i believe it will eventually happen. although im not sure if white moving first is enough of an edge. but chess will eventually be solved. BUT all we have to do is change the board just a little bit and that will lead to a more facsinating and new varient of chess.

I have my doubts we (Homo sapiens) will be around long enough to see the technology needed to "solve chess" developed. The energy/storage implications shouldn't be underestimated, notwithstanding quantum wizardry.

We have to know how much progress will computers make in next 100y to be able to say if chess can be solved. I do not think we can say for sure. I believe  somewhere next 100 years, AI might reach a point where is smarter than all humans on earth. That means, he or she can create something that we think is impossible and completely change physics and computers which might bring to huge progress in computers... so imagine processor power of year 2500 is reallity in 2100. Then chess will be solved in less than 100 years from now.

I do believe also in middle of century in middle game with few inaccuracy by black white will announce mate in 40 or so.

Although the thread is quite old I think it is still the right place to discuss as I don't want to open threads for nothing.

Whenever I come across a discussion on solving chess there are only two possible outcomes: White wins (see also Futurama robots) or a definite draw as a general opinion which I used to share.

Did it ever occur to you that we could also deal with a Trebuchet position? To support this point I want to link to a smaller version of Japanese Chess (Shogi) which is called Dobotsu Shogi ( https://de.wikipedia.org/wiki/D%C5%8Dbutsu_Sh%C5%8Dgi )

This one is already solved and it turned out to be a Trebuchet position where the one who moves first loses.

What do you think about this? In principle it is White who creates a weakness of a hole first.

While such a situation as you describe is theoretically possible, it is almost certainly not the case with chess. If there was a way to gain an advantage as Black, it would have been discovered by now, and it would have been discovered by humans long before engines came along. 

So, while it's possible, the odds of chess being a win for black, or even the odds of chess giving black any sort of objective advantage due to the first move are astronomically small. 

For deterministic computers with any architecture, the answer is a fairly safe "no" for a very long time, not because it is too complicated, but because it difficult to justify the investment.

To solve a 2 player game it suffices to provide a strategy for one player that can cope with all responses by the other. As a result it is something like the square root of the game tree complexity that indicates the computational demands. For English draughts (8x8) this is about 10^31, and the square root is a manageable less than 10^16. For chess the game tree complexity is 10^123, so the square root is more than 10^61, which incomparably more ground to cover. For example, the world's fastest supercomputer can only manage 33 quadrillion FLOPS per second. Let's forget that each position requires a lot of FLOPS, and observe that 10^61 is about 10^45 times bigger than this. So the time needed to solve chess by the massive Tianhe-2 would be much more than 10^45 seconds, which is about 10^38 years or about 10^28 times the lifetime of the Universe. Do you fancy paying the power bill? (Note this is a large understatement because of the simplified calculation).

If Moore's law could continue for ever (in truth, it has already stalled, with increased power being due to parallelism these days, which means more cost, more power, even with improved efficiencies), it would take 2 centuries before the world's largest supercomputer would be powerful enough to solve chess. Suffice it to say supercomputers are not going to continue past where they require the power of a country.

With Dobotsu Shogi it is the same as the two Kings have the same movements available as in chess. Nevertheless it is a Trebuchet. I don't think that a possible perfect game would necessarily end up in a king vs. king position. This is in my opinion already an oversimplification but of course it is possible.

Even for this small variant the win is a long series of single moves for the winning party which - maybe cannot easily been found by humans as well. This can be also seen for reversed openings. Sometimes the advantage of an extra tempo vanishes as one needs to commit for a setup with White which would still be covered for with Black resulting in better counteraction for Black and so on. But I also agree with Smyslov Fan that the probability is very small - but amusing.

Many games get solved in far more manageable ways than direct enumeration of response and move-tree pruning.  Mathematical invariants have been used to solve quite a number of games, and their uses have only grown.  In fact, invariants have solved infinite games, where the decision tree could never be handled by physical hardware.

