Chess will never be solved, here's why

How can you be sure the 50 move rule has no effect on a final solution? You'd have to know what the final solution is first. That's like saying you can be pretty sure chess is a forced win for black. At best, it's a hunch. But maybe pretty sure is a hunch, I don't know. 

The reason it matters is because there are countless positions yet to be discovered that are forced mates in 200, 300, 500, maybe even 900 or more. And many of those will have forced mate solutions where the 50 move rule would prevent the necessary progress to get the mate. And we just can't say for sure yet ALL of those undiscovered positions can't be forced from the opening position. 

Sometimes a game ends in a draw only to find out later (through computers) there was a forced mate solution to that draw. 

But we know it's a draw. Some of us anyway. Forced mates exist because of previous errors.

It's  a  draw

Unless the opening position isn't an error. We can guess it's a draw, and I guess it's probably a draw with some of the current rules, like the 50 move rule. I could be wrong, but that's my guess. 

I just feel that there is so much  unknown, particularly with very long complex endgames with a lot of pieces where there ends up being a forced mate. I haven't seen any good reason why at least some of those complex endgames can't be forced from the opening position. Or even more likely, a much longer and much more complex forced mate endgame that's yet to be discovered. I think everyone agrees we are going to see longer and more complex forced mate endgames as computers get more advanced. 

Good article

#3756

"How can you be sure the 50 move rule has no effect on a final solution?"
++ Because grandmaster and ICCF and TCEC games usually end before move 40. Because the 50-moves rule is almost never invoked before a 7-men position is reached. Because there is always a complelling reason to move a pawn or capture.

"You'd have to know what the final solution is first." ++ No.

"we just can't say for sure yet ALL of those undiscovered positions can't be forced from the opening position." ++ GM and ICCF and TCEC games rarely last over 40 moves. The 50-moves rule is almost never invoked before 7 men are reached.

"Sometimes a game ends in a draw only to find out later (through computers) there was a forced mate solution to that draw." ++ More often it is the other way around: a game is won and the computer points out a draw.

I don't doubt that more often "it is the other way around". But it would have to be the other way around EVERY time to eliminate the possibility of a forced mate. 

I still think we are many decades, probably at least 150 years from any sort of solution. Since computers are still in their infancy it's far too early to speculate what they can or cannot do in the future. But the trend seems to be that they are getting better at solving chess problems. 

So today a forced mate in 500 is rare, but in the future we may see countless forced mates in 2000 moves. And it only takes one of those positons to be forced from the opening  to make things interesting. 

He tried to make you appear wrong even if he did not make any real objection to what you said. He used that technique over and over in this thread.

#3762

"I still think we are many decades, probably at least 150 years from any sort of solution."
++ GM Sveshnikov said chess can be weakly solved in 5 years, but the impeding factor is money to rent the cloud engines and pay the (grand)master assistants.

"So today a forced mate in 500 is rare"
++ 500 is a fantasy. Games usually end before move 50.
In the last Candidates' tournament the average game lasted 47 moves.
In the ICCF world championship the average game lasts 39 moves ending in draw either by agreement, 3-fold repetition, or a 7-men endgame table base draw claim.

This is the longest world championship game ever. It should have been a table base draw.
The 50-moves rule was never close to being invoked.
https://www.chessgames.com/perl/chessgame?gid=2127373  

This is a famous example where black missed a forced checkmate.
The 50-moves rule was never close to being invoked.
https://www.chessgames.com/perl/chessgame?gid=1937789 

Hi

I think we may be talking about two different things. You are talking about grandmaster games, and tournaments. I'm talking about solving chess. 

I don't understand what things like "games usually end before move 50" have anything to do with solving chess. If there was a solution to chess in less than 50 moves, I think someone would have found it by now. 

#3766

"I'm talking about solving chess." ++ Me too. The closest we have to that are ICCF World Championship games, TCEC engine games, and human grandmaster games. If most of these were draws by the 50-moves rule, then it would be plausible that there might be wins without the 50-moves rule, but that is not the case: the 50 moves rule is rarely invoked before 7 men are reached. When two intelligent entities human, engine, or centaur play, then their games usually end before move 50. The rules of the game are such that good play automatically demands pawn moves or captures.

"If there was a solution to chess in less than 50 moves, I think someone would have found it by now." ++ Chess is a draw. It is harder to prove a draw than to prove a win. Losing Chess (64 squares, 32 men, 6 kinds of men, just like chess) is a forced win and needed only 10^9 positions. Checkers is a simpler game (32 squares, 24 men, 2 kinds of men) but needed 10^14 positions. We already have part of the solution: 99% of the ICCF World Championship drawn games are perfect games with no error from either side. However 1000 perfect games are not a full solution.

TCEC games average out to more like 60 moves as I recall when last I checked, and many go over a hundred moves.

You cannot determine a perfect game without solving chess, and saying that engines play perfect games because the same or other engines cannot discern any errors in those games is a form of circular logic you ought to understand, but apparently do not.

#3768
last ICCF World Championship:
Longest 102
Shortest 16
Average 39
Standard deviation 14
In TCEC games drag on longer, as there is no human entity to cut short obvious draws.

That was a fairly long winded way of saying "you are right, btickler", don't you think?

#3770
Yes, you are right: TCEC games last longer than ICCF games.
However that does not indicate something fundamental, it is just that the TCEC engines play on in drawn opposite colored bishop endgames until they reach a 3-fold repetition, while the ICCF grandmasters just agree on a draw.
Here is an example
https://iccf.com/game?id=1164259
The ICCF grandmasters agree on a draw as they know further continuation is pointless. TCEC engines would play on.
This is 99% sure to be a perfect game.

#3768
"You cannot determine a perfect game without solving chess" ++ I cannot determine a perfect game, but I can apply statistics on 136 games and infer that 99% of ICCF games are perfect games with no error. That is based on error probability calculation.

"saying that engines play perfect games because the same or other engines cannot discern any errors in those games" ++ That is not what I do. I apply statistics and probability calculation on ICCF tournaments of 136 games each. Each such game represents a few years of engine time.

Since you cannot determine that, what you are really saying is that ICCF games are played to the highest current standard of play, without discernable errors in hindsight being detectable.  It doesn't matter how long you run the engine for past a certain point because of diminishing returns, the same way you could let Sveshnikov ponder one opening for his whole life and he still wouldn't be able to claim perfect play even for that opening.

#3773
"ICCF games are played to the highest current standard of play" ++ Yes

"without discernable errors in hindsight being detectable"
++ Yes, but also with error probability calculation indicating 99% probability of 0 error.

"you could let Sveshnikov ponder one opening for his whole life and he still wouldn't be able to claim perfect play even for that opening."
++ Sveshnikov claimed his 1988 book The Sicilian Pelikan had B33 fully analysed to a draw.
'"By publishing a monograph on the 5...e5 system in 1988, I practically exhausted this variation. Since that time only some details have been developed, without introducing anything particularly new: the evaluations of the main lines have hardly changed. I described everything in such detail, that it became hard playing 5...e5 even against first category players...'

Carlsen and Caruana and their teams of grandmasters and cloud engines tacitly agreed to that as in their World Championship match games Carlsen allowed the main line 7 Bg5 and Caruana avoided it.

ICCF games also confirm this:
https://iccf.com/game?id=1164361 
This game too is 99% sure to be a perfect game with 0 error.

else... it is plausible anyway, but he tries to make you think otherwise.

Period. Mathematically proven. No doubt. 100% sure... So why nobody wrote a paper to declare chess ultra-weakly solved is a mistery.

For calculations tygxc and tygxc only believes correct.