Finite Moves

Hi ya!

I'm new to chess, and I have a question, more mathematical than tactical.

Chess has a finite number of moves and positions. Sure, it's big, or even bigger than graspable, but it's still finite. If this is true, then a computer should be able, theoreticly, to analyze every possible move, from the first to mate, and always find a winning one, thus finding an unbeatable strategy.

Am I right? Is there just to many moves to analyze, but it's theoreticly possible? Or have I missed something?

And sorry if it's an old thought, but I didn't find an answer. :)

Puss och kram

Well, the strongest engines are pretty much unbeatable by humans already.

But chess is far from being solved by computer.

I think all positions with 7 pieces have been solved.

8 pieces are many times more, and a few years of.

9 Pieces many many time more than 8, probably not in a decade.

32 pieces? Very, very distant future, if ever.

I think it definitely will be one day. I suppose it's a question of processing power. If you consider where couputers were as little as 20 years ago, and then think about where they might be in another 20, 50, 80, 100 it doesn't seem unreasonable to expect that it will be solved one day. In fact, I think it is probably unrealistic to think it won't.

A computer solving chess and people being able to perform in the same way are two vastly different things though.

I think I read that a 32 piece tablebase would need a harddrive the size of the earth if you could store a gigabyte in a space the size of an atom.

definetly chess has finite positions to analyze. But just from this you can't assure that computers can find an unbeatable strategy. Because computers don't work with strategy. They analyze positions and tell you, with human criteria who is better.

What computers are going to be able to do is to decide who's better and how and the best way to win, but they will never be able to teach that to humans in any way. We have for example the checkers. If i'm not mistaken they found out the result of checkers playing perfectly 4 or 5 years ago. Still, the checkers are still played professionaly because there are too many possibilities.

Well, chess has a lot more xD

Yes, it's theoretically possible, but the practical obstacles are enormous.  If you stored and referenced every legal position in a database where one position takes just 1 bit of memory, you couldn't store it in a memory device the size of 100 galaxies... so I woudln't hold by breath heh.

Getting it mostly solved with a 90% confidence of best play by combining amazing software, hardware, and database (or something silly like this) is much more reasonable.  100% solved is as far away as wormholes and teleportation

But as rooperi said, they play very well now anyway, so the implications of chess being mostly solved, when it happens, will be more of a technological milestone than something chess players will concern themselves with.

Well, that was kind of what i was thinking, that it just has to many possibilities to analyze right now, but do you agree on that it is THEORETICLY possible?
Now, again, I don't have a great understanding of chess. Yet. ;)

Yeah, i totally agree. Computerspeeds are still rising terrifyingly fast, and has no reason to stop doing so, as i see it.
But, yes, even if it was possible to "solve" every move, noone would be able to play with that information, except a super-computer.

Wow, that's a lot. I'm hoping you just didn't pull that number out of the air, because of you didn't then, yeah, that's silly-lots! Guess the Singularity is the only hope of solving it, hehe.

That won't ruin the game though, except for anyone playing the supercomputer.