Chess will never be solved, here's why

Yes, it is extremely likely that chess is a draw. The probability is so close to one it is reasonable to be certain. This is not the same as proving it, of course!

If I may use a loose chess analogy, it is virtually certain that Carlsen would beat a 100-rated tyro in a chess game. But he has not beaten him unless he plays a game against him and wins.

We don't have a deductive proof about anything in the natural world (only about models of it). Proof is restricted to the abstract.

Solving chess is abstract, so could have a proof, but it just happens to be impractically complex.

Although I agree with your opinion, it is also impossible to prove you right.

I wish I could understand whats going on here with all these thousands of paragraphs. Gravity? really?!. how the hell did it come to the gravity

If carlsen can still play to enjoy, i definitely can play to enjoy and so the “solution of chess” doesn’t matter

you are completely right! tygxc is arguing that this proof: "Such a proof itself would take trillions of years on a very fast computer, not that it exists." HAS ALREADY BEEN COMPLETED.

hey tygxc wheres the Nf3 strategy steal?

noticing you dodging that.

I thought nf3 was the best move since it has the most dynamic choice of opening

basically, tygxc is claiming that any position that can be made so that its white's move, white can reach that position. he claims that white can lose a tempo (despite blacks best efforts). but unfortunately, with NF3 NF6 start, theres literally no way to lose the tempo.

heck, its impossible for even Tygxc's own example for white to lose the tempo

White can't necessarily lose tempo but there's still no way to win if played correctly I think for now it is just too complicated though

Besides black chooses the opening not white white can only choice d4 e4 c4 nf3(those are the best moves ) it's kinda funny when people think that black losses too much tempo but not really since you chose the type of position(sorta) you want to play white you don't get to do that (sorta )

musk has that covered

Found this thread, just here to leave my thoughts.

Finding the game-theoretic value of chess is simply not possible with current computational power (and not with any kind of future computational power by brute force, though it may turn out there is a way to simplify the process of solving chess).

Chess engine developers have come up with a way to quickly verify that our engines are generating legal moves, by exhaustively searching all positions that can result from a given position following a certain number of moves. Some people have made it into a game to pointlessly optimize performance on this test, known as "perft". We have only made it to depth 15, i.e. 7.5 moves into the game. Doing so required looking through a total of over two thousand quintillion positions (to be specific, it was 2 015 099 950 053 364 471 960 positions). Of course, lookup tables were used to speed up the process. See: https://www.chessprogramming.org/Perft_Results#Initial_Position

As for the total number of possible chess games, that is almost certainly greater than 10^10000, a number which would be represented as a 1 followed by 10000 zeroes, and which I have no desire to post here. See: https://wismuth.com/chess/longest-game.html

Strongly solving chess means going through all these games. Once you see the scale of the numbers, it seems fairly evident that solving chess is a long way off.

Why are chess engines so strong? / How do chess engines search so deep?

You may have noticed that it's entirely possible for chess engines to search very deep (I can get depth 40 using Stockfish on my computer in a few minutes). However, this is actually just because chess engines don't search the whole game tree, instead only focusing on the most promising opportunities. Doing such is called "pruning" and without it, my chess engine can only see two moves ahead when playing a blitz game.

It is possible to construct positions in which Stockfish will miss a win due to pruning, though actually creating such a position would require quite a bit of scrutinizing of Stockfish's code. There's also positions in which you can get Stockfish to miss a win because it can't search deep enough. Courtesy of Yosha Iglesias, I'll give an example of that:

I'd like to learn how to make a chess engine. Can you provide me with some resources?

Assuming you know how to program, check out the Chess Programming Wiki: https://www.chessprogramming.org/Main_Page

Stockfish's code may also be a useful reference, though it is strongly unadvisable to copy it: https://github.com/official-stockfish/Stockfish

@9238

"How detectable do you think one blunder in 1000 games would be in your sample?"
++ One error is simple to detect:
if a game ends decisively, then an error has been made somewhere.

"conclude from your sample that the error rate is zero"
++ I conclude that the 104 ICCF WC Finals draws are perfect games with no errors.
I do not conclude anything about previous or future games.

"uncertainty of inductive reasoning" ++ Most science is inductive. Keppler derived his laws of celestial motion from astronomical observations by Tycho de Brahe. Only later came Newton with his theory of motion and gravitation, from which the same laws could be derived.

Likewise we know how stars of some kind behave, what they consist of,
what happens inside them, how old they are. That is by observing some stars, not all stars.

@9239

"1 tempo isnt enough to win "
++ You can queen a pawn, but you cannot queen a tempo.

@9240

"you claimed every position could be strategy stolen"
++ No, I did not claim that.
I claim that for any tentative strategy to win as black,
there exists a white strategy to win by stealing it.
If 1 e4 c5 were a black win, then 1 c3 e5 2 c4 would be a white win.
If 1 d4 Nf6 2 c4 g6 3 Nc3 d5 were a black win,
then 1 Nf3 d5 2 g3 c5 3 d3 Nc6 4 d4 would be a white win.
There cannot exist a consistent black strategy to win, as white can steal it.

@9242

"1. g4 is supposed to be near to losing." ++ Near losing does not exist, it is either losing or not.
1 g4? indeed loses by force, and it is the only white first move that loses by force.

"Another mis-spent tempo and it would lose." ++ No. 1 g4? losing by force has nothing to do with tempo, but with weakening the king's side. White can afford to lose 2 tempi,
and black can afford to lose 1 tempo. That is the drawing margin.
https://www.chessgames.com/perl/chessgame?gid=1026344

https://www.chessgames.com/perl/chessgame?gid=1768345

"white's extra tempo is gradually anulled with good play by both sides." ++ Yes

"a bad move has to be played to allow the zugzwang." ++ Yes

"Such a proof itself would take trillions of years on a very fast computer"
++ No. It is unnecessary to strongly solve Chess to weakly solve Chess,
and it is unnecessary to weakly solve Chess to ultra-weakly solve Chess.
Checkers for example is weakly solved, not strongly.

@9245

"We don't have a deductive proof about anything in the natural world"
++ We have deductive proofs from models like Newtonian mechanics, Maxwell's Laws,
Navier-Stokes equations, Schrödinger's equation, Einstein relativity etc.
The models are abstract, but apply to phenomena in the natural world.

"Solving chess is abstract" ++ Chess is in the natural world: we move wooden pieces on a wooden board and at the end it is either checkmate, or a draw.

AlphaZero has acquired chess knowledge from no human input but the Laws of Chess.
https://arxiv.org/abs/2111.09259