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

Only Chess Forum

  1. 14 Jul '17 10:36 / 1 edit
    Greenpawn34 brought up the subject of the deep blue program while on the subject of Garry Kasparov. I was wondering, would anyone here care to speculate on the outcome of deep blue vs a recent ICCF champion under correspondence time controls. With an average of 3 days to ponder a move, the human would be able to find some very strong moves, and eliminate many of the mistakes that OTB players make because of shorter time controls they must play under. Who knows, maybe the machines are not the stronger players after all.
  2. 14 Jul '17 11:44
    Hi Mchill,

    Every now and then they conduct such experiments v strong players
    over at chessgames.com. where they challenge a GM and are allowed
    to use computers. They call themselves 'The World.'

    One I saw but refused to take part in was 'Them and their engines' v Simon
    Williams.which ran from September 2013 to January 2014 The moves
    were voted on by which computer move came up with the best evaluation.
    The game lasted 32 moves with a win for the 'World.' and took over 550
    pages (pages, not posts) to complete.

    The larger number of posts and pages were due to do posts like this
    which appeared everywhere..

    "According to Stockfish-4 at 40-ply:


    <1. (-1.79): 30.e4 Bc5 31.R1g2 h5 32.R6g3 Rad8 33.Rg6 a5 34.h4 Bd4
    35.R6g3 Bb6 36.Kb1 Rd6 37.Rg6 Bd4 38.R6g3 Bc5 39.Rh3 Bb6 40.Rhg3
    Rxc6

    2. (-1.79): 30.Kb1 h5 31.e4 Rad8 32.Rc1 Be7 33.Ka2 a5 34.Kb1 Ra8
    35.Rcg1 Bc5 36.R1g2 Rac8 37.R6g3 Rcd8 38.Rg6 Bb6 39.h4 Ra8
    40.R6g3 Rad8 41.Kc1 Bc5 42.Kb1 Be7 43.Rg6 Bd6 44.R2g3 Bc5 45.Rg2
    a4

    3. (-1.79): 30.R6g3 a5 31.e4 Bc5 32.R1g2 Rad8 33.Rg6 Bd4 34.R6g3
    Bb6 35.Kb1 Bc5 36.Rg6 Rd6 37.R6g3 h5 38.h4 Rdd8 39.Ka2 Bb6 40.Kb1
    Rd6 41.Rg6 Bd4 42.R6g3 Bc5 43.Rh3 Bb6 44.Rhg3 Rxc6 45.Rxd3 Rd6


    4. (-1.79): 30.R6g2 h5 31.Kb1 a5 32.e4 Bc5 33.Rc1 Be7 34.Rcg1 Rad8
    35.Rg3 Bc5 36.R1g2 Rd6 37.h4 Rdd8 38.Ka2 Bb6 39.Kb1 Rd6 40.Rg6
    Bd4 41.R6g3 Bc5 42.Rh3 Bb6 43.Rhg3 Rxc6 44.Rxd3


    5. (-1.79): 30.R1g3 h5 31.Rg1 Rab8 32.e4 Bc5 33.R1g2 Rbd8 34.h4 Ba7
    35.Kb1 Rd6 36.R6g3 Bb6 37.Rg6 a5 38.R6g3 Rxc6 39.Rxd3 Rd6 40.Kc2


    6. (-1.79): 30.R6g4 a5 31.R4g3 Be7 32.h4 h5 33.e4 Bc5 34.R1g2 Rad8
    35.Kb1 Rd6 36.Rh3 Bb6 37.Rhg3 Rdd8 38.Kc1 Bc5 39.Kb1


    7. (-1.79): 30.R1g2 a5 31.e4 Ra6 32.R6g3 a4 33.bxa4 b3+ 34.Kxb3
    Rb8+ 35.Kc4 Bf8 36.Bc3 Rd8 37.Rg1 Rxa4+ 38.Kb5


    8. (-1.79): 30.h4 h5 31.e4 Rad8 32.R6g3 Bc5 33.R1g2 Bb6 34.Kb1 Rd6
    35.Rg6 a5 36.R6g3 Rdd8 37.Kc1 Bc5 38.Kb1 Be7 39.Rg6 Bd6 40.R2g3
    Bc5 41.Rg2 Rd6 42.R6g3 Rdd8 43.Ka2 Bb6 44.Kb1 Rd6 45.Rg6 Bd4>"

    Sometimes the analysis went on for 80+ moves finishing with:
    "...and now the Tablebase indicates Black mates in 17 moves."



