Will computers ever solve chess?

To the caveman tic-tac-toe would be a complicated game. Playing it requires abstraction and strategy. Today young children solve tic-tac-toe (even without testing every possible game) learning that neither side has to lose. Today's chess will be tomorrow's tic-tac-toe.

Speaking of cavemen, where has Urg been?

So we went from not being able to solve tic tac toe, to solving it (total of 255168 games possible). I'd say the Greeks were able to solve this, so let's say the problem was posed 30.000 years ago and solved 20.000 years ago.

Checkers was solved in 2009, or 8 years ago. In total there were about 500 billion billion (500*10**18) games to be checked.

Now we plot this table in excel (because more advanced methods of prediction would be too damn hard for this thread), with t = 0 at 30.000 years ago:

0, 0
10.000, 255168
29.994, 500*10^18

And we set up an exponential curve (because that's how technology seems to develop, now doesn't it?) Now this returns the equation
positions = 0.00573*e^(0.00176*t)
with t in years

As discussed earlier, the absolute freaking minimum amount of chess games to be checked is 10**120. This returns t = 159.927. Keep in mind t = 0 was set at 30.000 years back so we subtract that. Now this means we might be able to solve chess in... 129.927 years from now*!


*being extremely generous to our predictions of what is even possible, but hey, who knows right? This is a number I would feel okay with on agreeing on to be fair. Yes, chess can be solved, the technology will be here in 130.000 years from now

Solved doesnt mean one side has to win. Solved just means that the end of the game is known before it starts. Assuming perfect play,  one guaranteed outcome, and assuming one side makes any mistake, another guaranteed outcome. Tic tac toe is like that. It is solved. that doesn't mean the first person who moves wins. It just means the end result of the game is known before it starts.

Then it's not really solved.  What if I play the 2nd best move, which is only .01 less of an advantage?   This is a ridiculous topic.   No game can ever be solved.   They are games, not equations.

Jeff

Maybe it's just that you dont like the definition of the word "solved". A solved game just means that the outcome is known, assuming some conditions like perfect play, one side makes a mistake, etc. That's what solved is. Checkers and tic tac toe are solved because there is now no guess work if those games (or puzzles) are played perfectly. Think of it like another problem.  Divide 8 into 24. It's a problem that can be solved. It doesn't mean you have to win, it just means that if you do the problem perfectly and dont make any mistakes, the outcome is known before you begin. Chess isn't like the super bowl, where one side has to win. Chess is like a puzzle, where two people are competing to come as close to solving as possible.

I forgot to comment on your question. In the example of checkers where you play the second best move which is only .01 less of and advantage we now know the outcome before the next move is made. You will lose. Solved means if you ask ANY question like that, there is a correct answer. There is no guessing.

When a move has a different result, the difference in value is huge, not 0.01. Small differences in value are most common a matter of computers not being able to calculate to the end of the game (and occasionally examples of them getting the assessment wrong). 

Maybe someone needs to define the word "solved" then, as it applies to this topic.  Does "solved" mean that white wins no matter what black does?   Does it mean that its a draw no matter what white does?   Does it just mean that there is a perfect reply to every possible move, but the outcome can go either way?   What exactly are we talking about here?

There are different types of solution. One is to determine what the correct value of a game of chess is, with perfect play. Another is to find a strategy to achieve the best result for one player (like the solution of checkers). The strongest type of solution is a full tablebase of every possible legal chess position. While such a thing can be imagined, it would require an astronomically large device to store it, and cosmologically long time to calculate.

First, we know with optimum play for both sides --the game is a draw.

There is no absolute proof of this but the evidence is overwhelming. Any strong player knows the game is a draw.

To "solve" means to look at all the variations and this is not possible.

The sun will explode and the earth devoid of human life before the game of chess will be "solved".

Better yet, a child savant "solves" chess. Don't laugh. The human mind has a greater chance of it than any machine.

I'm not talking abut a losing move,  I'm talking about the 2nd best candidate move, in a position.   Sometimes, Stockfish even comes up with two best moves, that have exactly equal point evaluations.  Although, I don't believe that Stockfish is very accurate in it's point evaluations.  I analyzed one of my games (after it was over), a couple days ago, where it said that one side would have a +1.5 advantage if move A was played.  However, after I made moves A and B of that variation, on the analysis board, then it only showed a +0.2 advantage for the same line.  Apparently it's move horizon (ply) all of the sudden saw something that it didn't see on the previous move.  So, those evaluations are only very rough estimates.   Even if we develop quantum computers with a trillion trillion MB and a near perfect chess program, I expect to see this same horizon effect.  Although, it probably won't vary as much.   

Yes. Checkers has been "solved". The 1st side to move is able to force a win in all variations.

Sometimes a human player can look at a position that the very top chess engines cannot solve--and that human can solve the position in just a few minutes.

Stockfish's assessments are better than any human player, by quite a long way. But, as I said, it is a crude approximation to true assessments like a 32 piece tablebase would have.

If you give it more time in your position, it will get a more accurate assessment.

The game of checkers has been solved and it is a draw--just as tic tac toe has been solved and it is a draw.

Many posters here have pointed to the impossibility of solving chess by brute force, (checking all possible positions).  Intuitive shortcuts are the only possibility, which the human brain is capable of but not computers. That is the goal of AI, intuitive thought by a machine. AlphaGo is well ahead of chess programmers. But the research is moving past the game of Go, to more practical and benificial tasks.

While true in exceptional cases, I don't believe human players can achieve results in games against the top engines any more. I think they are too smart to allow drawing strategies to work these days.

I have found a win in a position which the top chess engines did not find.

Also a draw in another position which the top chess engines did not find.

However, you are very correct, the top chess engines would almost always beat me badly. They would beat Magnus. They have some "unfair" advantages over us humans.