I beg to differ. If chess was solved as "winning" for any of the standard 6 openings (c4-f4+Nc3/f3)then it would only be for black, because black has the information advantage of knowing white's first move, white doesn't know how black will respond.
I came to this conclusion when Nakmura lost to Stockfish as white with +4 moves (I believe he opened with the King's Indian with Stockfish in the starting position).
Will computers ever solve chess?
Sort:
EdwardKingSolomon wrote:
I beg to differ. If chess was solved as "winning" for any of the standard 6 openings (c4-f4+Nc3/f3)then it would only be for black, because black has the information advantage of knowing white's first move, white doesn't know how black will respond.
I came to this conclusion when Nakmura lost to Stockfish as white with +4 moves (I believe he opened with the King's Indian with Stockfish in the starting position).
Black does not have a forced win. White does not have a forced win.
Alpha Zero told me this.
Chess is dead. Prepare yourself for memorizing 1000000000-moves opening preparation. It's over, too late no more chance you lose.
Dec 26, 2017
0
#2645
EdwardKingSolomon wrote:
I beg to differ. If chess was solved as "winning" for any of the standard 6 openings (c4-f4+Nc3/f3)then it would only be for black, because black has the information advantage of knowing white's first move, white doesn't know how black will respond.
I came to this conclusion when Nakmura lost to Stockfish as white with +4 moves (I believe he opened with the King's Indian with Stockfish in the starting position).
You should be able to see your thinking is flawed by considering the position after 1. e4 (or your favourite move) only and arguing that because white has the "information advantage" while waiting for black's move, he has the advantage. The truth is that moves can have positive value: the situation where a player would prefer to pass is called zugzwang and is not the norm.
And by logic the game is solved.
Alpha Zero solved chess. Solution: Lock the opponent's bad bishop out of the game, sacrifice material for a development advantage, force zugswang and then win in the endgame with the better king position. Such simple concepts, yet so difficult for a human, due to our limited calculation abilities.
It would not surprise me, but even then, a human vs. human position has to have clear plans. The ultimate moves may be just too chaotic for the human mind to get to grips with. An ultra-sharp collection of 'if, thens' that no-one will be able to memorise.
JeffGreen333 wrote:
Alpha Zero solved chess. Solution: Lock the opponent's bad bishop out of the game, sacrifice material for a development advantage, force zugswang and then win in the endgame with the better king position. Such simple concepts, yet so difficult for a human, due to our limited calculation abilities.
Locking the opponent's bad bishop out of the game is a strategic concept. That's why I think it isn't difficult for a human to grasp and execute.
Just look at the game Winter - Capablanca. Capablanca buried his opponent's dark squared but as you can see it doesn't require a lot of calculation to see that white can't activate his bishop anymore.
Strategic understanding is more important in positions like this than calculating.
Jan 6, 2018
0
#2650
Until 20. ... Na6, it seems black is just clearly losing, after that there is a quick mate. (The reason is that black needs to clear c8 to deal with a threat of forced mate and slow the loss).
s23bog wrote:
Is Tic-Tac-Toe unplayable? Or, how about checkers?
Tic-Tac-Toe pretty much is. With accurate play, every game will end in a draw.
ImLagTheFlagger wrote:
s23bog wrote:
Is Tic-Tac-Toe unplayable? Or, how about checkers?
My question was meant to be serious. @JeffGreen333 pointed out that tictactoe is a draw with accurate play. Let's say chess is solved by computers. But I doubt that this solution (if one exists) will be just a couple of moves long and will probably too much for human memory.
What are your opinions on that?
I believe that Alpha Zero is going to solve chess, if it hasn't already. It's more about concepts than memorizing moves though. There's no actual solution that can be repeated for every game. Each game has a different solution. The fact that Alpha Zero blew out Stockfish and won 28 games and drew 72 and 25 of the wins were with white, gives us a clue. The solution is probably that white has a slight advantage, but not a guaranteed win every time. With perfect play by both sides, chess is probably a draw.
ImLagTheFlagger wrote:
I believe that Alpha Zero is going to solve chess, if it hasn't already. It's more about concepts than memorizing moves though. There's no actual solution that can be repeated for every game. Each game has a different solution. The fact that Alpha Zero blew out Stockfish and won 28 games and drew 72 and 25 of the wins were with white, gives us a clue. The solution is probably that white has a slight advantage, but not a guaranteed win every time. With perfect play by both sides, chess is probably a draw.
Let's assume AlphaZero solved chess. What does that mean for us? Just because AlphaZero doesn't like the french defense doesn't mean people won't play it anymore because the "refutation" is several moves long for sure. Engines give us variations not concepts we can work with. Right know I can't see how and how much the progress in AI affects (will affect) us regarding chess.
I learned several concepts from watching the commentary on those games. I was going to list them here, but decided to keep them as trade secrets instead. They are all things that I might try in my own games, if the situation dictates it. In fact, I'm already doing one of them and it has worked every time, so far.
ImLagTheFlagger wrote:
As long as you don't provide any proof there is no reason to believe you so your last post is not relevant to the discussion right now.
I'd rather you not believe me than reveal the secrets that I learned. I had them all typed in, but then deleted them after I thought about it for a bit. That information is just way too valuable.
Jan 9, 2018
0
#2655
What they should do is to put Alpha- Zero against itself and see how it goes and keep lowering the time until it gets interesting.
ImLagTheFlagger wrote:
Maybe you should consider chess as a career (maybe as Carlsen's second) unless it's enough for you to use your secret knowledge in bullet games lol.
Not a bad idea (on the Carlsen comment). I don't play bullet though. I'm a daily and 40/2 tournament player.
Lots of things that were "obvious" in math turned out to be false (and several obviously false things turned out to be true...)
Jan 11, 2018
0
#2658
That's probably mostly due to people thinking things were more obvious than was justified. What are you thinking of? Naive set theory? Incompleteness?
i think that stockfish is good engine but in future anyone can make a super chess engine then in that computer chess engine anyone cannot beat the engine i think