Your network blocks the Lichess assets!

lichess.org
Donate

Can Chess be Solved?

Do we really need to solve chess? When an engine is down in material, should it not resign against perfect play? no because it still aims for a draw. That's two different outcomes. The game of chess does not just have one goal, but two. The second is the backup plan of drawing. So why solved the backup plan and the main plan. Aim to solve for one of them first. Maybe material imbalances need relative values, like we gave a pawn a value of one. Everything is relative to the position and the time to play. So solve for 1 second games. Solve for 2 second games. Solve the game by number of exchanges. Start with no exchanges.

Do we really need to solve chess? When an engine is down in material, should it not resign against perfect play? no because it still aims for a draw. That's two different outcomes. The game of chess does not just have one goal, but two. The second is the backup plan of drawing. So why solved the backup plan and the main plan. Aim to solve for one of them first. Maybe material imbalances need relative values, like we gave a pawn a value of one. Everything is relative to the position and the time to play. So solve for 1 second games. Solve for 2 second games. Solve the game by number of exchanges. Start with no exchanges.

I mean, technically if humans can survive without sleep, and had very, very, very big brains, then yes, chess could be solved

I mean, technically if humans can survive without sleep, and had very, very, very big brains, then yes, chess could be solved

@Toadofsky said in #10:

Define chess. That can be part of such concise question thread content, in a forum in my world view of what a forum might be... :) (long ago we had some exchanges about that you and me, I was a dreamer in that dialog).

So I guess I was trying to work on that. If betting on some consensus of the reality of that word or some absolute common-sense or eternal meaning. Then I would say no. and I don't really think this is worth the prowess...

I find dissecting the question into its multiple possibilities more debatable and stimulating.

I also think that even if solved (weakly or strongly?) that would not change a thing to the experience of chess in single human life trajectories. same problem as move-line chess strategy of learning. The side lines or the bush of it won't get any smaller as learning problem at individual level. Same brain biology.... or would we chip in the solution? (wink wink).

Curtains open:
"oh this is not the solved line I have prepared for, what an annoying opponent"

@Toadofsky said in #10: > Define chess. That can be part of such concise question thread content, in a forum in my world view of what a forum might be... :) (long ago we had some exchanges about that you and me, I was a dreamer in that dialog). So I guess I was trying to work on that. If betting on some consensus of the reality of that word or some absolute common-sense or eternal meaning. Then I would say no. and I don't really think this is worth the prowess... I find dissecting the question into its multiple possibilities more debatable and stimulating. I also think that even if solved (weakly or strongly?) that would not change a thing to the experience of chess in single human life trajectories. same problem as move-line chess strategy of learning. The side lines or the bush of it won't get any smaller as learning problem at individual level. Same brain biology.... or would we chip in the solution? (wink wink). Curtains open: "oh this is not the solved line I have prepared for, what an annoying opponent"

Thanks for the publication. I am very fond of Antichess now. Is there any robust study about solving antichess by brute force?

Thanks for the publication. I am very fond of Antichess now. Is there any robust study about solving antichess by brute force?

The question is never "can", it's always "how". At this point in time it's not practical [or practicable ; ) ], but chess can be solved as it is finite (limited).

The question is never "can", it's always "how". At this point in time it's not practical [or practicable ; ) ], but chess can be solved as it is finite (limited).

are there magnitudes of finiteness.. numbers so big that our small brains would rather use the notion of inifinity to make some sense of the mass oif information. Unless having a machine that can recreate a universe order of magnitude cases... but then would it be solved if we can,t even read the solution set? assuming instantatenous priniting.

it would be only another same size computers that could deal with that big printout. Is the solution existing if no one can grasp it?

I think tic tac toe finite and chess finite.. are not the same degree of problem. I think approximation by human brain of the problem, would make that clear. Tic tac toe finite.. fits in. chess finite does not fit in... I think finite is not a good measure of the problem complexity driving variable. Not even the tree explosion number of nodes... We need tools about position classes.

patterns. that can be clustered. where we count patterns as volume of such clusters... that our wet brain are likely to easily solve by similarity. That notion of similarity is absent from the counting measure that finite is restricting the question of "solution" to.

are there magnitudes of finiteness.. numbers so big that our small brains would rather use the notion of inifinity to make some sense of the mass oif information. Unless having a machine that can recreate a universe order of magnitude cases... but then would it be solved if we can,t even read the solution set? assuming instantatenous priniting. it would be only another same size computers that could deal with that big printout. Is the solution existing if no one can grasp it? I think tic tac toe finite and chess finite.. are not the same degree of problem. I think approximation by human brain of the problem, would make that clear. Tic tac toe finite.. fits in. chess finite does not fit in... I think finite is not a good measure of the problem complexity driving variable. Not even the tree explosion number of nodes... We need tools about position classes. patterns. that can be clustered. where we count patterns as volume of such clusters... that our wet brain are likely to easily solve by similarity. That notion of similarity is absent from the counting measure that finite is restricting the question of "solution" to.

