Giving knight odds to a GM not too far away?

[smirobth]

[
Perhaps against one of the weaker grandmasters, maybe. But I suspect the time will never come where a top 10 GM with knight odds at long time controls can't win almost every game against a computer no matter how powerful the program or computer.

I really reserve my judgement until I will see a selection of GM's take this on.
If they cannot affect early exchanges (and the computer should be programmed to avoid this), then I would REALLY like to see them win game after game even at long time controls.

The problem with being a piece down is you are in a damned if you do and damned if you don't situation; simply programming a computer to avoid exchanges means that when the human places a piece on an aggressive square, avoidance of an exchange will lead the computer to constantly retreat until its pieces are all very passive.

[M ANSARI]

I have no doubt that any GM can win any match against any engine with Knight odds with proper training. A knight is too much of a handicap and if the human is taught to simplify and accept exchanges even exchanges that are slightly worse, eventually he will reach a simplified position which will totally eliminate the computers advantage. For Rybka to win basically the human has to blunder, and although that can happen quite often at fast time controls, it would be less likely at long time controls. I don't like these matches because all they would say if a human won is that he managed not to blunder, and if he lost then he sucks big time.

[Ovyron]

Yes, that's why they're planning to play Knight odds against a 2200-2300 rated player.

[Ovyron]

Yes, that's why they're planning to play Knight odds against a 2200-2300 rated player.

The match is going (against the 2200 rated player) at a 5+'5 time control, and the first two games have been played. Rybka won both of them:

[Event "Blitz:5'+5""]
[Site "ICC"]
[Date "2008.01.19"]
[Round "1"]
[White "Rybka 2.3.2f16 mp"]
[Black "Rivaldo"]
[Result "1-0"]
[ECO "A00"]
[Annotator "Kaufman,Larry"]
[PlyCount "113"]
[TimeControl "5'+5"]

{1024MB, larry.ctg, RYBKA36} 1. Nc3 Nf6 2. Nd5 Nxd5 3. Nh3 Nf6 4. Ng1 Ng8 5.
Nf3 Na6 6. Ng1 Nb8 7. Nh3 Nc6 8. Ng1 Nb8 9. Nf3 {0} d5 {0} 10. b3 {0} Nf6 {28}
11. Bb2 {0} g6 {9} 12. e3 {25} Bg7 {8} 13. d4 {6} O-O {20} 14. Bd3 {0} Bg4 {61}
15. h3 {6} Bxf3 {51} 16. Qxf3 {11} c6 {9} 17. O-O {13} a5 {11} 18. c4 {10} Nbd7
{10} 19. Bc2 {18} e6 {15} 20. Rfd1 {10} Nb6 {15} 21. Bd3 {14} a4 {31} 22. Bc3 {
13} Qe7 {24} 23. Qe2 {11} axb3 {15} 24. axb3 {0} Rfc8 {18} 25. Qc2 {11} Qf8 {23
} 26. c5 {14} Nbd7 {9} 27. Ra4 {8} Rxa4 {23} 28. bxa4 {2} Ra8 {6} 29. Rb1 {8}
Ra7 {9} 30. a5 {9} Qa8 {13} 31. Rc1 {9} Ne8 {60} 32. f4 {9} Nc7 {25} 33. Rb1 {
44} Ne8 {12} 34. Rc1 {8} Bf6 {10} 35. Qa4 {10} Nb8 {73} 36. Ra1 {9} Bd8 {18}
37. g4 {4} Na6 {8} 38. Be1 {13} Qc8 {14} 39. Qd1 {6} Qa8 {12} 40. Qa4 {7} Kg7 {
15} 41. Qa2 {14} Nac7 {9} 42. Qb2 {4} Kg8 {16} 43. h4 {7} Na6 {12} 44. h5 {7}
Nac7 {8} 45. hxg6 {9} hxg6 {8} 46. Qg2 {17} Nb5 {12} 47. f5 {2} Bxa5 {37} 48.
fxg6 {0} Bxe1 {14} 49. Rxe1 {0} Ra1 {15} 50. gxf7+ {2} Kxf7 {10} 51. Qf1+ {2}
Ke7 {8} 52. Rxa1 {2} Qb8 {8} 53. Bxb5 {4} Qg3+ {10} 54. Qg2 {2} Qxe3+ {9} 55.
Qf2 {3} Qxf2+ {7} 56. Kxf2 {11} 1-0

[Site "ICC"]
[Date "2008.01.19"]
[Round "2"]
[White "Rybka 2.3.2f16 mp"]
[Black "Rivaldo"]
[Result "1-0"]
[ECO "A00"]
[Annotator "Kaufman,Larry"]
[PlyCount "93"]
[TimeControl "5'+5"]

