Will computers ever solve chess?

Sort:
fburton
Arjun316694 wrote:

Dude at the rate the technology is increasing right now? Probably 5-10 years before chess will be solved.

Every desktop computer I have had starts (from cold) and becomes fully usable no faster than the one before, and sometimes slower. Go figure!

mdinnerspace

After achieving an equal position, Blacks counter-attacking chanches at winning are superior to Whites. Ya all got the question backwards. Should ask if Black has a forced win! After 1. e4 e5. White is in the throes of defeat.

Elroch

Laughing

To be serious, the statistics make it clear white has better practical chances regardless of whether the players are human or computers.

Senior-Lazarus_Long
  • Senior-Lazarus_Long
  • on 2/14/16, 6:57 AM.
  • 2Fwww.chess.com&pfname=&rpctoken=30167643" frameborder="0" marginwidth="0" marginheight="0" scrolling="no" width="100%" data-gapiattached="true">
    0
     

mdinnerspace

Statistics! Bah humbug! Good for predicting an increase in tomorrow's temperature, useless in predicting a result in chess. Get back to your charts and graphs Elroch!

Recent results for Black are greatly improved, getting better all the time. Graph that! Won't be long before Black is prefered!

mdinnerspace

Statistics do not "prove" White has a better practical chanch at winning. They only show white has won more games than black. The gap is decreasing substantially as players better understand the game. There is much to be said for Blacks counter-attacking chanches.

Elroch

The onus is on you to demonstrate something which most professional players would say is false and which contradicts the stats.

Clever players can do well with black, but the best players do a lot better with white.

Senior-Lazarus_Long

I think it's a draw with perfect play by both sides.

troy7915

  We believe, at this point in time, that one side needs to make an error in order for the opponent to win. But we don't have the complete picture. That picture may prove it possible for one side to win no matter what the other does.

  But at this point, the best we have is a belief. That's the fact of it.

Senior-Lazarus_Long

SashaRocks wrote:

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

Computers will be making computers.

troy7915
mdinnerspace wrote:

After achieving an equal position, Blacks counter-attacking chanches at winning are superior to Whites. Ya all got the question backwards. Should ask if Black has a forced win! After 1. e4 e5. White is in the throes of defeat.

  Not quite. After 1e4 e5 it is white who has that little thing called initiative. Black is a long way from equalizing if white keeps the initiative with moves such as 2Nf3, 3Bb5 and so on.

troy7915
[COMMENT DELETED]
troy7915
troy7915 wrote:
[COMMENT DELETED]

  There's nothing smart about computers. They're just raw calculation at work.

mdinnerspace

By playing the "Blacks advocate" is my way of saying the threads title makes an assumption that is false. Solvable implies White to play and win. Best play and the game will forever be a draw. I think the majority of effort has been put into finding ways to win as White. Recent good results by top players with Black indicates imo Black is finding new counter-attacking chanches.

troy7915

   Again, the big picture ( all possibilities) has not being seen. With a partial picture all we have is beliefs.

u0110001101101000
FirebrandX wrote:

1) As for chess getting solved in the sense of a 32-piece TB, that will likely never happen

2) it doesn't need to in order to play a perfect game.

Reading these forums is sometimes like:

https://www.youtube.com/watch?v=ilcRS5eUpwk

troy7915

  No one knows that. It's more accurate to say 'We don't know'.

u0110001101101000

No, the max rating should be under 10,000. And rating is not an objective measure anyway.

Back to the crazy pills Cry

Elroch

It is not clear to me what the Elo rating of perfect play is. People say computers have solved chess to a draw, but if you look at the matchups of the top two chess engines (Komodo and Stockfish), there are over 24% decisive games. And about 25% of these decisive games were won by the weaker program. It is clear that these two programs are a fair distance off perfection.

Note that, despite the inhuman perfection and never-tiring, never-wilting nature of computers, this draw rate is only modestly higher than human world championships (where 66% of games have been drawn in the matches since 1966, rather than 76% in this recent computer match-up).

SmyslovFan

The argument against computers ever solving chess because there's not enough physical space for the memory could be resolved in another way:

A solution doesn't have to have every possible answer available instantly, but it should be able to access any such answer. There are plenty of mathematical proofs that don't require every x value to be plugged in. For instance, it would be possible to create an 8 piece tablebase and work forward. You don't need to save every single possible game, just have it analysed. When a set of games from a specific position have been analysed, that position can be said to be "solved" and then the memory can be used for a new tabiya. It would be time consuming, but at least theoretically possible. 

I believe that an unbeatable computer, that is a computer that can't be beaten even by itself, will probably be available in the next 20 years. Solving chess from a practical perspective will not be so interesting then. So perhaps, chess will never be solved in the sense that Elroch means. But it will be solved in that the draw will be confirmed.