Chess will never be solved, here's why

@8042

There is statistical proof that ICCF WC Finals draws are > 99% certain to be perfect games with optimal play from both sides. They also represent 2 years of play with engines piloted by ICCF (grand)masters. So at least these games are suitable as seeded lines.

If one of these falls under the < 1% cases with 2 errors that undo each other, then that will become apparent during the solving process: then there will appear some line that does not end in a 7-men endgame table base draw.

It will be solved by some over powered AI

"If one of these falls under the < 1% cases with 2 errors that undo each other, then that will become apparent during the solving process: then there will appear some line that does not end in a 7-men endgame table base draw."

, yes, and there are10^17 possible lines, taking up 10^17 positions each.  

your explanation of what weakly solving chess isnt far off from how it would be done.  however, you dont really know what that entails.  at best only a couple orders of magnitude can be confidently reduced.

@8045

"there are10^17 possible lines, taking up 10^17 positions each"
++ No, there are 10^17 relevant positions in total. The whole tree has 10^17 nodes: either branching points or leaflets.

yeah, 10^17 in the final tree.  but you have to find that tree first.

@8047

"you have to find that tree first"
++ Yes: for black play the top 1 Stockfish move after 17 s calculation on a 10^9 nodes/s cloud engine, for white explore the top w (e.g. w = 4) Stockfish moves.

Let us take an example with this ICCF game as seeded line.
1 e4. Grandmaster E with cloud engine e will work on this.
Grandmaster D with cloud engine d will work on 1 d4.
Grandmaster C with cloud engine c will work on other first moves that do not transpose, notably 1 c4 and 1 Nf3.
Grandmaster E and cloud engine e will never have to look at any position with a white pawn on e2, shrinking the search space.
1...e5. Grandmaster E selects this. It is probably that 1...c5 draws as well and 1...e6 and/or 1...c6 might draw as well, but a weak solution only calls for one strategy to achieve the game-theoretic value of the draw.
From now on no positions with a black pawn on e7 are relevant. This shrinks the search space.
2 Nf3. Grandmaster E will study alternatives 2 Nc3, d4, Bc4 later with other seeded games if they do not transpose. There is no need to investigate 2 Ba6?
2...Nc6. Grandmaster E selects this. It is probable that 2...Nf6 draws as well, maybe even 2...d6 draws, but a weak solution only calls for 1 strategy to achieve the draw.
3 Bb5. Grandmaster E will study alternatives 3 Bc4, 3 d4, 3 Nc3 later if they do not transpose. There is no need to investigate 3 Ba6, Nd4, Nxe5, Ng5, Nh4, or Ng1.
3...Nf6. Grandmaster E selects this.
It is probable that 3...a6 draws as well, but a weak solution only calls for 1 strategy to draw.
4 O-O. Grandmaster E will study alternatives 4 d3, d4, Qe2 with other seeded games. From now on no positions with white castling rights need to be considered. That shrinks the search space.
4...Nxe4. Grandmaster E selects this. It is possible that 4...d6 draws as well, but a weak solution only calls for 1 strategy to draw.
From now on no positions with 32 men are relevant. That shrinks the search space.
5 d4. Grandmaster E will study alternatives 5 Re1, Qe2 with other seeded games.
From now on no positions with a white pawn on d2 are relevant. That shrinks the search space.
5...Nd6. Grandmaster E selects this. Other moves might draw, but only 1 way to draw is needed.
6 Bxc6 Grandmaster E will study alternatives 6 dxe5, Bg5, Ba4 with other seeded games.
From now on no positions with 31 men are relevant. That shrinks the search space.
6...dxc6. Grandmaster E selects this. No need to consider 6...bxc6: only 1 way to draw needed.
From now on no positions with 30 men are relevant. That shrinks the search space.
7 dxe5 There are no valid alternatives to consider.
From now on no 29-men positions are relevant. That shrinks the search space.
7...Nf5 Grandmaster E selects this. Only 1 way to draw needed.
8 Qxd8+. Grandmaster E will investigate alternative 8 Qe2 with other seeded games.
From now on positions with 28 men are no longer relevant. That shrinks the search space.
8...Kxd8. Only legal move. From now on postions with 27 men or with black castling rights are no longer relevant. That shrinks the search space.
Now that 26 men are reached, grandmaster E may decide to launch his engine e.
9 h3. Engine e investigates alternatives from its top w moves (e.g. w = 4) after 17 s of calculation at 10^9 nodes/s.
9...Be7. Engine e takes the move from the seeded game.
If it leads to a 7-men endgame table base, then it is right in retrospect.
10 Nc3. Engine e investigates alternatives from its top w moves (e.g. w = 4) after 17 s of calculation at 10^9 nodes/s.
10...Nh4 Engine e takes the move from the seeded game.
If it leads to a 7-men endgame table base, then it is right in retrospect.
11 Nxh4 Engine e will investigate alternatives from its top w moves (e.g. w = 4) after 17 s of calculation at 10^9 nodes/s.
After this move positions with 26 men are no longer relevant. That shrinks the search space.
11...Bxh4 Engine e takes the move from the seeded game. If it leads to a 7-men endgame table base, then it is right in retrospect.
After this positions with 25 men are no longer relevant. That shrinks the search space.
12 Be3 Engine e will investigate alternatives from its top w moves (e.g. w = 4) after 17 s of calculation at 10^9 nodes/s.
12...h5 Engine e takes the move from the seeded game. If it leads to a 7-men endgame table base, then it is right in retrospect. After this positions with a black pawn on h7 are no longer relevant. That shrinks the search space.
etc. ...
57 Bxa1 Engine e investigates alternatives from its top w moves (e.g. w = 4) after 17 s of calculation at 10^9 nodes/s. 

