Please turn on javascript in your browser to play chess.
Only Chess Forum

Only Chess Forum

  1. 01 Dec '05 22:49
    this is a question from another thread (kasparov-deep blue movie), but i'm curious on this point, so i'm posting it here.

    with computer growing ever stronger, will there come a time when chess has been solved? Bowman mentioned the "tree" of chess and it got me wondering: will a computer one day solve the tree- that is, knows exactly the right move and all the correct responses to all the small branches on which a game inevitably walks to it's end.

    if the number of games is finite (excluding draw endings like two kings), and technology keeps increasing it's scope of computability, it seems reasonable to think that chess "can" be solved. is it only a matter of time? are there unreasonable assumtions here i've overlooked?

    i don't know much about computers or chess, so i'd like to hear other's opinions on the matter.
  2. Standard member Bowmann
    Non-Subscriber
    01 Dec '05 22:55 / 1 edit
    Originally posted by Darth Sponge
    Bowman mentioned the "tree" of chess...
    But BathSponge didn't comprehend a word of it.
  3. 01 Dec '05 22:59
    No, it is theoretically impossible to solve chess. However, it will not be long before computers completely surpass humans in chess ability.
  4. Standard member ark13
    Enola Straight
    01 Dec '05 23:09
    Originally posted by exigentsky
    No, it is theoretically impossible to solve chess. However, it will not be long before computers completely surpass humans in chess ability.
    That's not true, it would just take far too long considering our current and forseeable level of computer technology. It's not theoretically impossible to know the solution to chess, just unreasonable to think that we'll solve it.
  5. 01 Dec '05 23:15 / 1 edit
    Originally posted by Bowmann
    But BathSponge didn't comprehend a word of it.
    I understood. you said "the number is too large" that each position has to be evaluated individually... and then of course each choice leads to a whole new set of possibilities which creates a ridiculously large number of possible games.

    the number is too large for modern computers... but what about computers of the Future? is it reasonable to predict that a computer will one day have the power to evaluate each and every position right down the whole tree? if the number of chess games is finite, it's finite. and then it's only a matter of time before a computer is powerful enough to grasp chess in it's entirety.

    i'm posting this question to learn, so please point out exactly what i've failed to comprehend.
  6. 01 Dec '05 23:50
    The fear of a 'computerized death' of chess is pretty common and understandable, but I don't share it. Computers are excellent at the purely analytical side of chess, but in the other areas they still fall pretty far behind. Computers take all of the life out of the game, so I hope that I am right. But I take heart in the disproving of a 'drawing death', a theory held by Capablanca and Lasker. Capablanca once said ,"Within 10-15 years any good player will be able to draw any game".
  7. Standard member Ravello
    The RudeĀ©
    01 Dec '05 23:52
    Originally posted by mufin78man
    Capablanca once said ,"Within 10-15 years any good player will be able to draw any game".
    Fortunately he was wrong!
  8. Standard member Wulebgr
    Angler
    02 Dec '05 00:01
    Originally posted by Darth Sponge
    this is a question from another thread (kasparov-deep blue movie), but i'm curious on this point, so i'm posting it here.

    with computer growing ever stronger, will there come a time when chess has been solved? Bowman mentioned the "tree" of chess and it got me wondering: will a computer one day solve the tree- that is, knows exactly the righ ...[text shortened]... don't know much about computers or chess, so i'd like to hear other's opinions on the matter.
    Start with some good data with respect to computers and chess, and you'll be able to frame the question more effectively.

    Look at http://www.cs.berkeley.edu/~flab/chess/statistics-positions.html.

    How long does it take your computer to evaluate 2097651003696806 games 5 1/2 moves long? When we get to 40 moves, the number of games the computer must process exceeds 10^120.

    Will computers solve chess? Not in my lifetime.
  9. 02 Dec '05 00:27 / 1 edit
    Originally posted by Wulebgr
    Start with some good data with respect to computers and chess, and you'll be able to frame the question more effectively.

    Look at http://www.cs.berkeley.edu/~flab/chess/statistics-positions.html.

    How long does it take your computer to evaluate 2097651003696806 games 5 1/2 moves long? When we get to 40 moves, the number of games the computer must process exceeds 10^120.

    Will computers solve chess? Not in my lifetime.
    Not in anyone's life time when 10^120 is more then the number of atoms in the known universe. That means that even if you could possibly store a chess position within one atom. You couldn't physically create a computer big enough.
  10. Standard member Bowmann
    Non-Subscriber
    02 Dec '05 00:47
    Originally posted by lausey
    That means that even if you could possibly store a chess position within one atom. You couldn't physically create a computer big enough.
    Actually, the positions could then be stored within a space smaller than the earth. But that's beside the point.
  11. Standard member Bowmann
    Non-Subscriber
    02 Dec '05 00:49
    Originally posted by Darth Sponge
    I understood. you said "the number is too large" that each position has to be evaluated individually... and then of course each choice leads to a whole new set of possibilities which creates a ridiculously large number of possible games.

    the number is too large for modern computers... but what about computers of the Future? is it reasonable to pr ...[text shortened]... 'm posting this question to learn, so please point out exactly what i've failed to comprehend.
    Go and re-read my post in the thread you already pointed out.
  12. 02 Dec '05 01:23
    i reprint this from the stupid ass spirituality forum

    many millions of millenia from now,after all the grandmasters and chess engines have worked out every possible chess game move progression into a draw,or discarded the line as unsound.christ and satan shake hands and say , good game.
    aspviper666

    and the answer to ur thread is yes it will be solved but even though trillions of chess games have been played to date...all variations have not been reached yet it would take possibly millions of millenia
    wtf knows??wtf cares??not me
  13. 02 Dec '05 02:05
    Kasparov weighed in on this a few months ago on the Charlie Rose show. He said that humans would always be able to beat computers because computers will always lack an essential quality of good chess-playing that humans have -- intuition.
  14. 02 Dec '05 08:57
    Actually, it IS true. And lausey explains why, "Not in anyone's life time when 10^120 is more then the number of atoms in the known universe. That means that even if you could possibly store a chess position within one atom. You couldn't physically create a computer big enough." This was basically my reasoning too.
  15. 02 Dec '05 10:05
    Originally posted by exigentsky
    Actually, it IS true. And lausey explains why, "Not in anyone's life time when 10^120 is more then the number of atoms in the known universe. That means that even if you could possibly store a chess position within one atom. You couldn't physically create a computer big enough." This was basically my reasoning too.
    Well, actually, 10^120 is the Shannon Number which is an estimation of the game-tree complexity of chess. Number of legal positions are estimated between 10^43 and 10^50 which would still make a computer bigger than practically possible. A moot point really as Bowmann pointed out.