don't like the nomenclature chosen.  It's not the greatest.  It is sound, however, and I am sure there is some evolutionary reason for the clumsiness of the terms for the average reader.

I just understand that all your shortcuts require determinations to reduce the 10^44 positions that are going to be roughly equivalent to just traversing the positions, say plus or minus a handful of orders of magnitude at best.

You don't seem to get this.  Your arbitrary reductions will require a massive amount of computation to apply your arbitrary criteria to each position simply to eliminate it from consideration.  Worse, at the end, your "good enough" shortcuts will produce a flawed answer that will not be accepted as a solution anyway.  Like building a house out of 2x4s and then staple-gunning cardboard for walls, then trying to sell it for a million dollars.

You can produce *something* in 5 years...but not a solution for chess.  More like a watered down attempt at a forward moving tablebase to meet the retrograde analysis in the middle.  The forward moving attempt will summarily eliminate valid positions in order to get around the fact that you are not working backwards from mate, which by force culls all of the invalid positions for you.  Not so for your ill-conceived plan.  What you propose would make engines stronger players, but the results would not even come close to being a solution for chess.

You don't have to believe me.  The proof is and will continue to be in the lack of anyone who knows how to build such systems paying the slightest attention to your plan.

I think the more I read this thread the more obvious the reason chess will never be solved is people cant even agree on what we want "solved" to mean XD.  Though I'm sure each and everyone can tell you exactly why how they feel about it is "the correct way".  The older I get the more "truth" seems to be as malleable as "morality". Fun thread to read though lots of nested philosophical ideas and such applicable to other arguments. When people cant even agree on operable definitions for the words used in the arguments you know its finna be a good time XD. 

We need Speedtalk.

What is that supposed to mean? The lines that no one has explored down to the tablebases yet, are unkown elements. You are excluding a priori that one, any one, of those lines may falsify the assumption that the game value is a draw.

"Equalising tendency" is too ambiguous. We could only say: 
1) S is a set of positions known to be draws;  
2) if both players play optimal moves from the inital position, it is possible to reach only positions belonging to S;
3) therefore, the initial position is a draw too. 
But then the question would be: how do we know that the positions in S are draws? How can we prove statement 2?

Where do you read "computer-assisted" in my statement? The point is: how do you know that a statement holds true in any possible case, without an exhaustive proof that the statment holds true in any possible case? Because of a tendency? It would be inductive reasoning, overgeneralization. It is good for hypotheses, but not for conclusions.

But that's exactly the argument you are trying to refute: conceiving the "unconceivable".

You and tygxc are about absolute knowledge. To me the only real proof is an exhaustive one, but it might be impossible to achieve that in practice, not only in case of computer-assisted proofs. If one million mathematicians do all agree that T is a theorem, they might all be mistaken. If we all see the sun rise, we might all be hallucinated.
Seriously, true scientists are skeptical in nature: they are aware that they can be biased and that what may seem universal can be in fact just a special case; it's already happened. What makes the difference between scientists and non-scientists is the level of accuracy, universality and stability of the conclusions, they seek. This make them use mathematical models of course, they try to falsify statements, they reproduce experiments, seek agreement with other scientists, etc.
A statement like "chess is a draw because of the equalizing tendency" really cannot be considered scientific, or nearly as reliable as a computer-assisted proof by exhaustion.

The King of Projection speaks.

The fact that you can never seem to muster anything but crude insults tells the real story.  At every reply, I respond with civil discourse and observation, and you end up turning to one word insults, being unable to operate at the same level.  It's a tired refrain.

 "you think that's air your breathing right now?" Suffice to say I don't think we are gonna agree on what tasty wheat tastes like.  Whether we point to god or science as an authority people can still choose not to subscribe to either.  Still a fun thought experiment 

Sorry for what I didn't take offense to anything you have said?  I do think maybe its not worth either of your time with the back and forth can have some fun semantical debates without getting upset with one another I think. 

Maybe you need to read your own posts.  But of course, in Opti-land, saying that nobody in the thread is anywhere near your mental ability is just a statement of fact and not an insult to everybody here, right?  So when you review your behavior, it's all statements of "facts" by you about being superior, and ergo insults only come from other quarters.  Is that about the size of it?

You are the bull in the china shop.

Delusional thinking.  It's your defining characteristic.

"IQ"?

Isn't that something kids get measured to give a hint of their potential to develop other skills that are more substantial than doing little puzzles?

Also something narcissists tend to exaggerate.

The quoted text is more substantial than the comment. It points out that there is a proof that an optimum strategy exists for games in a class including chess (I think the full description is finite, deterministic games of perfect information, all defined in texts. They left determinism implicit), and refers to the established name of the theorem. It also draws attention to the fact that there are three logical possibilities and what they are.

Note "perfect information" and "deterministic", as conventionally defined, are independent. While poker is neither, other games can be one but not the other. You could transform a game where the state was visible but the result of actions uncertain to one where the state was uncertain and the uncertain part of the state was what altered the effect of actions, so maybe they are being cleverer than me and realising you can do without the additional category.

What's amazing is that the 10^120 fact that everyone throws around doesn't even scratch the surface of every possible chess position... The famous 10^120 positions assumes a game that goes on for exactly 40 moves. Considering the fact that with the 50 moves draw rule the actual longest possible chess game is several thousand moves, you can only imagine how many orders of magnitude higher the actual number of chess positions are. Maybe something like 10^200 which is a number almost impossible to imagine

#3922
"Maybe something like 10^200 which is a number almost impossible to imagine"
++ Please do not produce such nonsense.
It has been proven there are 10^44 legal chess positions
https://github.com/tromp/ChessPositionRanking

It is obvious the three sample positions make no sense.
A better estimate thus is 10^37
https://arxiv.org/pdf/2112.09386.pdf

The number needed to weakly solving chess is much smaller.
Losing Chess has been solved using 10^9 positions.
http://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/LCsolved.pdf