In a different sense. You've spoken of weak and strong solutions, which are variations.
will chess ever die out once its fully solved?
Sort:
A weak solution, incidentally, would be one that contained every variation, because there's no need for that whatsoever. A strong solution would be one where there's an effective, algorithmic filter which removes irrelevant lines from consideration. So, you see, words can have opposite implications depending on the paradigm. I was referring to strong and weak solutions being variations of types of solution. You propose that strong solutions contain every possible line and I prefer the opposite, because not only is there no need for it but it immeasurably slows the analysis for zero gain.
Jan 7, 2022
0
#744
Optimissed wrote:
The number of possible games is irrelevant to an intelligent conversation regarding "solving chess" but so is the total number of possible positions. The only relevancy is the number of relevant positions, which looks a bit circular ..... but a position that is relevant to a solution is one that is reached by a series of moves which doesn't contain any obvious blunders. And, of course, that cuts out an enormous majority of positions from consideration. What's more, another large majority can be removed by cutting out all those which are reached by a series of moves which don't make any "chess sense".
So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
In reality, before chess can be properly "solved", it's necessary to analyse it algorithmically to mathematically express (depict) positions that are meaningful in a chess sense rather than in merely a random/legal sense. This is in itself a big job but it would provide a filter that is necessary if meaningful (relevant) positions are to be automatically identified.
This may be a bit difficult to follow, I don't know. But it's how things are.
I like this comment, but I disagree with this part : So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
I think its quite possible a neural network AI would destroy even a grandmaster using the opening moves 1.a3 2.b3 and probably even 3.c3 depending on what black plays. From this perspective it is the strength of the entity that determines the relevance of the position. Similarly, any nonsense looking opening position might be relevant to beginners. The problem seem to be the subjective determination of relevance rather than the frequency with which that position might be reached. There may be any number of positions that have only been reached once that are nevertheless significant. In my opinion that's just one of things that makes the game very interesting. I agree, however, that the mathematical determination of possible branches, positions, etc. is irrelevant to the art or enjoyment of the game. The question itself is a math problem, which might be interesting to a mathematician. I have an old problem book on combinatorics which contains a variety of chess puzzles, but the whole question is one of PROOF. So perhaps our mathematically inclined friends might be interested in stating their ideas in the form of proof, which is of course much better than simply declaring oneself correct because the answer is on the internet.
Jan 7, 2022
0
#745
Optimissed wrote:
A weak solution, incidentally, would be one that contained every variation, because there's no need for that whatsoever. A strong solution would be one where there's an effective, algorithmic filter which removes irrelevant lines from consideration. So, you see, words can have opposite implications depending on the paradigm. I was referring to strong and weak solutions being variations of types of solution. You propose that strong solutions contain every possible line and I prefer the opposite, because not only is there no need for it but it immeasurably slows the analysis for zero gain.
You said that before. I wasn't sure what you meant. Now you've clarified, I suppose it's fine, but that isn't the meaning commonly used. The strong form has more information in it--that's what's commonly meant by the people working on it.
Jan 7, 2022
0
#746
Euclid5050 wrote:
Optimissed wrote:
The number of possible games is irrelevant to an intelligent conversation regarding "solving chess" but so is the total number of possible positions. The only relevancy is the number of relevant positions, which looks a bit circular ..... but a position that is relevant to a solution is one that is reached by a series of moves which doesn't contain any obvious blunders. And, of course, that cuts out an enormous majority of positions from consideration. What's more, another large majority can be removed by cutting out all those which are reached by a series of moves which don't make any "chess sense".
So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
In reality, before chess can be properly "solved", it's necessary to analyse it algorithmically to mathematically express (depict) positions that are meaningful in a chess sense rather than in merely a random/legal sense. This is in itself a big job but it would provide a filter that is necessary if meaningful (relevant) positions are to be automatically identified.
This may be a bit difficult to follow, I don't know. But it's how things are.
