Upgrade to Chess.com Premium!

Nakamura defeats Rybka in a loooong game


  • 4 years ago · Quote · #21

    WilsonYiuWahWong

    Any decent player would of drawn that position. It makes no sense for the computer to sacrifice all those pawns for nothing; there isn't even compensation!

  • 4 years ago · Quote · #22

    Elubas

    Isn't Naka supposed to be like the best bullet player in the world or something?

  • 15 months ago · Quote · #23

    o0o0o0ohhhui

    why didnt he get queens why did he get all rooks

  • 15 months ago · Quote · #24

    the400blows

    Naka was just teaching Rybka some manners. Any computer with manners would have resigned much sooner.

  • 15 months ago · Quote · #25

    FirebrandX

    WilsonYiuWahWong wrote:

    Any decent player would HAVE drawn that position. It makes no sense for the computer to sacrifice all those pawns for nothing; there isn't even compensation!

    The computer was set to a very high "contempt" value for blitz play. This means it will do almost anything to avoid the 50-move draw rule, even if it means losing the game. Nakamura knew this was the case, and simply counted to 49 after he locked down the pawns. He knew that on the 50th move, the contempt setting would force the engine to sacrifice a pawn rather than draw. The default setting for the engine would have easily drawn that game.

  • 15 months ago · Quote · #26

    dpnorman

    This is actually a really awesome chess game- showing how computers can't play blocked positions at all. It should totally have been a draw.

  • 14 months ago · Quote · #27

    FirebrandX

    PinkMistzAsf wrote:
    FirebrandX wrote:
    WilsonYiuWahWong wrote:

    Any decent player would HAVE drawn that position. It makes no sense for the computer to sacrifice all those pawns for nothing; there isn't even compensation!

    The computer was set to a very high "contempt" value for blitz play. This means it will do almost anything to avoid the 50-move draw rule, even if it means losing the game. Nakamura knew this was the case, and simply counted to 49 after he locked down the pawns. He knew that on the 50th move, the contempt setting would force the engine to sacrifice a pawn rather than draw. The default setting for the engine would have easily drawn that game.

    He didn't have to count. He had no winning chances unless it sacrificed anyway.

    Yes he did, in order to win with still time on his clock. He would have been a fool not to, and even he would tell you this. He needed to be ready to recapture on move 50, and he counted to make sure of that.

  • 14 months ago · Quote · #28

    FirebrandX

    dpnorman wrote:

    This is actually a really awesome chess game- showing how computers can't play blocked positions at all. It should totally have been a draw.

    Read my comment on post #25 to see why this normally doesn't work, and as such, is NOT a demonstration of how computers can't play blocked positions "at all".

  • 14 months ago · Quote · #29

    schlechter55

    I start to like Nakamura. I didn't know he has a fine humor.

  • 14 months ago · Quote · #30

    schlechter55

    It is wrong to say that Nakamura knew from the beginning that he would win. He just demonstrated for a large part of the game, that his position was impenetrable. (He knew that computers keep giving evaluations like +0.50 or even +1.0 after - wrongly - closing the last files.)

    His humor came out when he sacrified one after another two exchanges ( without being forced to do so=. It takes high skills (more than just elo 2100) to see that the position was still equal.

    He couldn't know that Rybka would be 'provoked' and make some terribly wrong decisions (sacrifice of two pawns, to open up lines for his rooks). Because of the (still) limited scope of the white rooks and the active Black king, the white position was lost after that.

    I guess, a program of today (FIVE years later, when we also have faster processors), would reject such odd decisions with, say, ply 20.

  • 14 months ago · Quote · #31

    FirebrandX

    schlechter55 wrote:

    It is wrong to say that Nakamura knew from the beginning that he would win.

    Let me explain it more simply: Nakamura knew what he needed to set up to win. That's why he lost dozens of games in a row in order to reach the contempt exploit. I don't know how many times and different ways I need to explain this, but a lot of people commenting on this thread are making a lot of frankly naive assumptions. You first had to be there to witness how this gets done, and second, you have to know enough about engine chess settings to appreciate the exploit to begin with.

    For the last time:

    1. This is not how you can beat an engine at default contemptless settings. The engine at default settings will draw the game instead of sacrificing the positional equality.

    2. Namkura knew the engine was set at a high contempt value by the operator, and as a result, knew the engine would always sacrifice into a game-losing position in order to avoid the 50-move draw rule.

    3. Namakura saved reaction time by counting the shuffle moves, thus allowing him to near-instantly recapture the sacrificed pawn on the 50th move to the draw rule. He'd done it several times before.

  • 14 months ago · Quote · #32

    schlechter55

    So you are saying Nakamura played a match, from which he lost several first games to find out the game settings of the engine , until he finally won JUST one game ?

    This makes his achievement not look so great. I think that Nakamura might have STILL not knowing the outcome, and being in the process of guessing the 'settings' and strength/vulnerability of Rybka. May be, he was lucky to hit an ill point for once ?

    I cannot believe that the settings were put at: 'reject all threefold repetitions'. It is well-known to club players that there are cases when such absolute preference can lead to a loss of a game.

    Anyway, the game is amusing, and nice.

    If the horizon of all variants that evaluate correctly an idea/plan/strategem is too far, then the program may not find it . Because it does not know strategy and abstraction. 

  • 13 months ago · Quote · #33

    macer75

    o0o0o0ohhhui wrote:

    why didnt he get queens why did he get all rooks

    How do u get a rating of 1219 in bullet when u can't tell the difference between rooks and bishops?

  • 13 months ago · Quote · #34

    TheGreatOogieBoogie

    Impressive technique from the greatest American chessplayer in the history of the world!  I really like how his play said, "With all due respect Mr.Rybka, but your bishop won't target my pawns in the endgame while my pawns cover where my bishop can't, and I shall keep this position closed enough where my knight pair trumps your bishop pair!"

    Seemed like a lot of useless shuffling though, but of course someone like me would perceive such as "useless" while Nakamura understood the deep strategic ideas behind it. 

  • 13 months ago · Quote · #35

    richie_and_oprah

    in other recent news:

    man walks on moon 

  • 13 months ago · Quote · #36

    CP6033

    gbidari wrote:

    That's one of the dumbest games I ever saw.

    agreed why didn't Nakamura just beat him? promote a queen or two and that is game. how much time did he have?

  • 13 months ago · Quote · #37

    CP6033

    personally i don't care if  computers can beat anything. Really just forget the computers and play with people although computers are nice for analysing

  • 12 months ago · Quote · #38

    11Cryn

    It seems that someone worships the computers and has contempt for the world's greatest human chess players.  Perhaps that is due to his 1063 rating in Online Chess.

    In the end, this game showed the brilliance and creativity of Nakamura.  He willingly gave up two exchange sacs.  He was even able to embarrass Rybka, by underpromoting so many times.


Back to Top

Post your reply: