Will computers ever solve chess?

I read it in the news.

https://psmag.com/the-artificial-intelligence-that-solved-go-52abb6188f7a#.vwxu7djzx

That's a misleading title, go is nowhere close to being "solved" in the sense that it plays a perfect game, that engine even still lost one of its games.  Google solved the problem of creating an engine that can beat the best humans at go, that's very different from solving the game of go.

Read this thread and you will understand why this isn't true: even a 10 or 100 times increase of calculation power, something we will never accomplish in the next 10 years, is only a drop in the bucket regarding the problem of solving chess

Gee, n9511 you said.

"That's correct. RPaulB's logic is silly. If you think about it for a few minutes, you will see why."  and the only logic that is silly is the one you used, it is spelled   S I L L Y   if you can not find it.   I assume you were referring to; Best moves ALWAYS lead to a Draw.    So your logic is best moves lead to a lose, sometimes, for which side by the way ?   Oh only Black can lose with best moves.  nice logic there too.

This is probably a bad attempt at trolling, but I'll respond anyways. You describe here how engines analyze positions, they do exactly this, yet they have never been able to prove a draw and of course neither have you. The problem is first and foremost that the amount of possibilities is too large to realistically construct a "proof" with this method, and second, you just assume that the best variations you find this way will be a draw, but this might just as well be a win for either player. Again, we don't know for sure until this has been done, but even if we can calculate all these positions before the solar system ends, we could never store the result of this calculation, because if we could store one position per atom, all the atoms of our solar system are only a fraction of the amount of storage space required....

Thanks.   I said there is always a draw.  You said there may not be a draw.   Where is your proft that what you said is correct ?   Maybe look at it this way.  We have the same problem in 960 chess,  Only there are 960 different starting positions.  So we have to solve it all over 960 times.  Lets start with just 20 different positions and let Black move first.  With your logic can Black force a win ?   Well that is EXACTLY the case if White moves first in a regular chess game.  There are 20 starting positions and Black moves FIRST.

Umm, no, there would be an enormous amount of overlap.

In fact a 32 man EGTB solution for classical chess would very nearly be a solution for 960 as well.

Here is another case.  "We don't know for sure until this has been done".  Did you apply your logic here ?   That would mean that you have done EVERY THING and now know you statement above is true.  You are doing exactly what you said I can not do,  your are making a general statement as if fact.

I think 960 is called 960 because there are 960 starting possitions.  I never counted them.  If there are less you could be correct.  Did you count them ?

The most reasonable argument that comes to mind for chess being a draw with best play is comparing the opening advantage to the drawing margin of endgames. The opening advantage is very small compared with what it takes to win most endgames.

Also one that I came up with myself that I like:

To argue chess is a win for either side is to essentially argue the initial position (or the position after white's first move) is zugzwang. And what might we model for zugzwang? The fewer pieces on the board, the more likely zugzwang is, so the initial position is among the least likely.

Another argument I made that I like is that winning positions correlate with asymmetry and forcing moves. And again, the opening position has neither of these.

---

These are not proofs, just reasonable observations.

There is no proof that there is no draw, and there is no proof that there is. I accept this fact, you just assume there is a draw and reason from there. Of course it is likely there is a draw, but the honest answer is simply that we don't know (yet). If we have a 32 men EGTB every possible position is solved, except that in chess 960 there a different castling rules, so yes, when you change the rules you'd have to start over, although in this case only a subset of the positions would have to be calculated as the overlap will be huge

I am stating a fact that takes the form of a general statement, yes, but it shouldn't be that difficult to see its truth, there is simply a difference between beliefs and proven truths and you can't treat them the same way. It's just like with mathematics, you may want to take a look at Fermat's Last Theorem for example, it has always been very likely to be true for centuries and mathematicians treated it as such, but they didn't know for sure until the proof was delivered. In the same way you don't know for sure that chess is a draw until you have a proof, even though it's very likely to be the case.

It may be useful or it may not, what makes you think it is?

Mr. Spock has already proven chess is a draw.  

But isn't it the point of solving chess exactly that, to know what positions are a wins, losses and draws?

Raise your hand if you think solving chess will ruin the game?

Now raise your hand if you think you can remember enough patterns that solving chess will matter?

Case Closed...

As a dolphin, I find this method of polling very unfair.

Well done Sir...well done!

Because the map is larger than the observable universe