When Chess gets "Solved"...

It occurs to me that chess might not be a puzzle, and by that logic, might not have a "solution".

I don't think chess will be solved for at least a few hundred of years and even if it is, so what? Only computers could take advantage of it since a human can't always know the best move for every single position.

Kami5909 wrote:

It occurs to me that chess might not be a puzzle, and by that logic, might not have a "solution".

I don't agree. Chess has, according to definitions of how players move the pieces and the stated goals of the game, certain moves that are preferable to others. I think that means that it is "solvable." Unlike most puzzles, though, I bet there's going to be way more than one solution.

solving chess would have no effect on human chess. nobody can even play current tablebase endings correctly, so the 'solution' would be only of slight academic interest.

"you managed to calculate all positions? wow... so what's the verdict, a draw? nice. ... *awkward silence* ...sooo... anyone up for a pint?"

but it would kill computer chess for good. there's no point in putting computers against each other if the result is known before the game starts. same with centaur chess.

like the old saying, just cause we have forklifts, we still weightlift

We had a pretty big discussion on this topic a few months ago, I believe. I'm too lazy to search for it right now. In a nutshell, it won't happen in my lifetime. Probably won't happen in my child's lifetime, either. The real problem is storage space for the solution. I forget what the estimated storage requirement was, but I remember that it was crazy huge. WAY beyond our current feasible capabilities.

Besides... why bother? Maybe for the academic challenge, I suppose.

What they REALLY need to do, is find a solution for quickly and easily removing crabgrass from my Bermuda lawn without killing the Bermuda. Now there's something that would interest me. LOL

I wrote a piece on this a while back. I'll summarize. Chess IS solvable. The fact that there is a FINITE number of pieces, each with a finite number of moves, in a finite space, MEANS it is solvable. Just as a rook and king vs king is a SOLVABLE mate in 14, the more pieces you add, the higher the number of total moves. 10 to the 120 power TOTAL moves. Once ALL the moves can be analyzed, a solution can be found. The problem is that even an aircraft hanger full of ROADRUNNERS slaved together, would take thousands of years to analyze all the moves.

The most powerful computer system in the world, built by IBM for the U.S. Department of Energy’s Los Alamos National Laboratory is named “Roadrunner,” by LANL after the state bird of New Mexico achieved performance of 1.026 petaflop/s.

WIth the EXPONENTIAL increases in computing power in the last 20 years (commodore 64 with a HUGE 64K of RAM and a blazing 1 mhz precessor) it may be a few hundred years before all the possible moves can be computed in the span of a human life.

Just as checkers has been solved by CHINOOK http://www.cs.ualberta.ca/~chinook/project/

so will chess be. There will be the PERFECT move to be played that is stored in a monsterous database that will yield to a DRAWN game every single time. No matter what move you make, the database will counter it with a move that will either result in a win for it (if your move was less than perfect) or a draw (if your move was perfect).

It wont happen in our lifetime...but it will one day.

Hmmm, well, most of the pertinent info has been pointed out here already, but I would like to add a couple things.

1) The Result- the most probable result, according to current speculation, is that the 'solved' game would be a draw, with a small but decent chance that White wins. I would love it if the game started in Mutual Fatal Zugzwang, such that it is actually a black win from the start position. This, however, is extremely unlikely.

2) Whatever the result is, there are going to be multiple "correct moves" in any given position. i.e., if the game is found to be a draw, then on any move, any move that is made that maintains the final outcome of a draw is just as good as any other move, and thus can be considered a "perfect move" from an objective standpoint.

"solving" chess? there are like 20 possible starting moves. Assuming that the computers start with the best possible moves, it will eventually end in a draw by insufficient material.