I like this comment, but I disagree with this part : So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
I think its quite possible a neural network AI would destroy even a grandmaster using the opening moves 1.a3 2.b3 and probably even 3.c3 depending on what black plays. From this perspective it is the strength of the entity that determines the relevance of the position. Similarly, any nonsense looking opening position might be relevant to beginners. The problem seem to be the subjective determination of relevance rather than the frequency with which that position might be reached. There may be any number of positions that have only been reached once that are nevertheless significant. In my opinion that's just one of things that makes the game very interesting. I agree, however, that the mathematical determination of possible branches, positions, etc. is irrelevant to the art or enjoyment of the game. The question itself is a math problem, which might be interesting to a mathematician. I have an old problem book on combinatorics which contains a variety of chess puzzles, but the whole question is one of PROOF. So perhaps our mathematically inclined friends might be interested in stating their ideas in the form of proof, which is of course much better than simply declaring oneself correct because the answer is on the internet.
What would you like proved?
The solution to checkers (weak form) is available for review in published papers. Jonathan Shaeffer is the lead man in this.
The solution to chess has not been completed so no proof will be out there.
The expected size of the task, and the work in progress on it, is out there in very many papers which contain proofs as far as they go, along the lines "the problem is this size, etc." And mostly that's a matter of combinations. There are only 64 squares and only 32 pieces etc., is the starting point.
Edit: also, the other thing that is most obviously out there is tablebases, and work on those, which are effectively solutions of smaller versions of the problems (up to 7-man as of now, 8-man is about 100 times bigger).
Whether a task of this size can ever be solved is also unproven, and likely to remain so for a long time. The most convincing proof would be to solve it, but that's a long way off.
Beyond that, there is plenty of work in CS journals about the eventual end of Moore's law, due to components can't be smaller than a countable amount of atoms...so it's not solvable that way, but nobody is sure about what quantum computers can or can't do.
But a lot of work is going on, whether directed specifically at this or at more general problems. I'm sure the world be more immediately interested, for example, in a quantum computer for breaking crypto, but...it would have application for other things too.
Jan 7, 2022
0
#747
JubilationTCornpone wrote:
What would you like proved?
The solution to checkers (weak form) is available for review in published papers. Jonathan Shaeffer is the lead man in this.
The solution to chess has not been completed so no proof will be out there.
The expected size of the task, and the work in progress on it, is out there in very many papers which contain proofs as far as they go, along the lines "the problem is this size, etc." And mostly that's a matter of combinations. There are only 64 squares and only 32 pieces etc., is the starting point.
Edit: also, the other thing that is most obviously out there is tablebases, and work on those, which are effectively solutions of smaller versions of the problems (up to 7-man as of now, 8-man is about 100 times bigger).
Whether a task of this size can ever be solved is also unproven, and likely to remain so for a long time. The most convincing proof would be to solve it, but that's a long way off.
Beyond that, there is plenty of work in CS journals about the eventual end of Moore's law, due to components can't be smaller than a countable amount of atoms...so it's not solvable that way, but nobody is sure about what quantum computers can or can't do.
But a lot of work is going on, whether directed specifically at this or at more general problems. I'm sure the world be more immediately interested, for example, in a quantum computer for breaking crypto, but...it would have application for other things too.
I'm sorry, maybe I was confused. I'm not wanting any kind of proof. I was making a suggestion for the resolution to what appeared to me to be a dispute in your conversation regarding a mathematical fact that could be substantiated by the proof, if that proof exists. However, you say there is no proof on the central question (only estimates I see), therefore you couldn't possibly be having such a dispute, and it would appear that I've read your conversation wrong. I admit I only skimmed over it because I'm not looking at those kinds of questions right now. If I was remiss, please excuse my error. It's definitely an interesting problem.
Jan 7, 2022
0
#748
Euclid5050 wrote:
JubilationTCornpone wrote:
What would you like proved?
