Chess will never be solved, here's why

@Kotshmot
Are you aware that many people don't regard your country as existing at all?
Finland-denial is more popular than Australia-denial.

tygxc why cant you just answer basic questions?

just want to reiterate how this claim is literally pure delusion - the claim that tygxc responds to is that what if chess has a winning line that no one knows about?

tygxc's response just ASSUMES that the ICCF would already know the winning line LMFAO thats literally in violation of the premise!

ah yes, as evidence for ICCF being perfect, let's just assume they are already perfect so they cant make errors!

I'm thinking young lawyers in law school debating societies are both trained and required to defend ridiculous positions.
How crazy a position can actually be defended?
Any.
Especially if the 'jury' is the same person crazily defending his crazy anti-logical anti-mathematical anti-science positions.
T will always be able to defend his nonsensical positions to 'self' as jury.

@7xz you want to defend this?

Well ... arguably, ICCF (grand)masters aren't exactly superior at chess, on their own. In the top field, the strongest player I've seen is below 2300 FIDE.

If we were to tell a top FIDE player (Carlsen, for example) that a 2300 understands chess at a high level, he'd raise an eyebrow and feel inclined to disagree ...

This means that the decisions they make aren't coming from a position of superior human understanding; they're coming from trial-and-error exploration of what the engine(s) tell them works, or doesn't.

So "spotting errors" depends mainly upon the infallibility of the engines involved.

Historically, engines prove to be inferior, time and again, to the newer generations of engines. Even the play of the revolutionary AlphaZero, from 2018, is now considered subpar, by today's standards.

David Silver, co-creater of AlphaZero, declared in 2020 that AlphaZero would lose convincingly to future versions of itself, due to the forward march of computational power, combined with the continual learning of its neural network.

Time marches on, and the strength of engines continues to climb - marking previous peaks in play as lower than originally thought.

We can now look back at previous ICCF WC draws and find mistakes that both the players and engines failed to notice. The question is: will the same hold true many years from now, looking back on the current ICCF WC draws?

If our answer is "No", then the argument would be that no more improvement can be found. The current iterations of Stockfish (and its peer engines) are the highest any engine can climb, and there will be no future strength gains by any engine.

This would also suggest that the current hardware/software that the ICCF WC players are using would be enough to compete against all future ICCF WC competitors, regardless of how advanced the hardware/software may become for the others in the field ...

However, if we were to say, "Current hardware/software might not prevail against future improvements in hardware/software" - then we would be conceding that the current hardware/software may not be finding the best moves ...

What if we removed the 50 move rule as well? Since that rule is a practical thing to avoid endless random moves in tournament games but here random moves is the goal! Could a chess game continue for trillions of moves before inevitably checkmating or repeating moves?

@MaetsNori it should be brought to your attention that this stuff has been pointed out to tygxc dozens of times already. tygxc's response will be "no, we know that they wont improve further because we already know that they are perfect now because they drew 116 times"

either that or he will ignore u entirely and try to deflect.

@11392

"ICCF (grand)masters aren't exactly superior at chess" ++ They do not need to be superior at over the board chess. They need to be good at analysis with engines, need to be good at positional play. They do not need a good memory, strong nerves, or physical stamina. See them like the seconds who assist the top player grandmasters. They complement the engine like the weaker Vainstein (not even a master) helped the stronger Bronstein and the weaker Dokhoian helped the stronger Kasparov.

"We can now look back at previous ICCF WC draws and find errors that neither player noticed."
++ That is possible, but back then there were decisive games.

"will the same hold true many years from now, looking back on the current ICCF WC draws?"
++ No, because now they have reached perfection: 0 error/game.

'If you could change one thing in the world of correspondence chess, what would you choose/do?' - 'It is necessary to radically reduce the time for thinking.' - Dronov

"no more improvement can be found" ++ No. Improvement is quite possible.
Achieve the same 0 error/game at less than 5 days/move and/or without human.

"there will be no future strength gains by any engine" ++ There can and will.
Maybe 0 error/game at 5 hours/move, 5 minutes/move, 5 seconds/move...