I think the "large enumeration" fallacy is the most frequent error in these kinds of discussions.  We already have numerous chess invariants.  They are used to prove endgames all the time and problem pieces all the time.  If chess is going to be solved, it is far more likely it will be done using invariants, possibly in conjunction with a system strategy.

Good luck with the "mathematical invariants" approach in chess. The problem is that the rules are so arbitrary and relatively complex that there are not going to be any useful truly general rules beyond those that constrain the evolution of material in a game (pawns do not go backwards and require captures to change columns; pieces only leave the board except pawns promoting; bishops don't change colour). Chess has left-right symmetry (apart from castling), but that's a minor saving.

True, you could get very reliable probabilistic results by pretending you know there are no exceptions to some general rule (or even that there are unlikely to be any examples), very much like statistical primality proofs, but this is not solving a game, it is merely finding the answer with high probability.

Very relevant to this problem is Landauer's principle, which determines the minimum amount of energy required for computations according to the laws of physics.

One elementary operation requires k T ln(2) Joules, where k is Boltzman's constant and T is the absolute temperature.

For example, suppose a computer was at 1 Kelvin and did 10^60 elementary operations (a number that appeared above, which should have a large multiple to be fair), the energy requirement would be about 10^37 J which is many trillions of years of total human energy consumption.

So clearly 1K is not cool enough. A problem is that in order to operate at stupendously cold temperatures, it is necessary to compute a lot slower. This is not what we want, with a problem that is due to take a trillion years already.

But it has already been a fruitful approach for quite some time.  We do know useful invariants for all the pieces, and have used them to provide mathematical proofs of many endgame strategies without the need to consult tablebases.  We can calculate the minimal number of moves a piece will need to reach a given location without enumerating all possible moves, and even account for blocking moves, using equations on the board data.  We have generating functions for a variety of piece combinations and are building rules for piece collaboration that accounts for more and more complex circumstances.

So it's not as if this is a stalled approach, or one that hasn't been already shown useful.  And it is far more likely an approach to solution than enumeration.

I now have a better idea of what you mean: improvements in efficiency based on constraints analogous to those I mentioned. But wouldn't you agree this can only reduce the logarithm of the complexity by a modest factor? This may be great if you need an improvement in speed of 100 but not so good when you need to reduce 10^60 to something manageable.

10^58 is a hundred times better than 10^60! 

I am in the habit of thinking in logarithms, when it doesn't look so good. 

The problem here is not that we would be dead before a computer has solved chess. The problem is that the Universe would be dead. 

Well, no.

Moore's Law has already failed for desktop computers.

At the rather steady rate at which the speed of supercomputers has been increasing (by massive parallelisation, massive budgets and massive power consumption!) it would be a couple of hundred years before they were able to solve chess.

However, this will not happen because the limitations of physics get in the way (only a short distance along that progression, such a computer would need a country's electricity grid to power it.)

No

No

Perhaps, if someone as smart as a computer could make a computer that would do it.

really complicate, Ins't are a only way to win, or draw, or lose, always, the computers have to follow a pattern, Because that work with numbers, And, to calculate all the variants, and do it perfect, the computer can't solve it, that implicate, we solve it first, all computer need programation, but, we don't need calculate all the variants of the chess, because, some combinatios isn't are the best  way, to the knowledge, but the computers have to learn how think, maybe have to learn some combinatios, but, sound impossible, but, yes he can do it, but that take too much time, the combinatios all have to be learned, why? because, the computer have to know how play that, but now the knowledge of the chess is really limited, but now is impossible to do it, if we will can travel into the time, why not solve the chess?, where the combinatios of the chess is much little of a infinite, and that find into the time, how we can control the time? I say, the computers, can do it, but, we will never do it.

I didn't even realize chess had a solution or an answer. The game of chess, what people have thought of as the best moves in scenarios, and whats accepted as the best strategies has evolved ever since chess was invented. And I believe it will always continue to do so. Why wouldn't it? There are so many different combinations and variations that people will be having new ideas on how to best their opponent as long as people are playing the game.