Quantum computers and chess

[Uri]

Could a quantum computer together with artificial intelligence solve chess?

The game tree complexity in chess is 10^120 (Shannon number). The number of legal positions in chess is estimated to be between 10^43 and 10^50. But there is no need to calculate every move. Some moves are obviously losing. So we could use alpha-beta pruning and other pruning techniques to prune the moves which are useless.

[Ovyron]

Some moves are obviously losing.

But for a 100% proof you can't prune these moves. There's no possible way to know if:

1. e4 e5 2. Bc4 Nf6 3. Bxf7+ Kxf7
*

Is the only way to win for white, unless you solve for this obvious black win as well.

[Uri]

But for a 100% proof you can't prune these moves.

But that's what top GMs do. They don't look at all the moves. The look only at the most promising lines which is based on their chess knowledge, experience and excellent pattern recognition skills. But computers usually don't think in general terms, they think more in tactical terms. This is because they can accurately calculate concrete variations, but have difficulty thinking in general terms.

[Ovyron]

But that's what top GMs do.

GMs don't play perfect chess...

[Uri]

I doubt if chess can be solved by brute force calculation alone. In games like Go or in chess variants like Gigachess computer brute force calculation alone doesn't stand a chance against human knowledge and intuition.

http://history.chess.free.fr/gigachess.htm

[Edmund]

Quantum computers are often presented as a solution for all our problems. But as far as I know only algorithms for some specific tasks have been written (Factorisation - decryption). Does anyone in this forum know, whether quantum computers could actually be of any use for solving chess? Unfortunally I dont know enough about the principle of these machines.

[Uri Blass]

Quantum computers are often presented as a solution for all our problems. But as far as I know only algorithms for some specific tasks have been written (Factorisation - decryption). Does anyone in this forum know, whether quantum computers could actually be of any use for solving chess? Unfortunally I dont know enough about the principle of these machines.

The fact that chess is not solved is not a problem that we need to solve.

I see no reason to be happy if chess is solved thanks to better hardware.
I think that it could be more interesting if we could simply stop the hardware progress and concentrate only on software that is the interesting part.

My opinion is that computers are already too fast.

I feel that it is unfair that programs can solve games when humans can never understand the solution.

If humans cannot understand the solution then they cannot be sure that the solution is correct because the humans that wrote the program that solved the game may have a bug in the program or maybe the hardware simply has errors and the program does not do exactly what we expect from it.

If a program does 10^20 calculations to solve some game then even one mistake may be enough to get wrong result and how we can be sure that there is no mistake.

The mistake does not need to be a mistake of the programmers and maybe there is simply probability of 10^(-20) that the hardware is going to make mistakes in every single calculation(practically programs usually do clearly less than 10^20 calculations so the probability that we see the mistakes is very small but when hardware get faster this probability may increase).

Uri

[Bill Rogers]

I have often thought about this problem and I think everyone is looking at the pacture the wrong way. There is no reason to start at the begining to think about how to solve chess. With several millions of games already played and recorded all that needs to be done is design a program to analyze these games to see what kind of moves led to the winning positions. This would be a feat in itself but I think it would work a lot faster than trying to start with e4 or what ever your first move might be.
We must remember that with a hundred years of masters playing chess a lot of bad moves have already been elimated and good lines have been tested over and over again.
This approach is very similar to the opening book that I am creating for my chess program. I have two seperater books, one for white and the other for black and only winning lines are feed in each.
Bill

[ozziejoe]

You mean "possibly winning lines". It is likely that with best play, your oponent can draw, and maybe even win

best
J

[Uri]

It's possible that chess might be solved thanks to advacements in software (in the fields of artificial intelligence, machine learning and pattern recognition) and hardware. Quantum computers could solve certain problems a billion times faster than classical computers.

Current computers don't play perfect positional chess but they are very strong in tactical chess. Humans understand positional chess better than computers.