Will chess ever be solved? • page 2/3 • General Chess Discussion • lichess.org

Yes, in theory classic chess will be solved. However, I very much doubt it will happen in any of our lifetimes. The amount of data processing/storage would likely fill the known universe. Also, with the amount of processing power necessary, wouldn’t humans be better served in solving something else?

aentrenus

#12

Antichess has been solved as a win for white

lizani

#13

It can be solved by working backwards. The Nalimov and Syzygy tablebases are 100% accurate for endgames with 7 or less pieces. Dr John Nunn used the tablebases to write definitive guides to certain basic endgames in his books 'Secrets of Rook Endings', 'Secrets of Pawnless Endings' and 'Secrets of Minor-Piece Endings'
Someday there will be a 8 piece tablebase ... and someday long after that a 9 piece .... and ultimately a 32 piece tablebase!

It can be solved by working backwards. The Nalimov and Syzygy tablebases are 100% accurate for endgames with 7 or less pieces. Dr John Nunn used the tablebases to write definitive guides to certain basic endgames in his books 'Secrets of Rook Endings', 'Secrets of Pawnless Endings' and 'Secrets of Minor-Piece Endings' Someday there will be a 8 piece tablebase ... and someday long after that a 9 piece .... and ultimately a 32 piece tablebase!

MolDiva

#14

The argument is all theoretical. It is redundant if machines and databases ever do it. The practical fact is that no human lifetime is long enough for any individual to ever perfect one's play without assistance. We do not live long enough. Therefore I am always enraged by notion that chess at top level is "played out" as an excuse to promote variants and frantic time controls as the game's future. This movement will weaken the entire player pool and weaken future champions. They can master a super machine to "solve" chess and I really do not care because they can not solve the human.

fuxx_de

#15

The technic to solve chess exists already. The problems are disk space and CPU power. It just takes to much time. The 7 piece tablebase needs 18.4 TB. The 8 piece is estimated to be in the peta byte range.

StevenEmily

#16

It is currently infeasible to solve chess in the traditional brute force way, so the only possibility that could be feasible is being clever and coming up with an insanely clever mathematical strategy, finding clever patterns and clever ways to combine those patterns together, eventually creating the starting position as a solvable pattern.

Generalized chess for arbitrary board sizes is EXPTIME-complete (exponential is way harder than polynomial time = NP-hard), so any strategies must be specific to the 8x8 grid's "small" size. It is hard to imagine that there is a solution, since it's like one of those real life combinatorial math problems that have many random details that don't really connect with math's current techniques, but you can't rule out the very very long future.

For this reason, I'd say the likelihood of chess ever being solved is highly related to the likelihood humanity survives like a billion years, or even the next 100.

It is currently infeasible to solve chess in the traditional brute force way, so the only possibility that could be feasible is being clever and coming up with an insanely clever mathematical strategy, finding clever patterns and clever ways to combine those patterns together, eventually creating the starting position as a solvable pattern. Generalized chess for arbitrary board sizes is EXPTIME-complete (exponential is way harder than polynomial time = NP-hard), so any strategies must be specific to the 8x8 grid's "small" size. It is hard to imagine that there is a solution, since it's like one of those _real life_ combinatorial math problems that have many random details that don't really connect with math's current techniques, but you can't rule out the very _very_ long future. For this reason, I'd say the likelihood of chess ever being solved is highly related to the likelihood humanity survives like a billion years, or even the next 100.

Ayushman_C

edited

#17

@sheckley666 @dervidymart I'm sorry, I meant chess games. Thanks for pointing that out! Unfortunately I can't edit the first post anymore because I've edited it too many times :(
Possible chess games = around 10^120
Atoms in universe = around 10^80

@sheckley666 @dervidymart I'm sorry, I meant chess *games*. Thanks for pointing that out! Unfortunately I can't edit the first post anymore because I've edited it too many times :( Possible chess games = around 10^120 Atoms in universe = around 10^80

Sonia_Beliaeva

#18

ну...я...незнаю я не поняла в чём вопрос!

petri999

#19

@lizani said in #13:

It can be solved by working backwards. The Nalimov and Syzygy tablebases are 100% accurate for endgames with 7 or less pieces. Dr John Nunn used the tablebases to write definitive guides to certain basic endgames in his books 'Secrets of Rook Endings', 'Secrets of Pawnless Endings' and 'Secrets of Minor-Piece Endings'
Someday there will be a 8 piece tablebase ... and someday long after that a 9 piece .... and ultimately a 32 piece tablebase!
Already 7 piece databases are incomplete. So in mathematical sense (solving in mathematical concept) there are already 7 piece setups which are not proven to wins or draws.
And each step hugely exceeds effort and memory requirements needed by the previous step . Unlikely that 8 piece will show anytime soon and 32 one will neve happen.

And solving chess will not happen within my lifetime and may well be impossible
https://en.wikipedia.org/wiki/Solving_chess
in particular estimates made already in sixties

Checkers was weakly solved in 18 years of computation and it A LOT simpler game than chess. I doubt even international checkers is solvable by tech available today

@lizani said in #13: > It can be solved by working backwards. The Nalimov and Syzygy tablebases are 100% accurate for endgames with 7 or less pieces. Dr John Nunn used the tablebases to write definitive guides to certain basic endgames in his books 'Secrets of Rook Endings', 'Secrets of Pawnless Endings' and 'Secrets of Minor-Piece Endings' > Someday there will be a 8 piece tablebase ... and someday long after that a 9 piece .... and ultimately a 32 piece tablebase! Already 7 piece databases are incomplete. So in mathematical sense (solving in mathematical concept) there are already 7 piece setups which are not proven to wins or draws. And each step hugely exceeds effort and memory requirements needed by the previous step . Unlikely that 8 piece will show anytime soon and 32 one will neve happen. And solving chess will not happen within my lifetime and may well be impossible https://en.wikipedia.org/wiki/Solving_chess in particular estimates made already in sixties Checkers was weakly solved in 18 years of computation and it A LOT simpler game than chess. I doubt even international checkers is solvable by tech available today

EmaciatedSpaniard

#20

@sheckley666 said in #3:

Actually, there are fewer possible positions than the number of atoms on earth. So, at least, it could be possible to store the solution in a future computer memory.

But i don't think storing all the positions 'solves chess'. You need to evaluate the positions accurately and determine all possible responses and follow all lines until they conclude in a win, draw, or loss. Everytime you add a ply, the number of new positions whose path you must follow goes up. I guess it goes up as a power law of the number of positions? So its exponentially increasing. I am guessing esp. for the middlegame things rapidly blow up, making the task computationally challenging, perhaps impossible with any reasonably available resources.

Chess IS solved (in this sense) from certain endgame positions, but i'd venture we are a long way from solving any position.

@sheckley666 said in #3: > Actually, there are fewer possible positions than the number of atoms on earth. So, at least, it could be possible to store the solution in a future computer memory. But i don't think storing all the positions 'solves chess'. You need to evaluate the positions accurately and determine all possible responses and follow all lines until they conclude in a win, draw, or loss. Everytime you add a ply, the number of new positions whose path you must follow goes up. I guess it goes up as a power law of the number of positions? So its exponentially increasing. I am guessing esp. for the middlegame things rapidly blow up, making the task computationally challenging, perhaps impossible with any reasonably available resources. Chess IS solved (in this sense) from certain endgame positions, but i'd venture we are a long way from solving any position.