Will computers ever solve chess?

not sure. if your opponent makes suboptimal moves then [usually] you do not have to play best moves to win. If your opponent makes one bad move you very well might have billions of sequences to win.

[of course i am probably not understanding post #3958?]

He didn't say "it might be possible one day"...he said the number of positions that need to be traversed is "far, far" less than discussed here.  Let's be clear about what claims are being made:

"The actual number of positions which need to be analyzed is far far smaller than the numbers which have been touted here."

So.  Unless he'd like to pony up a reduced number of positions and some plausible reasoning behind the reductions, he's full of something...

That doesn't need to be done now for a claim to be correct. Just because something isn't done now, does it mean that it can't be done in the future. Remember, some people doubted that the Wright Brothers could fly.

So.  Unless he'd like to pony up a reduced number of positions and some plausible reasoning behind the reductions, he's full of something...

Made me laugh out loud btickler, ok you got me I am full of it bro...

Post #3958 was just about estimating the complexity of a perfect strategy. I gave an intuitive argument, coupled with some simple points about stochastic strategies, why a winning strategy needs to be able to cover a very large number of positions. (This would not be true, for example, if the second player's play was very predictable).

One thing that's interesting, is that it appears that very well-played games (above Grandmaster level) may be much longer than typical games between experts. Most of us are aware that Shannon used 40 moves as the average length of a chess game, but the recent tournament between the top-2 engines has an astonishing average game length of 100 moves! (details: stockfish-wins-chess-com-computer-championship).

This could mean a few things: As we search for "more perfect" chess games, the complexity may increase dramatically due to the much longer length of these games.

But on the other hand, this doesn't discount the possibility that a much shorter perfectly played game exists. In fact such a game may already have been played (but we don't know it). The challenge is proving that it was in fact perfectly-played. Proving this would require (for example) testing every possible refutation. But other methods exist, such as deductive analysis, neural network synthesis, and so forth.

Shannon's famous paper of 1940 left things hanging, with the last chapter entitled "Another Type of Strategy". He made it clear that simple brute-force analysis is not strong enough solve chess, but the question remains completely open as to whether other strategies exist.

Note the wording, "is far far smaller", not "may possibly be far far smaller".  So, it's a definitive assertion.  Clearly, ChessSpy knows something nobody else seems to have discovered in 200 pages of posts.  Show your work, or dial back your exaggerations/hyperbole...those are the 2 viable options here.  Continuing to let your unsupported statement stand as-is is like letting your clock run out in Live Chess:  cowardly.  Play on or resign.

This is what @chessspy said:

"Suggesting that a computer HAS to use a brute force approach (analyze every possible move) is somewhat facile. It is quite possible that some improvement of the alpha-beta pruning that already exists will be sufficient to solve chess..."

This is what you said:

"If it's so "facile", go ahead and show us how to eliminate, say, 30 orders of magnitude in the number of positions..."

Here is the merit of your demand:

No merit.  Not doing this isn't proof that it won't be done in the future.

The quote I gave is the relevant quote for the point *I* was making.  The only reason I leveraged "facile" is because it shows the hypocrisy of his opinion of the too-simple "other side" of the argument versus his own too-simple and completely unsupported assertion. 

Maybe you should stick to losing your chess variant matches and such.  Logic is not your strong suit.

 Come come Mr Tickler.

There is no need to be harsh.

What I said is quite reasonable, but I will reiterate if you need clarification:

 Let us assume that most games are decided by move 40:

Most opening books give a number of quite playable options at move levels up to 20 or so ply:

Endgame outcomes are decided for all positions with fewer than (7?) pieces.

40 -(10+7) leaves a lot fewer positions to solve than 40 moves do,

You leveraged facile because it went against your already lost argument that brute force is the only way to evaluate chess. Others moved on a long time ago, while you're still stuck at pre-Shannon (1949) game knowledge.

You seem to be making a lot of assumptions here. The one about opening books does not hold water. They may give a number of playable options up to move 20 but these are very usually not the very best options. [i know as have had opening books published myself]

If an opening book gives a line up to move 20--it is usually just to give the reader an idea [one idea] on how to play a very certain line.

Also most games are decided by move 40--this is only true as most games are played by very unskilled players who make blunders very early.

 Hi David,

Yes, I made a lot of assumptions.I felt it was easier to understand if I used concrete numbers but all I was trying to show was that the complete and total number of possible moves in a game of chess, which is a very large number is not necessarily a barrier to a computer solving the game.

What I was trying to show was that as some possibilities which are known to lose, or lead to a clearly inferior position from the opening can be discounted, and a lot of endgame positions are solved, the total number of positions which need to be tried are far fewer than one might at first suppose, and if Moores law is still holding then at some point, with the number of unratified positions becoming less and less, then at some point in the future chess which is necesariiy static as far as the number of moves posible to win is concerned then inevitably chess will follow tic tac toe and checkers.    

"Far fewer" is not enough. You need to reduce the logarithm of the number of positions a lot, and that is almost certainly impossible.

if computers solve chess it will be AI like Alpha go. but my guess is it will last at least five hundred years from today on until we have lets say 16 men tablebases. We can then do the rest by hand. 32 men tablebases, this will never happen. Edit, and of course they will prove what everyone knows already, that chess is draw.

 I clearly see the strength of both Lam and Nil's arguments and I agree that some new wrinkle like applying AI techniques to solving chess along with massive processing power will make some inroads as will the larger tablebases. Surely this cannot be more difficult than unscrambling human DNA?

As to forecasting the future, particularly where new technology is concerned, let me offer this as a cautionary tale.

The year is 1984 0r 5

CB radio is in full swing and the only problem is that people can listen-in, but one would expect that could be solved, and it is free. I was coming to the end of a full-time computer course and on the 'jobs' board was a vacancy for a 'carphone' salesman, selling mobile phone contracts at $2000 per year and $1 per minute of time. The commission for each sale was $400 (for $ read GBP)   these phones were only portable in the very broadest sense being about the size of a small suitcase or a very large brick. I laughed, after all, who would spend such a very large sum on something which with a bit of tweaking could be had for free? (via CB).  Well, plumbers and builders with home offices and other call-out jobs thought differently, they could not make them quickly enough to assuage demand and the salesmen made a lot of money. So much for predicting the future.

The way I see it is that there are only 20 possible games that can be played in chess. The pawn can be moved one square or two to start, so with 8 pawns, that's 16 possible moves. The knights can move to 4 different places., which totals 20. So a computer that can win those 20 games will have solved chess.  

Alpha Zero 2.0

Chess is infinite: There are 400 different positions after each player makes one move apiece. There are 72,084 positions after two moves apiece. There are 9+ million positions after three moves apiece. There are 288+ billion different possible positions after four moves apiece. There are more 40-move games on Level-1 than the number of electrons in our universe. There are more game-trees of Chess than the number of galaxies (100+ billion), and more openings, defences, gambits, etc. than the number of quarks in our universe! --Chesmayne