Chess will never be solved, here's why

@9723

"a chess position isn't perfect information"
++ It is: you know all pieces from both sides. It is unlike Bridge, Stratego, Backgammon (you know the pieces, but not the dice), Dominoes, Poker...

@9727

"statistics isn't solid proof 100% of the time" ++ Last year I had to use statistics,
i.e. a Poisson distribution, as there were some decisive games in the ICCF WC Finals.
This year 105 draws out of 105 games, no statistics needed.

"computers aren't smart enough to do that yet"
++ The ICCF grandmasters with their engines now are.

"they like playing a bunch of petrovs which is a draw"
++ There were only a few Petrov Defences.

"What were the openings played?" ++ Ruy Lopez, Italian, Petrov, Najdorf, French, Queen's Gambit, Catalan, Nimzovich Indian Defense, Queen's Indian Defense, Reti...
The interesting thing is that there is more than one way to draw for black.
So even if there were a nonperfect game, there is still a fallback option to draw.

@9731

"What about the kings gambit" ++ Nobody is foolish enough to play King's Gambit in an ICCF World Championship Finals. It is in all probability a white loss, and even if white manages to draw, it is months and months of suffering. They try to win as white, not draw. Even then they draw.

@9732

"If you get your information out of that book then why should I want to read it?"
++ It is a paper, no book, and it would prevent you from talking nonsense.

"Oh is it 17 GMs now? It was three a couple of years ago wasn't it?" ++ I originally calculated 3 grandmasters with 3 supercomputers can weakly solve Chess in 5 years as Sveshnikov predicted. Now however the 17 ICCF (grand)masters with their engines do the job.

"close to perfect accuracy" ++ They came closer to perfect accuracy in previous years,
but now they have reached perfection: 105 draws out of 105 games.

"given that computer games will be much longer than normal games"
++ No. Shortest 15, longest 73, average 38, standard deviation 11.

"which is about a year" ++ They started 20 November 2022. 31 games are still ongoing.
Each of the 17 ICCF grandmasters uses several engines. If I were to qualify for the ICCF World Championship, I would buy, rent, borrow, or steal whatever computer power I could.

"1000 games should be enough."
++ There will be 136 when it is over. Next year another 136 unless they stop organising it.

"I don't think looking at every possible continuation of every possible continuation is what we need to do." ++ Of course not.

@9736

"How do you get to the 7-men tablebase draw" ++ White plays what he thinks is his best move. Black plays what he thinks is his best move. They play move after move average 5 days/move, with engines. Then they arrive at a 7 men endgame table base draw, or a prior 3-fold repetition.

"there's no way to seperate good lines from bad"
++ If it ends in a draw, then it was a good line for black.

@9735

"kings gambit is only slightly worse"
++ Players that qualified for the World Championship try to become World Champion.
They try to win. They do not play King's Gambit.

Dang it I typoed again

@9740

"average length of games is 76 moves to a tablebase"
++ No, from initial position to table base draw or 3-fold repetition average 38.

I agree with tygxc that it might be possible to do what he proposes (my guess is not as quickly as he claims) and find lines that draw, in the limited amount of openings he will examine and following only the lines his gaggle of GMs choose to pursue.

I would say this comes as close to "solving" chess as you could come to perfecting weather prediction--if you only considered certain locales at a certain season and when prescribed weather conditions prevail in nearby upwind locales.

The answer to the question of the topic would be:

Because those who are trying to solve it are all idiots, including AI of course, because it has been programming by the same.

@9746

"I think that's a mistake. Engine based games tend to be longer than human ones."
++ You are mistaken. The human ICCF (grand)masters agree on a draw when neither has any chance left to win. Also they claim a 7-men endgame table base draw when they reach 7 men.
The shortest draw was a 3-fold repetition in 15 moves.
The longest draw was a 3-fold repetition in 73 moves.
Average is 38 moves with standard deviation 11 moves.

@9752

"chess may be regarded as infinite"
++ No way. 10^44 legal positions is a huge number, but not infinite.

10^34 sensible positions is a huge number, but not infinite.

10^17 relevant positions to weakly solve chess is a huge number, but not infinite.

@9751

"it is impossible to follow all chess lines and to evaluate them all."
++ It is not necessary to follow all lines and it is not necessary to evaluate any.
Only follow the relevant lines.
Evaluate only by reaching the 7-men endgame table base or a prior 3-fold repetition.
Read Checkers is Solved to see how Schaeffer weakly solved Checkers.

@9750

"it has been programming by the same"
++ No. AlphaZero had no other human input but the Laws of Chess

#9705, #9716, and #9719 are worth noting as an example of a small step forward in understanding.

Regretably the other hundred steps could take a while.

That's optimistic!

Here's a question for @tygxc and anyone else who could improve their understanding of the relationship between empirical evidence and general knowledge.

Suppose you have a large urn that you know contains a huge number of balls that can be black or white and have been thoroughly mixed up. You take N balls out of it without looking. All of the balls are white. How big does N have to be before you know that all the balls in the urn are white?

