Chess will never be solved, here's why

Keep dreaming.

@4257

"Chess is finite."
++ Yes indeed: 10^44 legal positions of which 10^17 sensible, reachable, and relevant.

"No probability there, get it?"
++ Yes, there is no probablility: each position is either a draw, a win, or a loss.

"It has NO weak or strong solution."
++ It has. An ultra-weak solution is a formal proof that chess is a draw.
A weak solution shows how black can draw against all reasonable white tries to win and needs to consider all 10^17 legal, sensible, reachable, and relevant positions.
A strong solution is a 32-men table base and requires all 10^44 legal positions.

"These are all SOLUTIONS." ++ Yes

"All possible draws." ++ Yes

"All the lines in which your opponent with absolute knowledge still doesn't have the wining line." ++ Yes

"There are many of these lines whatsoever, I mean the draw lines."
++ There are much more lines than positions because of transpositions.

"Call it a domain of SOLUTIONS if you like."
++ It are paths to hop from one drawn position to the other until the 7-men endgame table base or a prior forced 3-fold repetition.

"The rest ARE NOT SOLUTIONS because at some point one side (or both sides) make(s) mistake(s) so that the forced win(s) is/are (was/were) possible."
++ Those are pitfalls from the safe path of drawn positions into the abyss of the lost positions.

"All this said, of course, is in a sense of availability of a forced mate aside the Fide foolish rules like 50moves or such."
++ The 50-moves rule is a practical rule to ensure that a chess competition is not delayed for one game lasting weeks. However the 50-moves rule can be ignored as it is nearly never invoked in practice before the 7-men table base of strongly solved positions is reached.
ICCF allows 7-men table base win claims that exceed 50 moves without capture or pawn move, but such claims never occur.

"And what is of the most relevance for this here thread is the probability that the chess already is - SOLVED."
++ We already have part of the solution. We already have a 1000 ICCF WC drawn games that are perfectly played. Top grandmasters Carlsen, Caruana, Nepo have prepared their world championship matches for months with teams of grandmasters and cloud engines and probably have solved at least large parts of the Petrov, Sveshnikov, and Marshall.

Nah, come on. You said that one cannot be 100% sure a position is winning unless they can beat stockfish in the position.

What I'm saying is that its possible to prove that a position is winning with other means than beating stockfish or any other robot in the position.

I'm sure we agree now that I laid it out like this, but your comment wasn't in line with it.

No, I agreed with Mar that it would be a start for Optimissed to make his point.

Robot is an inaccurate term here.

No one thinks about chess correctly (including myself). 'Engines' are highly flawed.

Every time the engine says I blundered or made a mistake yet won the game - I knew what was going on all along. Prove I didn't. 

If I had played what the engine thought was not a mistake I would have lost.

@4280
"its possible to prove that a position is winning with other means than beating stockfish or any other robot in the position"
++ Of course there are other means to prove and even more: beating Stockfish is no proof.
I know KQ vs. K, KR vs. K, KBB vs. K, KBN vs. K, certain KNN vs. KP, certain KQ vs. KR are all won. Everybody knows, as these have been strongly solved in the 7-men endgame table base. However, many will fail to win some of the latter. So all know, but some do not know how.

I know 1 e4 e5 2 Ba6, 1 e4 Nf6 2 Qh5, 1 e4 d5 2 Qg4 win by force for black and I am confident I can win these against Carlsen or against Stockfish.

I also know for sure 1 e4 b5, 1 e4 f5, 1 d4 g5 win by force for white, but I may well fail to win those against Carlsen or Stockfish. That is the difference between ultra-weakly and weakly.
Ultra-weakly solved means knowing if a position is a draw, a win, or a loss.
Weakly solved means knowing how to draw / win that position.

Some here do not understand the difference between ultra-weakly solved and weakly solved.
Some here still do not understand that weakly solving Chess requires far less positions: 10^17 than strongly solving Chess: 10^44.

@4272
"This happens more in theoretical physics where rigour is sometimes put aside in favour of progress."
++ There is no merit in rigour.
Take the science of meteorology. They collect data from weather stations, weather balloons, satellites, radar and then use supercomputers to reliably forecast the weather of tomorrow.
Many people depend on that for air traffic, agriculture, or just to decide on clothing and whether or not to carry an umbrella.
Now consider 3 possible weather forecasts for tomorrow:
A) It will rain.
B) It will not rain.
C) There is some chance that it will rain.
Rigour is to always forecast C): it is never wrong but also completely useless for any purpose.

Right, so you condemn those who solved checkers for their lack of merit? They were insufficiently sloppy to deserve praise?

There you go again. Same severe learning problem.

What does "the tablebase" tell you about the position after White's move 37 (shown) here?

(Game played in Tarrasch which enforces 50 move and triple repetition rules.)

and who exactly wins here?

or here under competition rules?

And what is "the tablebase"?