{1024MB, larry.ctg, RYBKA36} 1. Nc3 Nf6 2. Nd5 Nxd5 3. Nh3 Nf6 4. Ng5 Ng8 5.
Nf3 Na6 6. Ng1 Nb4 7. Nh3 Nc6 8. Ng1 Nb8 9. d4 {0} d5 {94} 10. Nf3 {0} Nf6 {10}
11. g3 {0} g6 {31} 12. Bg2 {24} Bg7 {7} 13. O-O {8} Ne4 {21} 14. c3 {19} O-O {8
} 15. Ng5 {3} Nxg5 {13} 16. Bxg5 {0} b6 {25} 17. e4 {15} Bb7 {11} 18. exd5 {8}
Bxd5 {8} 19. Bxe7 {3} Qd7 {73} 20. Bxf8 {20} Bxf8 {15} 21. Re1 {17} Bxg2 {12}
22. Kxg2 {0} Qd5+ {24} 23. Qf3 {10} Qxf3+ {5} 24. Kxf3 {12} Nd7 {5} 25. a4 {5}
a5 {20} 26. h4 {23} Bd6 {7} 27. Ke4 {17} c6 {24} 28. Kf3 {11} Kf8 {12} 29. Re4
{12} Re8 {34} 30. Rxe8+ {3} Kxe8 {7} 31. g4 {16} Nf6 {21} 32. h5 {9} Kd7 {24}
33. hxg6 {6} hxg6 {7} 34. c4 {8} c5 {25} 35. d5 {15} Kc7 {36} 36. g5 {5} Nd7 {9
} 37. b3 {3} Be5 {13} 38. Rh1 {6} Kd6 {21} 39. Kg4 {2} Bd4 {66} 40. f4 {2} Ke7
{36} 41. Re1+ {2} Kd8 {21} 42. d6 {3} f5+ {28} 43. gxf6 {8} Nxf6+ {6} 44. Kg5 {
2} Ne8 {10} 45. Kxg6 {6} Nxd6 {6} 46. f5 {4} Ne8 {10} 47. Re6 {1} 1-0

[S.Taylor]

Yes, that's why they're planning to play Knight odds against a 2200-2300 rated player.

The match is going (against the 2200 rated player) at a 5+'5 time control, and the first two games have been played. Rybka won both of them:

[Event "Blitz:5'+5""]
[Site "ICC"]
[Date "2008.01.19"]
[Round "1"]
[White "Rybka 2.3.2f16 mp"]
[Black "Rivaldo"]
[Result "1-0"]
[ECO "A00"]
[Annotator "Kaufman,Larry"]
[PlyCount "113"]
[TimeControl "5'+5"]

{1024MB, larry.ctg, RYBKA36} 1. Nc3 Nf6 2. Nd5 Nxd5 3. Nh3 Nf6 4. Ng1 Ng8 5.
Nf3 Na6 6. Ng1 Nb8 7. Nh3 Nc6 8. Ng1 Nb8 9. Nf3 {0} d5 {0} 10. b3 {0} Nf6 {28}
11. Bb2 {0} g6 {9} 12. e3 {25} Bg7 {8} 13. d4 {6} O-O {20} 14. Bd3 {0} Bg4 {61}
15. h3 {6} Bxf3 {51} 16. Qxf3 {11} c6 {9} 17. O-O {13} a5 {11} 18. c4 {10} Nbd7
{10} 19. Bc2 {18} e6 {15} 20. Rfd1 {10} Nb6 {15} 21. Bd3 {14} a4 {31} 22. Bc3 {
13} Qe7 {24} 23. Qe2 {11} axb3 {15} 24. axb3 {0} Rfc8 {18} 25. Qc2 {11} Qf8 {23
} 26. c5 {14} Nbd7 {9} 27. Ra4 {8} Rxa4 {23} 28. bxa4 {2} Ra8 {6} 29. Rb1 {8}
Ra7 {9} 30. a5 {9} Qa8 {13} 31. Rc1 {9} Ne8 {60} 32. f4 {9} Nc7 {25} 33. Rb1 {
44} Ne8 {12} 34. Rc1 {8} Bf6 {10} 35. Qa4 {10} Nb8 {73} 36. Ra1 {9} Bd8 {18}
37. g4 {4} Na6 {8} 38. Be1 {13} Qc8 {14} 39. Qd1 {6} Qa8 {12} 40. Qa4 {7} Kg7 {
15} 41. Qa2 {14} Nac7 {9} 42. Qb2 {4} Kg8 {16} 43. h4 {7} Na6 {12} 44. h5 {7}
Nac7 {8} 45. hxg6 {9} hxg6 {8} 46. Qg2 {17} Nb5 {12} 47. f5 {2} Bxa5 {37} 48.
fxg6 {0} Bxe1 {14} 49. Rxe1 {0} Ra1 {15} 50. gxf7+ {2} Kxf7 {10} 51. Qf1+ {2}
Ke7 {8} 52. Rxa1 {2} Qb8 {8} 53. Bxb5 {4} Qg3+ {10} 54. Qg2 {2} Qxe3+ {9} 55.
Qf2 {3} Qxf2+ {7} 56. Kxf2 {11} 1-0