"the current hardware/software that the ICCF WC players are using would be enough to compete against all future ICCF WC competitors" ++ Not if they reduce the time for thinking.

That doesn't quite answer the question of how exactly retrograde analysis helps determine the best moves. Working backward from every possible checkmate/stalemate/insufficient material position is like generating a massive game tree in reverse. But how does that somehow tell you what the best branch is going forward? If it gives you a massive gametree back from some 2 knights vs pawn checkmate position, how does that reveal which of those branches is the optimal set of moves the other way?

This is an interesting question and begs one to define a "mistake".

We could theoretically have a sample of, say 116 games, where current engines play each other and produce games that contain no moves, that change the theoretical outcome during the course of the game - This doesn't mean the engines are unmistakeable facing future engines, that find more challenging lines.

Making no mistakes against one opponent doesn't suggest you wont make mistakes against another one.

So, to clarify your position for my own understanding: you believe that chess improvements, regarding the moves themselves, will no longer be found; only improvements in speed will be made from here on out ...

you'll notice how tygxc did completely ignore the possibilities of improvement you presented, just as i predicted, and instead tried to deflect by redefining "improvement" when you were obviously exclusively talking about reducing errors.

Chess has mathematical patterns but the solution isn't mathematical.

Well, i feel its some kinda STEM. donchu ? and maybe even some kinda STEAM. a/w we needta start by saying how much (lets take e.g.) pawns are worth. e/o says uhhh....1 !! thats sooo dopey. before a game begins a d-pawn is probably worth more than a a-pawn right ? well, how much more ? well thats gonna be difficult. Accurately evaluating ea & every piece/pawn relative to ea & every accurately evaluated square it controls &/or occupies (or CAN !) needsta happen...PER PLY. our effort is to solve chess right ?...not guess as to how many legal positions there are...hello-hello-hello. iows thinking like a computer isnt gonna get u a/w. trust me.

@Maets-Nori
from your post:
"So "spotting errors" depends mainly upon the infallibility of the engines involved."
I would put it differently.
Fallibility. Not infallibility. With that second term suggesting 'binary'.
Fallibility. Degree of fallibility.
Infallibility of the engines never established. Might never be.
'Spotting errors' depends on degree of fallibility.

And - lets see if EE deletes all his posts here every night as he does in other forums.

Hmm, yes I did notice that he redefined improvement.

Even giving him the benefit of the doubt, though, I find it quite a leap in logic to conclude that no more chess improvements (regarding the moves themselves) can ever be found ... It suggests that engines have maxed out.

And yes, I know tygxc would clarify this by saying "engines + humans" - though I remain unconvinced that the humans have any positive impact on the games, at this point ...

Come to think of it .. if we were to declare that engines perform better with humans than without - then wouldn't that, in itself, imply that engines have room for improvement?

And around we go, with the circular debate ...

A good catch, I agree. That would have been a better way to phrase it.

I appreciate the correction.

the 'determining which moves are optimal' problem and 'constructing a very long mate chain' problems, while connected, are separate.

for 'determining which moves are optimal' - the retrograde analysis is not done ON the moves to be analyzed, but has been done BEFOREHAND in order to create the tablebase that is used, and referred to, in determining which moves are optimal.

it's not that a reverse tree has been created by a singular end checkmate position and traced backto the move to be analyzed, it's that a reverse tree has been created for EVERY SINGLE POSSIBLE end position, and ALL possible trees have ALREADY been traced to positions on the tablebase.

thus from reverse game trees to all positions, the true game tree has been created for your specific position.

Ive seen it for years, tygxc just asserts that some unjustified statement is absolute fact, people question and rebut it, tygxc ignores the rebuttal and introduces a new deflection with a new unjustified statement, each time refusing to address the actual content and instead making the merry go round. there isnt even a true 'cycle', eventually tygxc just stops responding when someone lands on a core fallacy that he makes (for example he just did that with elroch) and then starts talking to someone else and acts like he's facing all comers.

tygxc doesnt answer me anymore because ive seen the cycle too many times, he calls me a troll for the fact that i repeatedly point out his core fallacies. But i don't mind that. I'm here to make sure that tygxc doesn't mislead people, and I have been very successful at that.