Then players may want to move to different type of chess with new pieces on larger board like the link below

http://chess.computerwebservices.net/stealth.php

cheater_1,

I remember those old Commodore 64's. :) That was the first computer I ever bought. I tell my students about them now days and they haven't even heard of them. I don't miss that thing though. My Powerbook G5 and PC tower are just fine. The fact that my Ipod has roughly 1,500 times as much memory as my first computer strikes me as pretty funny. My PC 15,000 times as much memory, and I've just got a standard machine. Processor speeds? Maybe five to ten thousand times as fast? It'll be interesting to see what we're running with 25 years from now (that's how old the C64 turned this year).

I rather like chess the way it is...

artfizz wrote:

Chess learning computers are quite interesting - provided that they have started out from not even knowing how the pieces move - in the same way that people have to. 

Maybe you like to torture new players, but I always teach them how the pieces move and the object of the game. Surely computers (even "learning" ones) deserve the same courtesy.

In fact, when new players show interest in doing so, I'm happy to give them a few strategy tips such as the rough value of the pieces, the importance of king safety along with a few checkmate patterns, the value of the center, use all your pieces together, knights are good on advanced posts, watch out for weaknesses.

I could see arguments for and against letting computers have the same benefit of our learned wisdom. On one hand, that's how people are entering the game, shouldn't it be "fair"? On the other hand, perhaps it's interesting to see if computers have the capability of rediscovering these things. But in either case, you've got to give them the knowledge of how the pieces move.

you say that a man cannot remember all possible continuastions to win chess.

but i think otherwise.

the game theory of chess is eveolving. i believe that one day a realy smart guy would figure out a new rule to wining chess. and so they would find many once good moves blunderfilled. and so getting to the wining positions by human guess, rather than machiene calculation.

another thing, maby solving chess isn't that bad, even though it will be quite sad for me. i think that once people solve chess they would be capable of creating much more complex and problomatic games, in which the human brain would extend to it's limit to win, and machiene could not calculate the outcome by the normal meens it uses today.

Uritbon - Play Go

With regards to computers solving chess - what I would LOVE to see that neither e4,d4,c4,or Nf3 are the best starting moves.  Something hilarious like a4 would really bring a smile to my face.

Apart from that I agree the solution to chess would most likely have the game ending in a draw (perfect play by both sides would seem to imply that this would happen (this is part of my arguement that white moving first isn't actually an advantage) based on current blunder free results) but I am curious as to how many of the possible starting moves when played perfectly by both sides would result in a draw?

Further more, I am curious as to how many currently sound defenses would be thrown out the window if white played perfectly.  For example, many sicilian defenses lead to such sharp tactics that some one is bound to win, yet what happens when the game shows that one side will always win as the result.  Does e4 or c5 stop being a viable opening?

The possibilities are truly mind blowing.  As I'm sure someone else has said earlier new rules of development would no doubt come in to play.

Also I think it would be interesting to see the differences in solutions between a game under modern rules and something so simple as removing the first double move of a pawn.  (After all what do you need to speed the game up for if you already know its result)

Taken from the point of view o mathematics, the solution to chess EXISTS: it's a game with well-defined rules and total information. Furthermore the number of possible positions is finite (ok, this number is very large, but less than 2 times 32 elevated 64, i.e. 427 followed by 94 zeros). So every line must be finite: an infinite line is impossible due to the 50 moves rule and the finiteness of the total positions.

Now we know this, various question arise:

Firstly, will this solution ever be found? Frankly I don't know how far recent research has arrived, maybe someone can tell me?

Secondly, if we find a solution, what kind of solution will it be? One can imagine two fundamental types of solutions: the first one is to explore all the possibilities. In this case I won't be too much worried because only computers could memorize all possible lines, and even the fastest ones would take years to examine the complete chess tree that can arise from each position. And such a computer would loose to anybody in a game of a couple of hours.

The second kind of solution would be the discovery of an ALGORYTHM which solves chess. To explain this, let me pose a simple example: take a big number, say 837460137519237641049751213481238947198347109382471.

Is it a prime number? That's a difficult question! To answer you have to divide it by all the smaller numbers, maybe in a couple of days you can tell me...

