Chess will never be solved, here's why

I would think it's possible to make an opening book long enough to reach to the point that a tablebase could take over. If that counts, then there ya go. (?)

2 things.

first, Go is much more complex than chess.  it took 20 years after deep blue to get the same level for AlphaGo.

second, a massive number of permutations doesnt necessarily mean that something cant be solved.  checkers had 10 ^20 and was still solved.  of course, chess is much much more complex, but the big number alone should dissuade us.

Your premise assumes that the efforts put into beating the world champs for Chess and Go were the same.  This is not the case.  Solving Chess was much better PR for IBM than solving Go would have been, so a lot more resources were brought to bear.

In terms of actually solving, IIRC Go has more positions...but evaluating Go positions should take less CPU horsepower than evaluating Chess positions.

That is an absurd statement.

@7279

"lets say its achieved and lets assume its draw.is computer going to play for draw everytime?
++ You cannot play for a win. You can only play to not lose.
To win your depend on your opponent making a mistake.

"black's best moves as computer play can  easily be memorized"
++ Think again. For a simple game like Connect Four you can memorize all moves, but not for Chess. A weak solution needs 10^17 positions. Maybe the solution tree can be simplified to 10^9 positions, i.e. 10 million games. No human can memorize that.

@7269

"For chess the rules are the basis of the axioms."
++ Yes, exactly, the Laws of Chess are the axioms.
That is why this paper produced theorems by boolean logic operations on said axioms.
https://arxiv.org/pdf/2111.09259.pdf 

His comment was fine, if uninformed.  Yours was absurd.  I've bolded it for you. 

You're not going to last here long.

Go ahead and dig through some source code, we'll wait.  I hate to inform you, but both Checkers and Go have simpler mechanics that make individual position evaluations easier for brute force computation.

@7260

"You could also then tell us which version or versions you propose to solve."
++ The Laws of Chess

"If it's not enough to win, it's not an advantage"
++ The first move is an advantage, called the initiative and representing 1 tempo.
That is not enough to win: to win a pawn is needed.
White has many good moves that draw and few bad moves that lose.
Black has few good moves that draw and many bad moves that lose.

"I wouldn't have thought there's much you can do with 1/3 pawn"
++ 3 tempi are the equivalent of 1 pawn. You can sacrifice a pawn if you get 3 tempi in return. You can give up 3 tempi to win a pawn.

"Under FIDE rules it just takes checkmate"
++ You cannot win by a direct attack on the king if your opponent defends optimally.
Queening a pawn is a more feasible way. That is why a pawn is enough to win, and so is a bishop, but a tempo is not enough to win. You cannot queen a tempo. You can trade a bishop for a pawn and then queen the pawn.

@7307

"both Checkers and Go have simpler mechanics that make individual position evaluations easier"
++ The whole point of solving is to not depend on evaluations, but rather to calculate all the way until the 7-men endgame table base.

A

Shoo.  That has nothing to do with his statement being absurd.  Would you care to argue that "we are just guessing" and that "one page" may suffice to bridge our way to tablebase results?

Ahhh, the old "everybody's opinion has equal validity" argument...

You can decide something is absurd if you like.  It matters little, if you continue to display this level of understanding (or lack thereof).

Oh no...Stockfish 15.1 release just blew Tygxc's premise out of the water.  What will the future bring?

https://stockfishchess.org/blog/2022/stockfish-15-1/

"This release also introduces a new convention for the evaluation that is reported by search. An evaluation of +1 is now no longer tied to the value of one pawn, but to the likelihood of winning the game. With a +1 evaluation, Stockfish has now a 50% chance of winning the game against an equally strong opponent. This convention scales down evaluations a bit compared to Stockfish 15 and allows for consistent evaluations in the future."

No 1.0 point pawn evaluation, no 0.33 tempo valuation.  These valuations/equivalencies were removed for being obsolete.

Well, admittedly it depends on you...if you are oblivious to any flak that comes your way when you post illogical conclusions, you can last here a long time...ask Tygxc.

woy

Your "book" will stretch out of the solar system .  Not one page.  Not a mountain.  That's not a guess, and we *do* know.  Printing out a 7 man tablebase (18+ terabytes for the "abridged" tablebase, and 140 terabytes for the full tablebase) by itself would be taller than a mountain, and that's only 10^15 positions...which is 0.00000000000000000000000000042383684% of the best estimate for the number of unique and legal chess positions.

Thus...absurd statement and illogical conclusion.

You really have a problem understanding large numbers.  I thought the examples might help.  Apparently not.  So, in your world, "stretching out the solar system" reduces to "much shorter than a mountain" by changing from paper thickness to graphene thickness (not that your contortion of your own example isn't also absurd)?  Graphene is ~1 million times thinner than paper.  That's only 6 orders of magnitude. 

Let's apply that to the percentage I gave...that changes 0.00000000000000000000000000042383684% to 0.00000000000000000000042383684%.  Yeah, you're really making a lot of headway there.

Do you mean because that's not what you're talking about? Because you can easily get an opening book to carry you out fifty moves, and if you've only got seven pieces left on the board, then there you go.

So I assume you mean *all* positions. Mapping out chess so you can get from starting position to the end via any route?