Will computers ever solve chess?
Sort:
ProfessorPownall wrote:
This is disproved in several ways.
Enumerate them, then.
One reason chess will never be solved is the expense it would take to even try to solve chess.
As more and more players realize that chess is a draw with optimum play--there will be less and less reason to try and "solve" chess.
ProfessorPownall wrote:
@Elroch .. speed of calculation does not increase it's rating,
Ermm...BS. Of course it does. If Stockfish can reach 50 ply instead of 40 within the given time control, the rating will be higher, This is why TCEC championships are run on high end hardware, and not on your smartphone. Please tell me you don't really think your Droidfish app can play at the same rating as Stockfish on a beefy server...
Jishnu_Chatterjee wrote:
NO. The time and energy required to "solve" Chess is not possible in this Universe.
This is correct.
However there is a 2nd reason why chess will never be "solved"
The 2nd reason is as humans become better at chess--more and more will realize that chess is a draw with optimum play. So the amounts spent to try to solve chess [which can never succeed anyway] will not be worth the time and effort and money as we already know just is a draw with optimum play for both sides.
Computers won't solve chess.
They now consider 0.35 to be a win.
In the future, maybe 0.2 will be enough for the win
They now consider 0.35 to be a win.
In the future, maybe 0.2 will be enough for the win
Jul 25, 2017
0
#1727
Curvybanana wrote:
Computers won't solve chess.
They now consider 0.35 to be a win.
They now consider 0.35 to be a win.
No, they do not.
In the future, maybe 0.2 will be enough for the win
In the future, maybe 0.2 will be enough for the win
0.35 is an absolute draw!
Jul 25, 2017
0
#1729
vickalan wrote:
Elroch wrote:
On an entirely separate topic, I wonder how strong a program would be running on a modern supercomputer. These are about a million times faster than a high end desktop computer.
If doubling speed gained 40 Elo points, such a computer would be about 800 points stronger.
I think it is safe to say no-one is ever going to justify the budget (never mind porting the code) to test this, but perhaps there is a potential 4000 Elo machine out there!
The Lomonosov supercomputers in Moscow were used for several months to develop the 8-piece endgame tablebases. But I wonder if a major Chinese or U.S. university would ever allocate their supercomputers for a chess study though.
On the other hand, solving chess would be the type of thing that would draw worldwide attention, and bring international acclaim to whoever does it first.
To my knowledge, there are no 8-piece tablebases. The 7 piece Lomonosov tablebase (barring one dull class) was completed in 2012 and is 140 terabytes in compressed form. 8-piece ones would be huge by comparison (more than 10 petabytes).
Jul 25, 2017
0
#1730
aravinds_ll wrote:
0.35 is an absolute draw!
0.35 is a slightly higher probability of a win.
'slightly' is correct. But unless the opponent makes a mistake, not even Komodo will be able to convert it into a win.
Jul 25, 2017
0
#1732
It is a dead end to search for mistakes from the end of a game. Simple reasoning. It is called "logic". How s23bog makes his claim his beyond me.
1. A checkmate is found at move 50.
2. A "mistake" is found at move 42 by the losing side.
3. The "search" comes to an end for that game.
You are back to searching through all possible games. This time ALL games must be analyzed. There is no "pruning" available. Every branch must be searched from every possible checkmate position unless the one in a gadzillion chance is hit upon.
Jul 25, 2017
0
#1733
s23bog wrote:
Let me propose an example, if I may. A particular game is played that white wins. The game lasts about 100 moves. Black makes two mistakes in the game. The second mistake was on move 35. A pawn weakness was created that was later exploited and causes black to lose. The first mistake that black made was at move 1.
Where did white err?
1. 100 moves
2. mistake found at move 35
3. 65 irrelevent moves were analysed in a fruitless search.
The objective is to find a forced checkmate. Searching from the end to the start is
ASS BACKWARDS. The only solution by this approach is to search all possible positions, which has been stated enough times as impossible.
Jul 25, 2017
0
#1734
I say the search ends for that particular game. It ENDS. Period
Your approach demands that every possible position be analysed starting with every possible checkmate position. When a mistake is found, that line gets dismissed. the search has ended along that branch.
There is no "pruning" possible which is possible (according to some) by starting the search from the beginning.
Jul 25, 2017
0
#1735
You clowns not solved chess yet?
While you waste time on this topic I am making the game engine for my game.
Writing the code in good old C using Visual Studio and using magic bit boards for sliding pieces.
Jul 25, 2017
0
#1737
s23bog wrote:
"Get back to me when you have found a game with no mistakes as far as can be guaranteed from the end of the game. This game can be one that was actually played, or one that occurred in analysis."
LMAO
OK
Will do. When I "prove" White is a forced win I'll do just that !
Jul 25, 2017
0
#1738
Correct. Basically, the problem is retrograde analysis which at the very least includes all positions the opponent can reach against some specified algorithm for selecting moves (using such an algorithm cuts the positions down a lot).
Jul 25, 2017
0
#1739
WHAT ? I am at a loss to answer that question. It's makes no sense. I'm convinced you're head is "up in smoke" ! LOL. literally too much of your favorite herb.
Here's how it works:
1. An hypothesis is made:
Chess is a forced win for White
2. By experimentation and analysis a solution is found.
3. A "proof" is written and tested.
To suggest I need to "prove" move one is correct (before any experimentation) shows a lack of understanding the "Method" required to verify a hypothesis.
There are two sides presented:
1. Every possible position needs to be tested for a result to be verified.
2. It is possible to "prune" the tree of it's gazillion branches to find a solution.
"Retrograde" analysis is possible as Elroch points out. But it requires all possible positions be analysed. Which has been agreed is not possible.
s23bog wrote:
A single process to "solve chess" would be nice, but it isn't necessary to actually solve the game. All the work that has already been done over the past thousand years can be built upon. There is no need to throw it all out and start over "the right way". What has happened has happened.
The solving method doesn't need to just spring up out of nowhere. We can make minor improvements to what has been done. We can reorganize things to make them easier to work with. We can find little tweaks to things to make them better. It doesn't have to be all or nothing.
The problem is that all the work which has ever been done in the past thousand years points very clearly that the result of chess is a draw with optimum play by both sides.
Almost all of the very best players recognize this.
So why would anyone spends trillions of $ to try and solve chess when it is quite apparent that chess is a draw? [just as checkers is a draw].