True or false? Chess will never be solved! why?

Sort:
TheGrobe

Except that it's a very flawed assumption to assert that Moores law will continue to hold for the next 520 years.

EmperorPiece

Chess doesn't need to be solved, it's not a problem. It's a GAME.

TheGrobe

The question of whether there is a forced win from the starting position or if the game is drawn with best play from both sides is the "problem". This is the question at the heart of "solving" chess.

Ziryab

Chess is 10^43 problems, a fraction of which are solved. Some of these "solutions" have changed over the past century or two. Here's one that was solved by Alexei Shirov but that continues to defy computers:

 

Shirov Test

 
Alexei Shirov found the winning ideas in a position that often has stymied chess engines. The game was Topalov - Shirov, Linares 1998.

Black to move

Photobucket - Video and Image Hosting
 
Jion_Wansu

True or false? Chess will never be solved! why?

True. Chess is too complex.

ponz111

Depends on your definition of "solved"  Most really good players know that with best moves for both sides--chess is a draw.

RG1951
ponz111 wrote:

Depends on your definition of "solved"  Most really good players know that with best moves for both sides--chess is a draw.

        I'd have thought that, in a lot of board positions, it is impossible to prove with certainty that any given move is best. Bear in mind that, up to the 25th move, for example, mathematicians say that the number of different positions that could possibly arise is astronomically huge.

fburton
NotAllowedTo wrote:

Pretty simple question: we have tablebases (perfect play / result) for 7 pieces (I think they're called Nalimov). For chess to be solved, we need tablebases for 32 pieces.

32 - 7 = 25.

Long way to go.

Or not?

Put like that, it doesn't seem that big a deal. Given the exponential nature of the beast however...

pfren

7- men tablebases are called Lomonosov, and 32-men tablebases would need to be stored some six times the atoms that consist Earth. So,  these tablebases may be invented by aliens which live in a much bigger planet... :P

Irontiger
pfren wrote:

7- men tablebases are called Lomonosov, and 32-men tablebases would need to be stored some six times the atoms that consist Earth.

Are you sure you do not mean six orders of magnitude bigger?

Even that sounds "little" to me.

varelse1

Computing power is growing exponentially. And from what I'm told, the big breakthrough is only another 6,7 years away.

Everyone who has underestimated the potential of computing power over the last 60 years has eaten their words. Try not to jump onboard with them.

Irontiger
varelse1 wrote:

Computing power is growing exponentially. And from what I'm told, the big breakthrough is only another 6,7 years away.

Everyone who has underestimated the potential of computing power over the last 60 years has eaten their words. Try not to jump onboard with them.

Moore's empirical law is powerful. But Shannon's entropy is stronger.

There is a physical limit to the amount of information that can be stored in a finite place.

pdela
pfren wrote:

7- men tablebases are called Lomonosov, and 32-men tablebases would need to be stored some six times the atoms that consist Earth. So,  these tablebases may be invented by aliens which live in a much bigger planet... :P

You don't need storage if you have enough calculate power as given a position previous ones have no influence over the game

Irontiger
pdela wrote:
pfren wrote:

7- men tablebases are called Lomonosov, and 32-men tablebases would need to be stored some six times the atoms that consist Earth. So,  these tablebases may be invented by aliens which live in a much bigger planet... :P

You don't need storage if you have enough calculate power as given a position previous ones have no influence over the game

If you are recalculating at each move, you have no storage issues, but electricity only moves at the speed of light.

pfren
pdela wrote:
pfren wrote:

7- men tablebases are called Lomonosov, and 32-men tablebases would need to be stored some six times the atoms that consist Earth. So,  these tablebases may be invented by aliens which live in a much bigger planet... :P

You don't need storage if you have enough calculate power as given a position previous ones have no influence over the game

Huh?

You either don't know what a tablebase actually is, or you have no idea how brute force calculation is done. I would rather risk to say both.

Oh, and Moore's law is dead since several years ago, but apparently some people failed to notice.

TheGrobe

Yeah, what pdela described is a perfect engine, not a tablebase -- and the only way to confirm you had a perfect engine is if you also had the 32 man tablebase to verify against.

TheGrobe

The whole point of a tablebase, by the way is to persist in storage a comprensive result set of some very intenstive and lengthy calculations so that you don't have to repeat those calculations every time a position needs to be evaluated.  Naturally, you don't need a tablebase if you can just generate the tablebase every time you want to evaluate a position.... you just need an eternity.

Irontiger
pfren wrote:
pdela wrote:

You don't need storage if you have enough calculate power as given a position previous ones have no influence over the game

Huh?

You either don't know what a tablebase actually is, or you have no idea how brute force calculation is done. I would rather risk to say both.

Well, if we forget about alpha-beta pruning and that kind of stuff, we can program brute force to be just what it means. Something like:

-to analyze position i: check if it is a checkmate, otherwise:

--sort out the m legal moves

--for j=1 to m, repeat on position i+1 that results from move j of that list

 

You only need to store the chain of moves in one line. (You can recalculate the "ordering" each time you climb up in the recursion chain)

 

Needless to say, this is ridiculously inefficient, even inflating current computing power by a factor of 1,000 (~2^10).

shell_knight
TheGrobe wrote:

Yeah, what pdela described is a perfect engine, not a tablebase -- and the only way to confirm you had a perfect engine is if you also had the 32 man tablebase to verify against.

A "perfect" engine is hard to even imagine though.  (Although I'm sure I'm preaching to the choir here).

shell_knight

Haha, ok, you could just have it calculate (nearly) forever.  I suppose that's one way to do the "perfect" engine.