When we think about mapping the world, we often consider roads and paths. Similarly, in chess, we have main lines and strategies that can be mapped. However, just as not all roads are navigable, not all chess paths lead to a win. Chess engines have solved certain endgames with limited pieces, but extending this to the entire game is a different story. They use heuristics and approximations to play effectively without fully solving the game. The ECO book is useful, but it does not mean the opening phase is solved. With the assistance of engines we can now use them to help us step forward in chess. An entropy chess positioning system (ECPS) is required to solve chess. We also need to update the pgn format with a standard like: {[%eval -0.4][%legal (20 20)]}. The values "(20 20)" is the entropy moves available for white and black. In the opening we have 20 legal moves each. https://lichess.org/study/DSF0P9Nv/zkEWPYLc

https://supercomputingchallenge.org/21-22/finalreports/45/team45-final-report.pdf

The universe also has patterns (like the orbits of planets and galaxies) that help us understand its complexity. In chess, identifying strategic patterns is key. Instead of trying to "solve" chess, we should focus on mapping its main winning lines with entropy. This would involve creating a dynamic map that updates over time, similar to how our understanding of the universe evolves. A map of chess's main lines could be incredibly valuable, but it requires ongoing effort to keep it relevant.

Ultimately, the goal should be practical navigation rather than complete solution. Focusing on finding winning strategies and updating our knowledge over time can help us build a better understanding of chess. This approach is similar to how astronomers search for exoplanets using tools like the Hubble telescope. By focusing on achievable goals and using technology to guide us, we can make progress in both chess and our exploration of the universe.

When we think about mapping the world, we often consider roads and paths. Similarly, in chess, we have main lines and strategies that can be mapped. However, just as not all roads are navigable, not all chess paths lead to a win. Chess engines have solved certain endgames with limited pieces, but extending this to the entire game is a different story. They use heuristics and approximations to play effectively without fully solving the game. The ECO book is useful, but it does not mean the opening phase is solved. With the assistance of engines we can now use them to help us step forward in chess. An entropy chess positioning system (ECPS) is required to solve chess. We also need to update the pgn format with a standard like: {[%eval -0.4][%legal (20 20)]}. The values "(20 20)" is the entropy moves available for white and black. In the opening we have 20 legal moves each. https://lichess.org/study/DSF0P9Nv/zkEWPYLc https://supercomputingchallenge.org/21-22/finalreports/45/team45-final-report.pdf The universe also has patterns (like the orbits of planets and galaxies) that help us understand its complexity. In chess, identifying strategic patterns is key. Instead of trying to "solve" chess, we should focus on mapping its main winning lines with entropy. This would involve creating a dynamic map that updates over time, similar to how our understanding of the universe evolves. A map of chess's main lines could be incredibly valuable, but it requires ongoing effort to keep it relevant. Ultimately, the goal should be practical navigation rather than complete solution. Focusing on finding winning strategies and updating our knowledge over time can help us build a better understanding of chess. This approach is similar to how astronomers search for exoplanets using tools like the Hubble telescope. By focusing on achievable goals and using technology to guide us, we can make progress in both chess and our exploration of the universe.

map of lines, is not explicit enough, rather map of positions as internal representation model.

I prefer position world to move chess world, it has more board information to navigate a map about position as internal representation might be more feasible for me than a map or evolving map of a position world.

It might just be me. And rats or bats. That does not mean that the moves are not there, but that they have an ambient space where the map is better as "ground" substrate to put pins....

I should mention that I am not talking about 2D map where the things are moving, the board. but a bigger internal space of positions where the positions are points. Since this is too big for both our retinas and our verbal language to mere point to, using words, we need other tools. Also, we are not really conscious of those things, or not to the point where we have common pointers to talk about them. we need bigger tools. And there already exist such tools in the scientific culture. Just not making it to chess for some reason. It is not within the current language of it. (or linguistic constrained psychology).