[Site "ICC"]
[Date "2008.01.19"]
[Round "2"]
[White "Rybka 2.3.2f16 mp"]
[Black "Rivaldo"]
[Result "1-0"]
[ECO "A00"]
[Annotator "Kaufman,Larry"]
[PlyCount "93"]
[TimeControl "5'+5"]

{1024MB, larry.ctg, RYBKA36} 1. Nc3 Nf6 2. Nd5 Nxd5 3. Nh3 Nf6 4. Ng5 Ng8 5.
Nf3 Na6 6. Ng1 Nb4 7. Nh3 Nc6 8. Ng1 Nb8 9. d4 {0} d5 {94} 10. Nf3 {0} Nf6 {10}
11. g3 {0} g6 {31} 12. Bg2 {24} Bg7 {7} 13. O-O {8} Ne4 {21} 14. c3 {19} O-O {8
} 15. Ng5 {3} Nxg5 {13} 16. Bxg5 {0} b6 {25} 17. e4 {15} Bb7 {11} 18. exd5 {8}
Bxd5 {8} 19. Bxe7 {3} Qd7 {73} 20. Bxf8 {20} Bxf8 {15} 21. Re1 {17} Bxg2 {12}
22. Kxg2 {0} Qd5+ {24} 23. Qf3 {10} Qxf3+ {5} 24. Kxf3 {12} Nd7 {5} 25. a4 {5}
a5 {20} 26. h4 {23} Bd6 {7} 27. Ke4 {17} c6 {24} 28. Kf3 {11} Kf8 {12} 29. Re4
{12} Re8 {34} 30. Rxe8+ {3} Kxe8 {7} 31. g4 {16} Nf6 {21} 32. h5 {9} Kd7 {24}
33. hxg6 {6} hxg6 {7} 34. c4 {8} c5 {25} 35. d5 {15} Kc7 {36} 36. g5 {5} Nd7 {9
} 37. b3 {3} Be5 {13} 38. Rh1 {6} Kd6 {21} 39. Kg4 {2} Bd4 {66} 40. f4 {2} Ke7
{36} 41. Re1+ {2} Kd8 {21} 42. d6 {3} f5+ {28} 43. gxf6 {8} Nxf6+ {6} 44. Kg5 {
2} Ne8 {10} 45. Kxg6 {6} Nxd6 {6} 46. f5 {4} Ne8 {10} 47. Re6 {1} 1-0

Sure! I am not surprised! I wish it was against atleast a low-rated GM. The title GM means much more, ANY GMis fine, but a 2200 is not teling me very much.
Were they hard fought games? I didn't manage to manually play through them, but I'm wondering if the human will win ANY or even draw.

Exchanging? Why not? But I also wonder if a 2200 player can understand how to fool Rybka into exchanging.

Oh! I wish Rivaldo luck and I'm sure Larry Kaufman knows that this 2200 player has chances, but also the blitz timing? Quite frankly, surely too much is being expected from the human! OK then, why not Larry Kaufman himself? He DID, and didn't win too much? Well? would EVERY GM win a match at these settings?

[Ovyron]

The match is going (against the 2200 rated player) at a 5+'5 time control, and the first two games have been played. Rybka won both of them:

Update (By lkaufman):

"The final match score was 4.5 to 0.5 in favor of Rybka. The sole draw occurred when Rybka had the initiative on both flanks but as she was still a piece down she repeated when Black did so. [.] It's quite clear now that a player must be much stronger than 2200 ICC to be a match for Rybka at knight odds at 5' plus 5" increment. [.]