that doesnt work lmao.  you prove absolutely nothing, literally by the second move you have a  false assumption

yes i saw i could have taken that if i would have stopped for a nanosecond, but who gives a f i won anyway

"1... e5 Grandmaster E selects this. It is probably that 1...c5 draws as well and 1...e6 and/or 1...c6 might draw as well, but a weak solution only calls for one strategy to achieve the game-theoretic value of the draw.
From now on no positions with a black pawn on e7 are relevant. This shrinks the search space."

"7...Nf5 Grandmaster E selects this. Only 1 way to draw needed."

yeah but you havent ruled out that black could be playing to win.

you have to search every single one and prove a draw/win for each black move.

your use of grandmasters is also redundant, because a computer can select moves better than any grandmaster could.

you also still assume that black is playing to draw and not win.

I hope i could play so magically and skillfully next time

@8053

"you also still assume that black is playing to draw and not win"
++ It has been established by now that Chess is a draw.
White has an advantage of 1 tempo called the initiative, but not enough to draw.
Black is at a disadvantage, but not enough to lose.
Chess cannot be a forced black win because of strategy stealing.
Formally you can count all draws and all black wins as black successes.
That has also been done in solving Losing Chess, which is a forced win for white. All black draws and black wins were counted as black successes.

"It has been established by now that Chess is a draw."

no it hasnt lmao.  if it was then that would be an ultra weak solve.  by definition.

"Chess cannot be a forced black win because of strategy stealing" 

this is factually incorrect. it is physically impossible to do a strategy stealing thing against black.  your example literally forced black to play a bad move.  plus, black can just mirror whatever white does.  impossible to strategy steal by definition.

@8052

"your use of grandmasters is also redundant, because a computer can select moves better than any grandmaster could."
++ No. The grandmasters work with seeded games that each represent 2 years of engine calculation.
Chess is most complex at 26 men. Computers get swamped from 32 to 26 men. So it makes sense that grandmasters with data bases handle 32 to 26 men and only then launch the engine.

"Chess is most complex at 26 men. Computers get swamped from 32 to 26 men. So it makes sense that grandmasters with data bases handle 32 to 26 men and only then launch the engine."

thats just factually incorrect what. 

you do realize that computers are FORCED to play suboptimal openings because they are literally too good at it otherwise?

@8056

"if it was then that would be an ultra weak solve"
++ For all practical purpose Chess has been ultra-weakly solved and the game-theoretic value of the initial position is a draw.

@8058

"you do realize that computers are FORCED to play suboptimal openings because they are literally too good at it otherwise?"
++ Yes, in TCEC slightly unbalanced openings are imposed to prevent all draws.
However, it is inefficient to recalculate what already has been calculated.
Each ICCF draw represents 2 years of engine calculation under supervision of an ICCF (grand)master.

"For all practical purpose Chess has been ultra-weakly solved and the game-theoretic value of the initial position is a draw."

the fact you said "for all practical purposes" means it isnt ultra weakly solved.

the thing is that tygxc is arguing for it on a theoretical level.