Now take the same number, is it even or odd? That's easy! It's odd! You just have to look at the last cyphre! This is a (very simple) example of algorythm that solves the problem, even if the number is very large.

In the case a simple algorithm for chess exists and will be found, it's clear that chess will loose all it's appeal...sad day that day. We would have to play baseball or invent some weird type of four-dimensional chessboard!

In any case, consider that chess has evolved during the centuries to solve "bugs" of the initial game...think about clocks, the en passant rule, etc. So one could imagine to introduce a very small difference in the rules which makes the solution not valid: no one can say that a solution found for "chess 2008" will be valid also for "Fisher chess" or some other kind of chess. Maybe a little difference could force mathematicians to solve a whole new problem, and those freaks could maybe take years to solve each different problem! And in the meaningwhile, we can happily continue playing a slightly modified chess game every year!

For instance, if they solve chess this year, I'm ready to propose "Albatro chess", or "Chess 2009", which I sometimes play with friends: You can promote pawns also to a king! When you have two or three kings, you don't loose when only one of them is checkmated! Question: in this case who is worth more, generally, a king or a queen?

It would be a gradual process with certain lines becoming forced white wins, and others  being draws.

e.g The Sicilian dragon = white wins .

I think computers can start with lines such as 1.e4 f6 - How long will it take to show that this is a forced white win? (This means there is ALWAYS a way for white to win - regardless of what black plays and these lines must be shown )

Probably not too long. ~20 years?

If people want to give up chess completely when this happens rather than playing chess  variants, that would be too bad.

Many chess variants :

http://www.chessvariants.org

http://www.zillions-of-games.com/

albatro wrote:

For instance, if they solve chess this year, I'm ready to propose "Albatro chess", or "Chess 2009", which I sometimes play with friends: You can promote pawns also to a king! When you have two or three kings, you don't loose when only one of them is checkmated! Question: in this case who is worth more, generally, a king or a queen?

This is a bit similar to this variant: Holy Grail, where after a  piece (the Prince) is dropped into the game, checkmating the king is not enough because now the prince becomes a king and that has to be checkmated. Additionally, pawns can promote to princes making it necessary to checkmate all of them! The Prince can be captured like any other piece and it cannot become King if attacked while the king is checkmated.

http://chess.computerwebservices.net/holy.php

When chess is "solved," thousands of fools will flock to memorize the "perfect line," brag about their immense chess ability, and proceed to wet themselves at the board when their opponents go "out of book."

The rest of us will continue the same way we do now, realizing that there are far too many positions in chess to memorize the perfect move to each and that a deep understanding of tactics and/or ability to analyze positions as they arise will benefit much more than committing thousands of positions to memory that may never appear in a game.  Good players today memorize openings because they are far less varied with far fewer strategies, and endgames because there are far fewer pieces, but midgames will never be worth the memorization.

After all, we already have computers that can analyze every position fed into it (though not all positions at once), and that hasn't destroyed the game yet.  I think chess will survive the rest of the computer age in much the same fashion.

zxzyz wrote:

Holy Grail, where after a  piece (the Prince) is dropped into the game, checkmating the king is not enough because now the prince becomes a king and that has to be checkmated. Additionally, pawns can promote to princes making it necessary to checkmate all of them! The Prince can be captured like any other piece and it cannot become King if attacked while the king is checkmated.

Man! I didn't know! And I thought I was the inventor of such brilliant idea! It's true there's not much to invent anymore, someone has already done it!

Anyway, I think this is the correct way to modify chess: after decades from the feminist revolution, chess is still retro: why the hell can a king have many queens, and not a queen have many kings?

More seriously: many kings is different that a single king and various princes: for example, a position in wich your king is under check and it's your opponent's move is forcefully illegal in "Chess 2008" and "Holy Grail", not in "Albatro chess"... The king is not anymore a king though, the concept of senatus would be more appropriate... Another possible modification: you're american, right? What about "Democratic chess" or "Republicain chess"?!