Here are the remaining games:

[Event "ICC w11 5 5 u"]
[Site "Internet Chess Club"]
[Date "2008.01.21"]
[Round "3"]
[White "Rybka 2.3.2 q16 MP"]
[Black "Rivaldo"]
[Result "1-0"]
[WhiteElo "????"]
[BlackElo "2219"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "38"]
[EventDate "2008.01.21"]
[TimeControl "5'+5"]
(remove White's b1 knight)
1. e4 d5 2. e5 e6 3. Nf3 c5 4. c3 Nc6 5. Bb5 Nge7 6. O-O a6 7. Ba4 b5 8. Bc2
Bb7 9. d4 Qb6 10. dxc5 Qxc5 11. Be3 d4 12. cxd4 Qd5 13. Re1 Qd8 14. Be4 Nd5 15.
Ng5 g6 16. Qf3 Qd7 17. Bxd5 exd5 18. e6 fxe6 19. Nxe6 Be7 {Black resigns} 1-0

[Event "ICC w11 5 5 u"]
[Site "Internet Chess Club"]
[Date "2008.01.21"]
[Round "4"]
[White "Rybka 2.3.2 q16 MP"]
[Black "Rivaldo"]
[Result "1/2-1/2"]
[WhiteElo "????"]
[BlackElo "2219"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "58"]
[EventDate "2008.01.21"]
[TimeControl "5'+5"]

(remove White's b1 knight)

1. f4 d5 2. e3 Nf6 3. Nf3 Bg4 4. Be2 e6 5. O-O Be7 6. h3 Bxf3 7. Bxf3 O-O 8. c4
c6 9. Qc2 Nbd7 10. b3 Nc5 11. Rb1 a6 12. Rd1 Rc8 13. d3 Qc7 14. b4 Ncd7 15. e4
Qb6+ 16. c5 Qa7 17. e5 Ne8 18. d4 g6 19. Be3 Ng7 20. Be2 Nf5 21. Bf2 Ra8 22. g4
Ng7 23. Kg2 f5 24. Be3 Kh8 25. a4 Kg8 26. Rdc1 Kh8 27. Rd1 Kg8 28. Rdc1 Kh8 29.
Rd1 Kg8 {Game drawn by repetition} 1/2-1/2

In view of White's initiative and space advantage on both flanks, I hoped Rybka would play on, but being a piece down she goes for the draw despite contempt setting.

[Event "ICC w11 5 5 u"]
[Site "Internet Chess Club"]
[Date "2008.01.21"]
[Round "5"]
[White "Rybka 2.3.2 q16 MP"]
[Black "Rivaldo"]
[Result "1-0"]
[WhiteElo "????"]
[BlackElo "2219"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[PlyCount "79"]
[EventDate "2008.01.21"]
[TimeControl "5'+5"]

(Remove White's b1 knight)

1. e4 c5 2. Nf3 d6 3. Bc4 Nc6 4. O-O Nf6 5. d3 g6 6. Ng5 e6 7. h3 Bd7 8. a3 Bg7
9. c3 O-O 10. Ba2 b5 11. Be3 h6 12. Nf3 Qc7 13. Qd2 Ne5 14. Nxe5 dxe5 15. Bxh6
c4 16. Rad1 cxd3 17. f3 Bc6 18. Bg5 Rfd8 19. Qe3 Qb6 20. Qxb6 axb6 21. Bb1 b4
22. cxb4 Bb5 23. Rf2 Rac8 24. b3 Rc3 25. Rb2 Rd6 26. Bd2 Rc8 27. Be3 Nh5 28. a4
Ba6 29. b5 Bb7 30. Rbd2 Nf4 31. Bxd3 Nxd3 32. Rxd3 Rxd3 33. Rxd3 Rc7 34. Bxb6
Rc1+ 35. Kh2 Bh6 36. a5 Bf4+ 37. g3 Rc2+ 38. Kh1 Bxe4 39. fxe4 Rc1+ 40. Bg1 {
Black resigns} 1-0"

Now he's trying to find a 2400 rated opponent that is willing to play.

[S.Taylor]

The match is going (against the 2200 rated player) at a 5+'5 time control, and the first two games have been played. Rybka won both of them:

.

Yeahhh!
Yeahhhhhhhhhhh!


THAT's right! Now try it with a 2400 elo rated player!
Yes! This first experiment was with a 2200 rated one, and the results were as good as 5-0 in favour of Rybka (see reason for draw). So add on another 200 elo!

THAT should about do the job!!!

[BubbaTough]

Instead of adding 200 ELO, I would find it more interesting to add 90 minutes to clock.

[S.Taylor]

Instead of adding 200 ELO, I would find it more interesting to add 90 minutes to clock.

That's what I mean!. More should be added than 200 elo points to the human. (my comment was, sarcastic of course!)

[Ovyron]

Instead of adding 200 ELO, I would find it more interesting to add 90 minutes to clock.

I told it to lkaufman at RF, and he replied:

"Yes, I think that would be interesting too. I could probably persuade one of our local players rated around 2200 FIDE to play a couple of tournament level games here at knight odds for a nominal prize."