There are DTC, DTM, DTZ50 tablebases etc. that won't generally agree even on ply count 0 positions, either on the result or the moves that should be played.

@4286
"By contrast deductive reasoning deals solely with probabilities of 1 and 0 - known as True and False in boolean logic - and is able to come to the conclusion that a conclusion is True - its probability is 1."
++ The conclusion that 1 e4 e5 2 Ba6 loses by force for white is true, its probability is 1.
The conclusions of this paper are true: they have been derived from nothing but the Laws of Chess by boolean logic
https://arxiv.org/pdf/2111.09259.pdf 

@4287
"so you condemn those who solved checkers for their lack of merit?" ++ No, I condemn those who say chess cannot be solved for rigour standing in the way of progress

"They were insufficiently sloppy to deserve praise?" ++ They deserve praise for their progress. Sloppyness deserves no praise. Rigour standing in the way of progress deserves no praise either.

With due deference to your many degrees in flower arranging, this is a heap of crap.

The application of boolean logic in AZ produces only numeric figures and recommended moves. It doesn't address the conclusions you ascribe to the paper nor even the conclusions in the paper.

And where does the paper say 1 e4 e5 2 Ba6 loses by force for white?

This discussion is about solving chess. Your subjective and personal reticence, regarding committing yourself to any firm decision on the objective merits of random series of moves, means that your comments, however intrinsically accurate they may be, are inapplicable to this discussion.

If you can't bring yourself to know that something like 1. e4 e5 2. Qh5 loses for white, but you only think it so, then logically, you cannot bring yourself to fully accept that any random series of moves, no matter how ridiculous, ends in a definite, forced result with best play, without tracing all lines to the end.

It means that on your say-so, it would be impossible to cut down the myriad permutations of moves in chess, to leave only relevant (i.e. sensible) moves. So we're left with the total number of permutations or variations, which is said to be over 10 ^100. A so called "strong" solution is therefore necessary, to produce the foolishly named "ultra-weak" solution of whether chess is a draw or not, with best play. Such a full solution, consisting of tracing all the possible chess lines to their ultimate conclusion, seems impossible to achieve, by any means. It would certainly be impossible to store the results, using present or foreseeable technology.

I can only conclude that your refusal to understand the processes that need to be involved in solving chess is wilful. But it seems rather an extreme method of negating tygxc's claims and it makes you just as incorrect, wrt answering this, as he undoubtedly is. It was unnecessary and pedantic to make your detailed explanations, which were obvious, although inapplicable to the question of solving chess. It's an absolute necessity to be able to assess positions to eliminate lines that are irrelevant to "best play".

@4291
++ A scientific paper is no 'heap of crap'.
I ascribe no other conclusions to the paper than what the authors conclude.
'human concepts can be accurately regressed from the AZ network after training,
even though AlphaZero has never seen a human game of chess'
AlphaZero corroborates human knowledge with no other input but the Laws of Chess and just by performing boolean operations, i.e. logic.

Another scientific paper endorses the use of game knowledge in solving a game:
'Next to brute-force methods it is often beneficial to incorporate knowledge-based
methods in game-solving programs'
https://www.sciencedirect.com/science/article/pii/S0004370201001527 

That is what I say the whole time.
We know (knowledge) that 1 e4 e5 2 Ba6 loses for white. There is no need to consider that.
We know (knowledge) that 1 a4 cannot be better than 1 e4 or 1 d4. There is no need to consider that.

Much as I disagree with your ultimate assessment, regarding timescale, at least you are commenting relatively accurately and on subject, when compared with Elroch and btickler. I would suggest that they're trolling.

@4295
Your post is a heap of crap.
The scientific paper ranks all initial moves in figure 5 and figure 31.
Hence it is clear that considering 1 e4, 1 d4, 1 c4, and 1 Nf3 as first moves is enough.
That 1 e4 e5 2 Ba6? loses for white is obvious. I even proved it is a forced checkmate in 82.

We can safely say that the probability that 1. e4 e5 2. Ba6 loses by force is 1. Unfortunately, Elroch doesn't understand that. Apparently, neither does MAR.

It doesn't help, though, because there are many positions which are probably losing or winning but the relevant probability is less than 1. To solve chess, an algorithm is necessary which cuts out the obviously losing continuations. It's there where the problem lies, because such an algorithm needs to be 100% accurate.

I therefore believe that "solving chess" by the means of tracing lines is literally impossible to achieve and such a solution would depend on an accurate, mathematical depiction of chess being achieved. Then, perhaps chess can be solved by means of equations. However, such a mathematical depiction of chess is impossible to achieve using known mathematical methods and it will have to wait at least until the full development of A.I. allows such equations to be achieved.

Although I think that Elroch wasn't contributing helpfully, it did serve to concentrate my own mind. I have now reached the conclusion that a full solution of chess must be regarded as impossible, using known methods. Elroch reached an implicitly identical conclusion, although, I think, accidentally.