I think it is almost conclusive that chess canNOT be solved.

[bnemias]

as far as the size of the disk farm needed, who knows...compression algorithms are steadily improving..
who can say?

Compression doesn't matter. The program will have already used the least bits possible to store positions.

[kranium]

as far as the size of the disk farm needed, who knows...compression algorithms are steadily improving..
who can say?

Compression doesn't matter. The program will have already used the least bits possible to store positions.

which program? rybka with it's persistent hash?
I'm not sure how it gets written to disk..

i believe Namilov's egtbs use a compression algoritm (huffman), the purpose of which i would assume is to minimize disk space, perhaps i'm wrong here (can't imagine it being for speed?)

[bnemias]

which program? rybka with it's persistent hash?
I'm not sure how it gets written to disk..

i believe Namilov's egtbs use a compression algoritm to save disk space...

The mythical program we are discussing. It will surely use the fewest bits required for storage, making compression useless.

[S.Taylor]

there are more (possible) chess positions than atoms in the universe?
are you sure about that?

Perhaps not the universe (10^79 atoms or so), but the earth consists of about 10^50 atoms which is about the same as the number of legal chess positions.
).

Well, that's one h-ck of a difference!

[S.Taylor]

there are more (possible) chess positions than atoms in the universe?
are you sure about that?
<:

shannon calculated 10 to the 120th chess positions

Well, since the number of atoms in our universe is about 10^80, you would still need 10^40 parallel-universes to store them all.

I didn't folllow.
So you do not agree about anywhere near as few as 10^79?

[S.Taylor]

there are more (possible) chess positions than atoms in the universe?
are you sure about that?

i'm a believer...
100 years ago they wouldn't have believed what is possible today...

It's well clever to be a believer, but of what?
Just as the numbers are finite, so is technology.

Now what about pruning?
Can obvious pruning reduce most of that number, without any risk at all of missing anything of value?

However, just because things we would never have believed 100 years ago would be seen, IS seen today, Now we need belief, to believe that what we only imagine now, will NOT happen again this time.

Unimaginable things did happen, but not EVERY unimaginable thing!

[Dr.Wael Deeb]

there are more (possible) chess positions than atoms in the universe?
are you sure about that?

i'm a believer...
100 years ago they wouldn't have believed what is possible today...

It's well clever to be a believer, but of what?
Just as the numbers are finite, so is technology.

Now what about pruning?
Can obvious pruning reduce most of that number, without any risk at all of missing anything of value?

However, just because things we would never have believed 100 years ago would be seen, IS seen today, Now we need belief, to believe that what we only imagine now, will NOT happen again this time.

Unimaginable things did happen, but not EVERY unimaginable thing!

He Shimon,you bad boy....again you opened an infinite can of warms

[Dirt]

Now what about pruning?
Can obvious pruning reduce most of that number, without any risk at all of missing anything of value?

So far as I know, this is unknown. Perhaps you can prove, without examining every possible position, something like:
An advantage of at least a queen and knight, with less than 14 pawns left on the board, and some other conditions, is always won if some material cannot be won back in less than six plies.

It doesn't look easy to me, though, and it's hard to believe that you can trim the tree enough to make a big difference.

[rlsuth]

Aren't you all getting confused with "solving chess" and storing "every possible position". The two are NOT the same. If chess is solved, you only need the one unique line that produces the best move from both sides. Even if you store the one line that leads to mate from the winning side only, it would reduce the number of positions because only the best first move would be stored, as well as the best response to each move by the losing side. In other words, you wouldn't be storing every line of every move, just a subset of the tree.

To calculate this set of moves, you'd start by calculating any line to the end, and then calculating a different move, working backwards, but pruning any line that isn't better than the best line so far found. You wouldn't store every line you'd calculated and you wouldn't need to store the entire tree at any time.

[Dirt]

Aren't you all getting confused with "solving chess" and storing "every possible position". The two are NOT the same. If chess is solved, you only need the one unique line that produces the best move from both sides. Even if you store the one line that leads to mate from the winning side only, it would reduce the number of positions because only the best first move would be stored, as well as the best response to each move by the losing side. In other words, you wouldn't be storing every line of every move, just a subset of the tree.

To calculate this set of moves, you'd start by calculating any line to the end, and then calculating a different move, working backwards, but pruning any line that isn't better than the best line so far found. You wouldn't store every line you'd calculated and you wouldn't need to store the entire tree at any time.

I don't think I'm confused, and I already agreed with this above. There are several different levels at which a game can be considered solved. Strongly solving chess without storing the result for nearly every position looks implausible, Weakly (or as the link has it, ultra-weakly) solving it is another story.

We could weakly solve chess, either prove it a draw or find out which side wins, with almost no storage. Alpha beta search with near perfect move ordering would only have to visit around 10^60 nodes. Assuming that the problem is sufficiently parallelizable for this to actually work, It would still take more processors than I think would fit on the Earth to finish in any sort of reasonable time frame.