Will computers ever solve chess?

  That was the point, that the 'available evidence' is partial, fragmentary. No one has the whole picture. Since evidence is partial, the resulting conclusion is a belief. You can see the logic of it, right? Because the remaining evidence, that we don't have, might contradict that conclusion.

  Thus, the only thing we can say, if we're honest about it, is to say 'we don't know'. There's no evil in not knowing something, if that is the fact. Pretending that we do know is not exactly a virtue: it's born out of a fear of not knowing everything and the resulting reaction of wanting to know everything. Yet we must look at the limits of what we know and what we don't know and not mistaken one for the other.

Nobody knows. Basically your question is: how fast computers will evolve, and will they have enough resources to calculate every single move in a chess game? Maybe. Arguing over science fiction is pointless.

  First of all, there is the present moment. At the moment, the 'evidence' is partial, so right now we don't know. Smyslov's made an argument that right now it's a fact, when in fact it's not.

  Secondly, even looking into the future, can a number that is bigger than the number of atoms in the Universe ever be calculated?!

I made the argument that it is accepted as a fact. I also pointed out that there is no 100% proof yet and there may never be an absolute proof.

Evolution is accepted as a fact by biologists. It's not 100% proven, but it is the theory that guides biology. 

Chess is a draw is accepted by every professional chess player. They play the game knowing that with best play the game is a draw, and strive hard to make their opponents play suboptimal moves. If they operated under the theory that chess was a win (for white) with best play, they would approach the game differently. 

Please, read carefully what others write before stating they are wrong.

  The fact that it's 'accepted by professional players' doesn't make it a fact. They simply operate under that belief. Nothing wrong with that. Just know that it's not a fact, but a belief.

  Why, do you think professional players are free from belief? That's all they got, in fact, one system of beliefs against another's, in many a case. When asked why he played 25...f5 in the 24th game vs Karpov, in '85, Kasparov couldn't say more than 'I didn't calculate anything, I just had to open the game, that's how I play, without knowing what's gonna happen--I gambled, isn't the whole game a gamble?' At least he sees that all he has is a system of beliefs, and doesn't call them facts. And Karpov's sytem of beliefs prevented him from playing f5 himself on move 23.

  Right now, because evidence is partial we cannot say what the end result is, with perfect play. We only assume. An assumption is not a fact. That's all.

  That doesn't answer the question, because the question assumes perfect play. The clock is an obstacle to perfect play on both sides.

  Yes, I am familiar with this trap: that we may not even exist and it's all hallucination. In a way, it is. But when you're out of the realm of imagination then you will never assume you're a part of it.

  When you're a part of it, psychologically, then this doubt is legitimate, it has its logic. But when free from imagination this doubt never comes up.

  Apart from this interjection, the problem with the number of atoms in the Universe deserves its own question marks: what Universe are we talking about? We cannot see it all, only a part of it. Then, how do we know the number of atoms the planets and other forms of matter contain?

There are three possible outcomes of chess. The argument that we don't know with 100% certainty that one result is more likely with best play leads to the argument that all three results are likely correct.

If Black to move wins, then from a practical pwrspective moves such as 1.d3 or 1.e3 would be best.

If White to move wins, then either 1.d4 or 1.e4 would be best.

If chess is a draw, then a wide range of openings designed to confuse the opponent would be best.

Chess players operate on the premise that chess is a draw with best play. The evidence is currently overwhelmingly in support of that premise.

We all know that a single decisive line would prove that premise wrong, but there is no candidate for that single line. There is absolutely no evidence that there is a single best line that leads to a greater advantage than any other line.

 I don't feel very optimistic towards your idea of calling a truce. But many variants exist already that change the objective of the game, you may want to look into Three Check (third check wins the game) or King of The Hill (checkmate or moving the king to a center field wins) and other chess variants

This would be a reasonable prior belief (before having any empirical knowledge of chess), but would be a very poor guess based on the empirical evidence. If you were to offer 1:1:1 odds on bets similar to that, you would be fleeced.

It's an interesting question what would be wise odds to offer on the tablebase value of the opening position in chess. Most GMs would give anything other than a draw very long odds.

Not really. It won't be solved without a device to do the computation, and such a device would be defined as a computer.

Not really so. It is necessary to deal with every possible opponent move in your selected strategy. As a result, the size of the problem is something like the square root of the size of the game.

To explain that, suppose every game had 100 moves and there are N moves in each position, then the number of possible games is N^200. But if your strategy selects a unique move in each position that might occur, the number of games in your strategy is only N^100, the square root of the above number.

Of course, a good position tends to leave your opponent with less options (by capturing pieces and leaving pieces with less freedom, plus extreme forcing against the king). But it is clear that a large fraction of realistic games allow the opponent a large number of legal moves. In really high quality games, this is exacerbated by the games being balanced. If the aim is to force a draw, this might help a bit by encouraging exchanges (but the most testing opponent will then be trying to avoid exchanges to make it more difficult).

It seems implausible that a perfect chess strategy would involve less than 10^50 positions due to this freedom of the opponent. Finding it is harder, because it involves finding this unique choice of move rather than it being given to you on a plate.

A neat trick is that if you can find a drawing strategy with white and a drawing strategy with black this solves the game. The reason this is neat is because each of these strategies may only have complexity around the square root of the size of the game. Otherwise a forcing drawing strategy only shows you the opponent has no winning strategy: it does not prove you have no forcing winning strategy.

Requires too much paper. 

As I have explained, as long as you accept that chess is arbitrary enough to make efficient mathematical proofs utterly implausible, a solution is of size around 10^50. This is a lot of data.

It's best thought of as a tablebase. You start with the initial position. You require a strategy that tells you what to do in each position, but the opponent can generate a large number of new positions on each move (eg 20 on move 1). The rules of chess mean that absolute general rules are few and far between: much of the interest of chess is in breaking simple rules to win games (even if most of it is about respecting those rules).

Your strategy can define just one move for you each time it is your move. But it has to be able to deal with every legal move the opponent might play. So it is not a path: at its simplest it is tree where each branching is an opponent move (and your unique moves are located between these branches).

The people who do not understand chess want to solve chess.

  If one doesn't know someting to be 100% correct, then it's a belief. A belief always comes with doubt in mind. After all, not being 100% correct means that Black could win, with perfect moves. My point is to see the difference between a fact and a belief, something with far deeper consequences in life than in chess.

You give the impression of not understanding the difference between practical play and solving.

I suggest reading up on the solution of the game of checkers, which was achieved in 2007. The game of checkers is a lot smaller than chess, but still massive computation was needed , and the principle I explained was what made it possible (i.e. you do need to cope with all opponent moves, but you can ignore a lot of your own).

See Solved games

Also The solution of checkers/draughts

No computer can do it cause the number of possible games is more than the elementary particles in the visible universe loool

YuriAdamov wrote:

No computer can do it cause the number of possible games is more than the elementary particles in the visible universe loool

Yes! The complexity is a lot more than the aSquare of that for draughts, and solving that took many years and hundreds of computers.