The solution to checkers (weak form) is available for review in published papers. Jonathan Shaeffer is the lead man in this.
The solution to chess has not been completed so no proof will be out there.
The expected size of the task, and the work in progress on it, is out there in very many papers which contain proofs as far as they go, along the lines "the problem is this size, etc." And mostly that's a matter of combinations. There are only 64 squares and only 32 pieces etc., is the starting point.
Edit: also, the other thing that is most obviously out there is tablebases, and work on those, which are effectively solutions of smaller versions of the problems (up to 7-man as of now, 8-man is about 100 times bigger).
Whether a task of this size can ever be solved is also unproven, and likely to remain so for a long time. The most convincing proof would be to solve it, but that's a long way off.
Beyond that, there is plenty of work in CS journals about the eventual end of Moore's law, due to components can't be smaller than a countable amount of atoms...so it's not solvable that way, but nobody is sure about what quantum computers can or can't do.
But a lot of work is going on, whether directed specifically at this or at more general problems. I'm sure the world be more immediately interested, for example, in a quantum computer for breaking crypto, but...it would have application for other things too.
I'm sorry, maybe I was confused. I'm not wanting any kind of proof. I was making a suggestion for the resolution to what appeared to me to be a dispute in your conversation regarding a mathematical fact that could be substantiated by the proof, if that proof exists. However, you say there is no proof on the central question (only estimates I see), therefore you couldn't possibly be having such a dispute, and it would appear that I've read your conversation wrong. I admit I only skimmed over it because I'm not looking at those kinds of questions right now. If I was remiss, please excuse my error. It's definitely an interesting problem.
Ah, no worries then. It's definitely an interesting issue. And it's definitely a good point to go look at what is out there. And definitely many people don't bother and then come along and say "well, I think this ..." And I am not even an expert...just a former undergrad who actually did read some papers.
On weakly and strongly solved games:
https://en.wikipedia.org/wiki/Solved_game
A strong solution of chess is not viable yet.
A weak solution is within reach of present cloud engines.
tygxc wrote:
On weakly and strongly solved games:
https://en.wikipedia.org/wiki/Solved_game
A strong solution of chess is not viable yet.
A weak solution is within reach of present cloud engines.
The latter assertion is purely an opinion based on some wildly optimistic and overreaching estimates. At present neither solution is actually possible.
#1007
It is all opinion until it is done, then it is fact.
At least I gave facts and figures in support.
Euclid5050 wrote:
Optimissed wrote:
The number of possible games is irrelevant to an intelligent conversation regarding "solving chess" but so is the total number of possible positions. The only relevancy is the number of relevant positions, which looks a bit circular ..... but a position that is relevant to a solution is one that is reached by a series of moves which doesn't contain any obvious blunders. And, of course, that cuts out an enormous majority of positions from consideration. What's more, another large majority can be removed by cutting out all those which are reached by a series of moves which don't make any "chess sense".
So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
In reality, before chess can be properly "solved", it's necessary to analyse it algorithmically to mathematically express (depict) positions that are meaningful in a chess sense rather than in merely a random/legal sense. This is in itself a big job but it would provide a filter that is necessary if meaningful (relevant) positions are to be automatically identified.
This may be a bit difficult to follow, I don't know. But it's how things are.
I like this comment, but I disagree with this part : So let's say white opens 1. a3 and then 2. b3 and 3. c3. If black were to reply by any similarly meaningless series of moves, all the positions reachable from such a series, except by transposition, are irrelevant, because one side isn't attempting to profit from the other side's weak moves. The idea that a6, b6 and c6 is black's best way to respond to a3, b3 and c3 isn't reasonable and can be discounted.