No it wouldn't because you've left out castling which means it wouldn't be solving At All.
I'm not arguing - I'm simply contradicting you.
Rightly.
That doesn't mean you're doing wrong doing what you're doing though.
I'm encouraging you to continue. Even though you'll continue Anyway.

--------------------------------------------
Got to qualify something I said earlier.
'Strongly solving' is kind of an oxymoron term and 'weakly solving' is kind of a self-contradicting term.
Even if they're accepted terms in the relevant community - they're poorly formed terms.
And the tablebase project is poorly formed since it chose to reject castling.
It becomes transparent.
If they had chosen to factor in castling their progress would have been more pathetic than it is and they would have lost funding. Apparently.
-----------------------
Much more ideal generic terms:
Perfect 'solving' versus 'No its not solved'.

A chess position is 'perfect information' if its legal and possible to exist under chess rules and the information is complete like which side is to move and which way the pawns are going.
To be absolutely complete - it might also have to be stated what the previous move is in case of en passant possible and also whether castling is still legal.
(although in tactics puzzles here there's an issue there).
And there must be 50 move information too. (in theory the tactics puzzles could have that issue too but they don't - its only when a player missses castling that he wrongly complains)
All of the information must be there or the information isn't perfect.
But in looking at games - it always is if the game information is complete and that's what defines the game. The list of moves.
For this project you don't need the clock times after each move.
The game is there - and perfect. (with move numbers too. Usually there just couldn't be a 50 move issue for so many reasons)
Its perfect information.
---------------------------------------
So solving should be perfect too to be True Solving.
And it often is.
But its not always possible. Just can't be done.
Perfect information but so often can't be known whether there's forced win or forced draw there.
And I think that would be obvious to most people who even briefly consider it.
But then - what about the actual solving and the actual nature of the actual 'game of chess 'solved' ' issue?
Instead of 'worrying about technology' what about the actual nature of the task?
--------------------------------------------------
Start with just two Kings on the board.
Its tempting to just say: 
'Hey that's just a dead draw and no need to categorize there.'
But yes there is.
Things aren't always what they seem to be.
Don't understand? Here's why.
King positions must be classified in order to solve properly when adding a piece or pawn and then classifying all positions with two Kings plus a single piece or pawn on board.
Because if either or both Kings are taking away a square or two from the 48 squares that a pawn could occupy - that affects the number of places the pawn could be. It could be 48 places or 47 or 46 - but would otherwise always be legal.
But this would also apply to Q or R or B or N of either color.
Why - because when a King is on a corner square there's 60 other places the other King could be.
But when he's on one of the interior 36 squares - there's only 55 places the other King could be.
And when a King is on one of the 24 edge non corner squares there's only 58 squares the other King could be.
In each of the three cases the piece could legally be on any of the other 62 squares - but the King restricitions still affect the number of positions.
Then when you add restrictions for a pawn and even 47 or 46 squares instead of 48 even just two Kings plus One other starts to get involved.
But then when you add a fourth man to the position - even then it starts to get Cruel.
Because many positions would be illegal and impossible.
Like both kings in check. Or the position couldn't have got there.
The computer would have to allow them and then process and find them illegal - To be completely perfect - it could be even Crueller. The computer has to indicate whether forced wins are possible under the 50 move rule and since it doesn't have the 'game info' it has to classify for 'the cutoff' on how many of those moves gone by.
-------------------------------------------
Gets harder and harder the more pieces are added.
By a factor of about 500 each time.
Ten piece types x many possible squares to add to.
Conclusion - the whole things 'in check'.
The sun would engulf the earth long before completion.
So the forum topic name is correct.
Chess will never be solved and there's why.
Soon the forum will reach 10,000 posts.

With all due respect, you shouldn't guess if you want to find out the truth, you should reason! One thing you can do is consider possible things that might happen and how that affects what you can deduce.

First let me explain that in this thought experiment what you are doing is inductive reasoning. You have empirical examples and you are trying to arrive at general knowledge.

The simple truth is that such reasoning always leaves uncertainty, except in the sole case where the empirical examples comprise all possible examples.

So in the example of the urn, you need to take ALL of the balls out to determine they are all white. If you think this is not so, take out all of the balls except one and tell me what colour the last one is.

You can't DEDUCE its colour. What you can do is have a confidence in what colour it is. If there was only one black ball in an urn with a million balls it would be unlikely that you would happen to take out the 999,999 white ones first and leave the black one. In fact the probability of this is exactly 1 in a million. But this is not impossible. Merely rather unlikely.

The relevance to this forum is when people try to claim that seeing the results of an incomplete set of examples PROVES something about the other ones. This is simply wrong. It merely provides statistical evidence. As a relevant example, it's just like when you try to solve a difficult mate in 2 problem, where you have checked 14 of the 15 responses to a candidate solution move and found you can mate in all of them, and you think the last move looks a bit silly. Anyone who thinks you can be SURE the last response to your move permits a mate needs to think again!