Quantum and Super computers will solve chess before 2050. There's no need to create a 32 men EGTB. Expanding Opening Theory and Growing EGTB will meet in the middle.
expanding opening theory so that it constitutes solving chess would mean makin an opening book with all the possible move so in fact that would 32 men table base. Which is not doable.
To solve chess the NN will need to be tuned by quantity of pieces left on the chessboard. A NN for every exchanged piece combination. There are 32 pieces, but 2 of them are King's. The 7 men EGTB exist, but what about the Opening Game Table Base (OGTB)?
The game has 32 pieces, but the 2 king's cannot be exchanged. Subtract the 2 kings and the 7 men EGTB and we are left with 23 pieces to solve. For starters we might need 23 NN's. Each turned according to the number of pieces left on the chessboard. We could have more NN's if the exchanges are also tuned. So it would be 2 to the power of 23 = 2.0880468e+31 NN's. That's what the Quantum computer will need to solve chess. Now think about the file size of those NN's.
Understanding chess is all that is required. A sentient AI could change the end game rules and count the static position to determine a winner. Time controls have limits and chess needs move limits too. A game could end in 40 moves. There are time controls with minimum number of moves in a given time limit. It also needs a rule for maximum number of moves in given time limit.
Conclusion: It's because of the lack of control on how many moves we can play in a given time control that chess has not been solved. Limit the number of moves to 50 like an EGTB and there is hope of solving chess. The present 50-move rule would need to change. No chess game should go beyond 50 moves. Then the engines could determine the WDL by evaluating the end positions. So it would be like limiting the game like a clock. What is left is to truncate all games played to 50 moves and evaluating the final position. Create a new WDL ratio to solve chess.
en.wikipedia.org/wiki/Endgame_tablebase
en.wikipedia.org/wiki/Fifty-move_rule
The game has 32 pieces, but the 2 king's cannot be exchanged. Subtract the 2 kings and the 7 men EGTB and we are left with 23 pieces to solve. For starters we might need 23 NN's. Each turned according to the number of pieces left on the chessboard. We could have more NN's if the exchanges are also tuned. So it would be 2 to the power of 23 = 2.0880468e+31 NN's. That's what the Quantum computer will need to solve chess. Now think about the file size of those NN's.
Understanding chess is all that is required. A sentient AI could change the end game rules and count the static position to determine a winner. Time controls have limits and chess needs move limits too. A game could end in 40 moves. There are time controls with minimum number of moves in a given time limit. It also needs a rule for maximum number of moves in given time limit.
Conclusion: It's because of the lack of control on how many moves we can play in a given time control that chess has not been solved. Limit the number of moves to 50 like an EGTB and there is hope of solving chess. The present 50-move rule would need to change. No chess game should go beyond 50 moves. Then the engines could determine the WDL by evaluating the end positions. So it would be like limiting the game like a clock. What is left is to truncate all games played to 50 moves and evaluating the final position. Create a new WDL ratio to solve chess.
en.wikipedia.org/wiki/Endgame_tablebase
en.wikipedia.org/wiki/Fifty-move_rule
Just listen to the commentary of super GM tournaments that chess.com hosts. Many super GMs already believe that chess is a draw, and that it takes a surprising degree of error to get to a loss. Rather than sap their enthusiasm, its been freeing in a way. They no longer feel like they have to play the optimal opening as white to get some tiny opening advantage, as they now know that such "advantages" are too small to actually lead anywhere with perfect play. Instead, they are content to find a less explored corner of opening theory that, even if it doesn't provide any opening advantage, is complicated, and they know better than their opponent.
If ratings or standings is all that is important than draws count and cause rating changes and standing changes. Even if you don't change the value of a draw, you still have a winner and a loser. The higher rated player drawing loses and the lower rated player wins. There is no need to call a drawn game a half point. All that is needed is to see the rating difference and don't match two players with the same rating.
The structural balance of the position vs future mobility advantages is all that matters.
See chess is solved if it ends in 50 moves and draws are evaluated as full points for the player with the smallest rating. Rules made chess unsolvable. Rules can be changed.
The structural balance of the position vs future mobility advantages is all that matters.
See chess is solved if it ends in 50 moves and draws are evaluated as full points for the player with the smallest rating. Rules made chess unsolvable. Rules can be changed.
NO need to worry man. We'll just BAN quantum computers from chess. It's evolution tbh
never., Chess isn't about memorizing actually, Real thing is how your own Mind works, then he will or will not see some things in Rule of choosing and understanding variants. There are Obvious Positions, but You will never calculate deep enough to solve all, because it's not about calculating all possible variants, much more Important is Understanding of what happening in current position Itself, independed of how complex it is, and it can be very complex , while real powerful machines play, but even they don't have real thoughts and understanding , that human could possibly have, if they had possibility to calculate and see all variants in an instant
@Cyncko-3000 said in #26:
> NO need to worry man. We'll just BAN quantum computers from chess. It's evolution tbh
"The quantum computers are cheating!"
-Kramnik probably
> NO need to worry man. We'll just BAN quantum computers from chess. It's evolution tbh
"The quantum computers are cheating!"
-Kramnik probably
@AsDaGo said in #2:
> Solving Chess960 seems to me like a very small step up from solving regular chess.
Technically solving the position with a computer would be a small step. Memorizing 960 times the lines for in-game use would be a MUCH larger step.
> Solving Chess960 seems to me like a very small step up from solving regular chess.
Technically solving the position with a computer would be a small step. Memorizing 960 times the lines for in-game use would be a MUCH larger step.
This topic has been archived and can no longer be replied to.