So it is speculation. But that is not forbidden, that is how one can build theories to make sense of the big finite or infinite world... One chunk of hypothesis at a time.. but not looking out, is a sure loop around promise.

map of lines, is not explicit enough, rather map of positions as internal representation model. I prefer position world to move chess world, it has more board information to navigate a map about position as internal representation might be more feasible for me than a map or evolving map of a position world. It might just be me. And rats or bats. That does not mean that the moves are not there, but that they have an ambient space where the map is better as "ground" substrate to put pins.... I should mention that I am not talking about 2D map where the things are moving, the board. but a bigger internal space of positions where the positions are points. Since this is too big for both our retinas and our verbal language to mere point to, using words, we need other tools. Also, we are not really conscious of those things, or not to the point where we have common pointers to talk about them. we need bigger tools. And there already exist such tools in the scientific culture. Just not making it to chess for some reason. It is not within the current language of it. (or linguistic constrained psychology). So it is speculation. But that is not forbidden, that is how one can build theories to make sense of the big finite or infinite world... One chunk of hypothesis at a time.. but not looking out, is a sure loop around promise.

When I created my version of the legal move annotation {[%legal (20 20)]}, I also tested it out on chess gui's and AI's. The AI's recognised it's use, but obviously Chess gui's did not, but it did not interfere. Maybe one day chess gui's or a pgn-spec will adopt the feature.

https://github.com/mliebelt/pgn-spec-commented

We still have decades to go before being able to map a chess solar system, because there are many different "solar systems" or variations of the game. Chess to be solved needs to be solve one part at a time like with sets of opening lines, middlegame strategies, and endgame techniques. Solving one tiny part of any of that needs to be logged on a checklist in a github, before claiming chess to be solvable. We still have not even made the checklist. So chess is far from being ready to be solved. Each chess variant is in a galaxy of it's own.

https://en.wikipedia.org/wiki/Portable_Game_Notation#Numeric_Annotation_Glyphs

When I created my version of the legal move annotation {[%legal (20 20)]}, I also tested it out on chess gui's and AI's. The AI's recognised it's use, but obviously Chess gui's did not, but it did not interfere. Maybe one day chess gui's or a pgn-spec will adopt the feature. https://github.com/mliebelt/pgn-spec-commented We still have decades to go before being able to map a chess solar system, because there are many different "solar systems" or variations of the game. Chess to be solved needs to be solve one part at a time like with sets of opening lines, middlegame strategies, and endgame techniques. Solving one tiny part of any of that needs to be logged on a checklist in a github, before claiming chess to be solvable. We still have not even made the checklist. So chess is far from being ready to be solved. Each chess variant is in a galaxy of it's own. https://en.wikipedia.org/wiki/Portable_Game_Notation#Numeric_Annotation_Glyphs

What is the ambient space equivalent for chess in your solar system analogy? Where things don't have to be, necessarily. As empty space between the things or positions in single game trajectories? Word or PGN syntax categories won't deal with that. And PGN alread allows comments for the "move" SAN syllables.

PGN is the mininal string necessary to recompute a position deep in any game or the whole game on step at a time but always having to crank from a FEN before one has the explicit position complete information. One is stuck with the prefiix all the time, to have position complete information (or is it perfect information, in another perspective, for me it is complete information like an EPD or a FEN, not a single piece change of location, a SAN move, complete in natural language common sense, yes, I go there, common sense... sigh).

I find that talking about notation standard is jumping the gun. We need a mathematic space that is not biased (like engine models can still be) or ininital condition and historical path dependent (as chess cutulral knowledge so far is, as trajectory of knowledge over such putative big space).
Words here would keep me loopig around until we taclke that bull holding both horns.... this is where I stop.

This is likely more looping around because we are stuck in a too "small" set of tools. Technology local optimum trap.

What is the ambient space equivalent for chess in your solar system analogy? Where things don't have to be, necessarily. As empty space between the things or positions in single game trajectories? Word or PGN syntax categories won't deal with that. And PGN alread allows comments for the "move" SAN syllables. PGN is the mininal string necessary to recompute a position deep in any game or the whole game on step at a time but always having to crank from a FEN before one has the explicit position complete information. One is stuck with the prefiix all the time, to have position complete information (or is it perfect information, in another perspective, for me it is complete information like an EPD or a FEN, not a single piece change of location, a SAN move, complete in natural language common sense, yes, I go there, common sense... sigh). I find that talking about notation standard is jumping the gun. We need a mathematic space that is not biased (like engine models can still be) or ininital condition and historical path dependent (as chess cutulral knowledge so far is, as trajectory of knowledge over such putative big space). Words here would keep me loopig around until we taclke that bull holding both horns.... this is where I stop. This is likely more looping around because we are stuck in a too "small" set of tools. Technology local optimum trap.