    A response from a punter: 'Thanking Simon Williams for his effort.'

    "It may not be polite to criticize our opponent's mediocre performance
    displayed in this game. But for the sake of players on our team trying to
    improve their playing strength it should be pointed out that white's
    opening plan reflects a rating well below expert level."

    ......they were openly using computers. What did they expect to learn?

    At the end of this charade we see this:

    "Thanks Ginger GM, and congrats to our team!."

    'our team' consisted of every computer under the sun.
  3. 14 Jul '17 11:51
    You mean have the ICCF champ play a magically resurrected Deep Blue running on 1996 technology? I think it would win. You'd still need a world class GM to be competitive with it in my opinion.
  4. 14 Jul '17 14:35
    There was a article in chess a while ago. Positions where the computers were not coming up with the best moves. Trouble is these positions exist but don't do most of us much good!
  5. Subscriber 64squaresofpain
    The drunk knight
    14 Jul '17 18:32
    Originally posted by greenpawn34
    [b]Hi Mchill,

    Every now and then they conduct such experiments v strong players
    over at chessgames.com. where they challenge a GM and are allowed
    to use computers. They call themselves 'The World.'

    One I saw but refused to take part in was 'Them and their engines' v Simon
    Williams.which ran from September 2013 to January 2014 The moves
    were ...[text shortened]... ger GM, and congrats to our team!."

    'our team' consisted of every computer under the sun.
    [/b]
    Those people are the worst,
    I see them all the time in stream chats while watching live GM tournaments...

    They all think they're smart,
    being super eager to pounce on every "horrendous blunder"
    which they claim was SO easy to see, even they could see it.

    Something like:
    "How could Magnus be so stupid, 'x' move was OBVIOUSLY winning"
    and then nearly all of them quote what stockfish is saying.

    To these guys, a -0.5 centipawn loss on a move is basically throwing the game away!
  6. 14 Jul '17 19:22 / 1 edit
    Originally posted by greenpawn34
    [b]Hi Mchill,

    Every now and then they conduct such experiments v strong players
    over at chessgames.com. where they challenge a GM and are allowed
    to use computers. They call themselves 'The World.'

    One I saw but refused to take part in was 'Them and their engines' v Simon
    Williams.which ran from September 2013 to January 2014 The moves
    were ...[text shortened]... ger GM, and congrats to our team!."

    'our team' consisted of every computer under the sun.
    [/b]
    Sorry folks, please allow me to clarify here: I was thinking in terms of 1 (one) ICCF world champion vs 1 (one) computer program (not necessarily deep blue) in a match under correspondence time controls. I think the human would stand a slightly better chance than the OTB player under the faster OTB time controls. No voting involved.
  7. 15 Jul '17 01:49
    The human would have a better chance but I cannot see it happening.
    It would take too long, the public would lose interest and asking some top player
    to take 4 months out, well you would need to pay them well.
  8. 15 Jul '17 17:57
    Originally posted by greenpawn34
    The human would have a better chance but I cannot see it happening.
    It would take too long, the public would lose interest and asking some top player
    to take 4 months out, well you would need to pay them well.
    Yes, it would take a long time. Too bad though. I would very much like to see a sweaty little band of computer programmers, and their over hyped pile of software packing for home in humiliation after getting roundly hammered by a strong correspondence player. It's a pity people have such short attention spans.
  9. Subscriber moonbus
    Uber-Nerd
    16 Jul '17 16:04
    Originally posted by mchill
    Sorry folks, please allow me to clarify here: I was thinking in terms of 1 (one) ICCF world champion vs 1 (one) computer program (not necessarily deep blue) in a match under correspondence time controls. I think the human would stand a slightly better chance than the OTB player under the faster OTB time controls. No voting involved.
    I agree that a strong correspondence player would have a better chance against AI chess than an OTB player. Two reasons: OTB time controls favor AI, and the multitude of opening variations which are programmed into AI cannot be memorized by even the best OTB players. Allowing a human player access to books and data bases, and correspondence time limits, levels the playing field. AI chess is essentially a consortium of GMs playing by proxy. No one player would consent to playing against a roomful GMs consulting with months of preparation, while the lone player had only 90 minutes on his clock.