Will computers ever solve chess?

Another "hello?", I guess.  I've never said it's impossible that chess will be solved in the future, only that it won't happen in our lifetimes, nor in any foreseeable future scenario using "hey, we'll soon be there" technology.  That makes it impossible at this juncture in human history.  You might as well have been born 20,000 years ago telling cavemen they could throw together a fighter plane with sticks and stones...that was also impossible.

We've also previously established this stance...you really are almost as forgetful as s23bog.

All it takes is a shift in technological ability. Once they can self-learn, there's no stopping them. We won't figure out how to make a machine that will solve chess, we'll make a machine that will teach itself the best method. That could happen tomorrow.

Why indeed -- it could happen YESTERDAY,

based on the wild-eyed ideas that inhabit this nutty thread.  

It is surprising how, 4000 posts in, people still seem confused about the contrast between the problem of playing chess well (attackable using AI, as exemplified by AlphaZero) and solving chess (dominated by the inherent computational complexity of the problem, as no approximations are useful). The latter is very similar to the problem of constructing a chess tablebase. For this problem too there is scope for cleverness, but this makes a very limited dent in a large computational problem (on a log scale ).

BTW, when I refer to a log scale, I am contrasting the difference between everyday computational problems and very extreme ones. In the former achieving, say, a factor of a hundred improvement in efficiency and speed (or space) would be a great achievement. For a huge problem, it hardly makes a dent, because it is only a small change to the exponent of a number like 2^100.

Surely it has. Steinitz said he could give God pawn and move odds, so he must have solved chess. 

One of two things will happen for computers to solve chess...

1 - it becomes capable of computing a googol positions in a reasonable enough time frame, starting from checkmate and working backwards (very much like creating an endgame tablebase). I suppose it could happen from the starting position and eventually every possible game has been played to checkmate, but it's the same method... brute force. We're obviously nowhere near this kind of processing, but once a googol is a reasonable number for a computer then we're in a different league. Although machines solve chess here, it was a human method.

2. A computer solves chess by creating an equation that gives mate in n for a given position. It can simply analyse every move, and apply the equation to each new position, to quickly calculate which move is the winning move. Is this possible? I have absolutely no idea, but if so, this is truly machine solving chess, and this is what I think *could* happen at any given time. I'm confident it won't, but confidence is not certainty.

It's rather like figuring out if a number is prime. The best method I'm aware of is essentially brute force with a hint of logic... divide x by prime numbers lower than the square root of x, and if none return a whole number, x is prime. But maybe there's a quicker way? Maybe we only need to make one calculation, instead of many. And if so, my money is on a machine figuring it out.

The above is called a 'trapdoor problem I think, easy to get into, (postulate) diffucult to get out.

Chess is large but finite, everyone agrees I think.

Computers have until the end of civilization to figure it out, one way or another.

Go which was supposed to be the one game computer programs would solve last is effectively bust, that is to say, the best program can beat the best master.

Chess is on the cusp of that. I think we will see it before too long.

What number of mathematical operations did you assume is required to solve chess to arrive at this conclusion?

That makes sense, and it's certainly possible. If the white side never makes a mistake and the black side is always defending ultimately the white side could force a win from the very beginning. But, sometimes even if the white side is up a bishop and a pawn at the end of the game it's still a draw. So the game always being a draw is very possible too.

Very true.

I don't know if this is true? is it?

Any finite solvable problem will eventually be solved. 

Thanks, but as you well know, the subtopics discussed here for 200 pages are more granular...I would think you would know that based on your numerous attempts to solve the world's energy problems with a ball of twine or whatever is laying around the house ...

Go is nowhere near being "solved" either, and chess engines were beating the best chess players long before AlphaGo was around, so in that sense, AlphaGo has achieved nothing in Go that has not already been done in chess by other computers.

Go back and find it yourself.  I'm done digging up posts every time you take another brain dump.  Or just use 10^46.7 like everyone else.  Either way.

How did your arrive at that number?

As usual, quoting the one phrase that supports your argument and ignoring the rest of the post that points out why your position is none too solid (to be kind).

Do you happen to live on a pirate's shoulder?  That would explain a lot.

There is nothing on this forum that substantiates the argument that chess will never be solved (not even within a human lifetime). Also not in academic or scholarly papers. Also not in any published work from a university or respected institution. This is despite the fact that chess is the most studied game in history.