Will computers ever solve chess?

• AI techniques don't help at all: chess remains large.
• Massive processing power: just not enough. Converting the whole planet to computers would be inadequate.
• Tablebases? 7 piece tablebases scarcely scratch the surface and every extra piece adds a factor of more than 100 in difficulty,  Another 10^50 times as much analysis (compared to the months of processing on a supercomputer to generate the 7 piece tablebase) and we will be nearly there! [I believe we could not currently store even an 11 piece tablebase using the entire digital storage on all devices in the world, but calculating it might be a bigger problem].

Chess is much bigger than you think.  Lots closer to the literature (in the link below) than most contributors to this thread realize.  Thanks again to @Elroch for injecting a semblance of "realism" into this discussion. 

https://www.transfinite.com/content/about5

The immediate question for chess is how will AlphaGoZero affect Opening Theory.  Hopefully that theory won't be so damned choreographed, in the wake of its influence.  Perhaps its creators will eventually reveal what inside their black box.

But Maybe Not.

The answer is simple. YES.

The future is quantum computers. Centuries away maybe, but one day they´ll be a reality. And they´ll definitely be able to solve chess.

That assertion appeared about 3000 posts ago.

Advance something original, please.

42

I think it's possible AI or neural networks might be involved in solving chess (if and when it is solved). But their conclusions will need secondary analysis.

One example of AI helping in big problems is the recent discovery of a star which has eight planets (same number as our solar system). The amount of data collected in space is so vast - no human can sift through it or understand it. So neural networks were used and "Keplar-90" was found. Our best telescopes were then used to confirm the discovery - and yes, there is such a star-system (the first discovered with 8 planets).

Wiki article: (https://en.wikipedia.org/wiki/Kepler-90)

Chess might be solved in this same way. If employed, and allowed to work on the problem for some time, it may announce once side has a forced mate. It would then need to be confirmed, for example by testing every possible refutation. If the game is not too long, confirming a (single) forced mate might be within the capability of existing computers. Is it guaranteed to happen? No. Possible: Yes.

Lol...you are the one who has repeatedly used Shannon's paper as if it were some kind of proof that there are non-brute force methods for solving chess (it's nothing of the kind)...so let's cut the hypocrisy, old man ,

You aren't really too bright, are you?  Every recorded game of chess ever played by mankind and engines to this point in history is far fewer than 1 trillionth of 1 trillionth of 1 percent of possible games.  You have no concept of the orders of magnitude involved.  It's completely beyond you, or anyone else who says we can just cull the numbers to something reasonable by eliminating "obviously losing positions".  Even if you eliminate 999,999,999,999,999,999,999,999,999 out of every 1,000,000,000,000,000,000,000,000,000 positions (clearly, ermm...unlikely), you would still have to evaluate about 10^20 positions.

P.S. Moore's Law was declared DOA years ago.  Please catch up...maybe put down the CB radio.

Do watch it with the insults btickler. 

Nope, it will always draw w/ perfecto play

Not correct. Shannon's paper probably is cited in more acadamic and scholarly papers on the topic of solving chess than any other paper in academia. Btw: Brute force is a basic and simple method to solve problems. It might appear esoteric and complicated to you because it is inefficient, but that doesn't make the method complex. It may serve you well to read his paper.

no cmutuor will solve chess

Well said - It's just too inefficient to use by itself for chess. Everyone else has moved along a long time ago. Except apparently btickler.

I've already read it, as we established ago the last time or two you trotted out his paper and waved it around...

Completely wrong...there is "pruning" of any positions that can be established to make a win impossible...this was also established long ago.  It's just that that number or pruned positions is insignificant compared to the total number of positions.  You guys need to eat more fish or take Omega-3 supplements so you can actually remember what you talked about last year...

I think it would always be a win for white because white dictates the game. Which means that when whites moves, unless a mistake is made, it's more powerful than black. Eventually leading to a certain win.

And do you remember the part where not all positions of chess need to be examined to solve the game?