Chess will never be solved, here's why

Nor any point.

Just to satisfy @Optimissed.

Yes, I noticed you don't take much notice of proofs, when they are provided.

Well I wouldn't like to implement any routine in an an EGTB generator that produces an extrapolated proof beyond the number of men it addresses. Not saying it's impossible, but it would have to be pretty ingeneous.

"Extrapolation"? Not a concept that seems to fit in here.

Do you mean in the context of an EGTB generator or in the context of #61?

You can add very few 9 piece positions to an 8 piece tablebase without covering a LOT of 9 piece positions that can be reached before it transitions to an 8 piece position.  There is no reason to believe in significant shortcuts.

So You're talking about in the context of an EGTB generator.

As I said, I wouldn't like to implement such a routine.

But I do, strangely, agree agree with @Optimissed that if chess is solved, old fashioned organic computers will play the major part. Why I said not impossible.  If someone has an idea that could solve chess in a more elegant way than the BFI methods proposed to date, it could rely on information extracted from an initial set of EGTBs and need silicon (or some such material) to complete . In that case you would be more inclined to say that the EGTB generators are subroutines of of the solving program rather than vice versa, but either way.  

There is some reason to believe in shortcuts. Many of the EGTBs could currently be replaced by prescriptive routines requiring no (or little) look ahead. You just need to find a way of replacing the rest.

I have at some point sketched my views on the way a quantum computer could be used to solve chess. Fundamentally it's like constructing an endgame tablebase while dealing with every possible position in parallel. This is technologically challenging, for sure, but not, for any identifiable reason, impossible.

Interesting to see whether humans or computers get there first. But I probably need to improve my lifestyle if I want to see the answer.

It is very safe to say unassisted humans can't solve chess!

Probably can't prove Fermat's last theorem either.

But I think you're probably right.

Unassisted computers most certainly can't solve chess.

#88
Fermat's last theorem was proved by a human.
The Riemann Hypothesis and the Goldbach Conjecture will probably be proved by a human.
The Four Color theorem was solved by a computer guided by humans.
Checkers was solved by a computer guided by humans.
Sveshnikov said 5 years for modern computers and human assistants to solve chess.
My calculations corroborate this.
The answer may come sooner than you think or fear.

"Probably can't prove Fermat's last theorem ..." was tongue in cheek (though I know at least one human who can't).

How is the 5 years worked out? Is that for both versions of chess?

https://arxiv.org/pdf/2112.09386.pdf
See table 3. Human assistants with ChessBase prepare 26-men starting positions 'tabya' to reduce the number of positions.
Every pawn move and every capture drastically reduces the number of relevant positions.
Checkers was solved using the square root of the number of legal positions.
http://library.msri.org/books/Book29/files/schaeffer.pdf 
It is thus plausible that the square root applies to chess as well, maybe more, maybe less.
Cloud engines reach 10^9 nodes / second.
There are 500 ECO codes A00 to E99.
Not all ECO codes are necessary: if 1 e4 e5 draws then it is unnecessary to look at 1 e4 b6 as well.
If black can draw against 1 d4 and 1 e4, then it is trivial that black can draw against 1 Nh3 as well.
Thus 5 years and a few cloud engines suffice to solve it.
That is the standard version of chess, with 50 moves rule and 3-fold repetition rule.

Thanks. Hadn't seen the first.

Didn't understand the relevance of the ECO codes, nor how you prove the statements in the following para.

Also the solution will obviously be very different depending on whether the 50 move rule is in force or not. It may not limit the length of games, but it does limit the length of forced mates. It isn't mentioned anywhere.

#92
1.08e37+6.14e36+3.19e36+5.66e35=2.07e37 legal sensible positions starting from 26 men tabiya
in analogy to the checkers proof: square root of this: 4.55e18 relevant positions 
cloud engine 10^9 nodes/s
4.55e9 s = 144 years on 1 cloud engine for the whole of chess i.e. all 500 ECO codes A00 to E99
Hence 0.29 year on 1 cloud engine for 1 ECO code e.g. C67
19 ECO codes suffice to prove a draw against 1 e4: e.g.
C67 C65 C54 C53 C52 C51 C50 C49 C47 C45 C34 C33 C29 C28 C26 C24 C22 C21 C20
That is 5.4 years on 1 cloud engine to prove the draw against 1 e4.
A 2nd cloud engine may likewise prove a draw against 1 d4.
A 3rd cloud engine may work on the relevant moves other than 1 d4 and 1 e4 not transposing.

Your 5 year clock is already down to 4.5 years ...

#94
5 years after funding for the cloud engines...

OK I see how you plan to use the ECOs - I'll need to think about that.

When the 50 move rule is in effect don't the position counts need to be multiplied by 100?

E.g. with the 50 move rule in effect this White to play position is a win

but this White to play position is a draw