I think its quite possible a neural network AI would destroy even a grandmaster using the opening moves 1.a3 2.b3 and probably even 3.c3 depending on what black plays. From this perspective it is the strength of the entity that determines the relevance of the position. Similarly, any nonsense looking opening position might be relevant to beginners. The problem seem to be the subjective determination of relevance rather than the frequency with which that position might be reached. There may be any number of positions that have only been reached once that are nevertheless significant. In my opinion that's just one of things that makes the game very interesting. I agree, however, that the mathematical determination of possible branches, positions, etc. is irrelevant to the art or enjoyment of the game. The question itself is a math problem, which might be interesting to a mathematician. I have an old problem book on combinatorics which contains a variety of chess puzzles, but the whole question is one of PROOF. So perhaps our mathematically inclined friends might be interested in stating their ideas in the form of proof, which is of course much better than simply declaring oneself correct because the answer is on the internet.
I get what you're saying but my point is that if white played 1. a3, 2. b3, 3, c3, perhaps a very strong player would win by mirroring those moves BUT there are bound to be stronger moves for black, which are so much stronger that the mirroring sequence for black becomes irrelevant. If black gets a sufficiently good position from normal moves, all the lines following from the mirrored sequence are irrelevant because they don't reflect strongest play by black.
Incidentally, my son the mathematician disagrees with even the possibility that a series of equations might be written to depict accurately the game of chess. He considers it beyond the bounds of possibility for a human to write such equations. So I'm sticking my neck out but I think one day, they might conceivably be written by machine intelligence. Not in the next 100 years though.
JubilationTCornpone wrote:
Optimissed wrote:
A weak solution, incidentally, would be one that contained every variation, because there's no need for that whatsoever. A strong solution would be one where there's an effective, algorithmic filter which removes irrelevant lines from consideration. So, you see, words can have opposite implications depending on the paradigm. I was referring to strong and weak solutions being variations of types of solution. You propose that strong solutions contain every possible line and I prefer the opposite, because not only is there no need for it but it immeasurably slows the analysis for zero gain.
You said that before. I wasn't sure what you meant. Now you've clarified, I suppose it's fine, but that isn't the meaning commonly used. The strong form has more information in it--that's what's commonly meant by the people working on it.
Yes, I understand that but thanks for mentioning it. It's just that I don't see a strong form of anything containing irrelevant information. My background is philosophy. I've little doubt my son would disagree with me. His background's maths. @Caproni .... apparently, Caproni was a brilliant and under-recognised Italian mathematician.
#1010
Please point your mathematician son to the
https://en.wikipedia.org/wiki/Four_color_theorem
This was the first computer proof of a mathematical theorem.
Appel & Haken proved the theorem coloring not all, but 1482 relevant maps.
Likewise chess can be solved not by all 4*10^37 possible positions but only the 10^18 relevant.
^^^ thanks, I read about that only in the past year and found it very interesting. One of my hobbies is collecting old maps. I'm pretty sure he'll know about it. If you find him on linked-in, you'll see, but I'll take your advice next time I speak to him.
When that happens, with quantum computing, the competitive chess will die, but only for machines.
There are several BOT competitions worldwide, but they will die, because every match will end-up in a draw.
For humans, the control will have to be more strict, and online chess will have to improve their detection tools to avoid unfair advantages from people that use computers to play. But chess is fun and as long as people will keep on having fun, they will keep playing it.
That's my thoughts!
Chess can't be fully solved at present computing speeds.
i completely forgot i made this forum, memories but oh well wishing everyone here in 2023 a happy thanksgiving with their families cheers!
For practical purposes chess has been solved and is a draw. See here
https://www.iccf.com/event?id=100104
So far 54 perfect games without errors, all draws.
That does not stop competitive chess tournaments.
Top grandmasters admit in interviews that they have a hard time remembering their deep engine preparation.
It's a variation of a solution isn't it?
The solution is the whole graph, containing all possible positions, linked together to account for positions which lead to other positons, allowing for transpositions, and for cycles. A variation would be just one particular path through the solution. So, there are "strong" and "weak" solutions, depending on whether they contain every possibility, but variations do not have this same characteristic--they can't, and can't be expected, to have every posibility--they inherently are only one.