Just discovered a Rybka 3 bug

[Uri]

In Playchess engine room, Phil Harris2 played someone who used Rybka 3 Dynamic but probably on weaker hardware.

Now Phil Harris's Rybka made two serious errors. It could probably draw the game hadn't it been for these two bad moves. Here's the game:

[Event "5m + 0s, rated"]
[Site "?"]
[Date "2009.04.29"]
[Round "?"]
[White "Phil Harris2, Rybka 3"]
[Black "Jefk, Rybka 3aqua"]
[Result "0-1"]
[ECO "B90"]
[WhiteElo "2708"]
[BlackElo "2656"]
[PlyCount "181"]
[EventDate "2009.04.29"]
[EventType "blitz"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. f3 e5 7. Nb3 Be6 8. Be3 Be7 9. Qd2 O-O 10. O-O-O Qc7 11. g4 Nbd7 12. Kb1 {0.00/0 0} b5 13. a3 Nb6 14. g5 Nh5 15. Na5 Rab8 16. Nd5 Bxd5 17. exd5 Nxd5 18. Qxd5 Qxa5 19. Bd3 g6 20. Be4 Qc7 21. Rhe1 a5 22. Rg1 Rfc8 23. Rd2 Bf8 24. h4 Qe7 25. Rdd1 Rc7 26. Qb3 Qd7 27. Rg4 Ng7 28. Bd5 Rcc8 29. Bd2 Qc7 30. Qd3 Nh5 31. Be3 Kh8 32. Ba2 Bg7 33. Qd2 Rd8 34. Bd5 Re8 {-0.20/15 35. Ba2 Rec8 36. Qd3 Rd8 37. Rg2 Rd7 {(Qd7) -0.21/14 4} 38. Bd5 {0.00/15 5} Re7{(Kg8) -0.22/14 9} 39. Be4 Re6 40. c3 {0.00/16 8} f5 {(Kg8) -0.21/14 7} 41. Bd5 {0.00/16 6} Ree8 {-0.25/15 0} 42. Rc2 {0.00/16 Qd7{-0.24/16 43. Bd2 {0.00/17 14} Rec8 {(Rbc8) -0.25/15 5} 44. Ba2 {0.00/16} Rd8 {-0.25/15} 45. Bd5 {0.00/17} Qe7 {(Bf8) -0.25/15} 46. Bc6 {0.00/16} Qe6 {(Qc7) -0.25/14} 47. Bd5 {0.00/17} Qd7 {-0.25/16} 48. Rh1 {0.00/17} Bf8 {(Ng3) -0.26/16} 49. Re1 {0.00/16} f4 {(Rdc8) -0.47/14} 50. Rh1 {-0.12/14} Ng3 {-0.39/14} 51. Rh2 {-0.23/16} Rdc8 {-0.41/16} 52. Be1 {-0.22/17} Rc5 {(Qf5) -0.39/15} 53. h5 {-0.23/15} Nxh5 {-0.41/15} 54.Rcg2 {-0.27/15} Be7 {(Rc7) -0.45/14} 55. Be4 {-0.19/15} Qe6 {(Kg7) -0.42/13} 56. Bf2 {-0.27/15} Rc7 {-0.42/16} 57. Bd5 {-0.29/16} Qc8 {(Qd7) -0.42/15} 58. Rg1 {-0.26/14} Qe8 {(Qd8) -0.42/13} 59. Rxh5 {0.00/13} gxh5 {-0.16/14} 60. g6 {0.00/15} hxg6 {-0.15/14} 61. Rxg6 {0.00/16} Qf8 {-0.20/15} 62. Bd4 {0.17/16} Bh4 {0.00/10} 63. Rg8+ {0.19/15} Qxg8 {0.00/9} 64. Bxg8 {0.19/16} Rxg8 {0.00/14} 65. Bb6 {0.19/18} Rd7 {(Rf7) 0.00/15} 66. Bxa5 {0.16/16} Rg1+ {0.00/15} 67. Ka2 {0.31/18} Kg7 {0.00/14} 68. Qf5 {0.31/17} Rf7 {0.06/15} 69. Qxh5 {0.31/18} Be7 {0.13/16} 70. Bb6 {0.31/18} Rg3 {0.20/15} 71. Bf2 {0.36/18} Rg6 {0.20/16} 72. Qh3 {0.36/18} d5 {0.21/15} 73. Qh1 {0.36/18} Bf6 {0.22/14} 74. Kb3 {0.35/17} Rg5 {0.26/14} 75. Qh2 {0.35/16} Rg6 {0.16/11} 76. Bc5 {0.35/16} Be7 {0.21/13} 77. Qf2 {0.09/16} e4 {0.05/13} 78. Qd4+ {0.09/16} Bf6 {0.00/13} 79. Qxd5 {0.09/17} e3 {0.00/15} 80. Qd1 {0.00/17} Rg2 {(Rg5) 0.00/14} 81. Qf1 {0.01/17} Rd2 {0.00/14} 82. Qh3 {0.08/17} Rd5 {(Bg5) 0.00/14} 83. Bb4 {0.00/15} Re5 {(Rg5) 0.00/13} 84. Qg4+ {0.00/12} Kh7 {(Bg5) 0.00/12} 85.Qxf4 {0.00/14} e2 {0.00/14} 86. c4 {0.00/14} bxc4+{0.00/14} 87. Ka2 {0.21/15} Kg6 {0.07/13} 88. Qg3+ {0.15/16} Rg5{0.00/13} 89. Qe1 {0.27/17}? Rg2 {(Re5) -0.80/11} 90. Qh1 {1.34/13}? e1=N {(Rg5) -9.47/11} 91. Bxe1 {-10.85/12} 0-1

As you can see 89. Qe1 and 90. Qh1 are fatal mistakes by white because after 89. Qe1 black can play Rg2 followed by Rd7 and ...Rd1. I'm a weak player and even I could see that mistake. 90. Qh1 is also a big error as black can promote to a knight which protects the rock and threatens mate after ...Rxb2.

Old programs like Shredder 10 and Toga II 1.4.2JD could see that these two moves are a mistake. I hope these bugs will be corrected in Rybka 4.

[M ANSARI]

Something seriously wrong in this game. The score simply does not seem correct. It is not 2 bad moves, it is a bunch of complete blunders on both sides which continue throughout the game. Nothing to do with Rybka or bugs in Rybka. On move 38 white plays Ba2? leaving a full rook hanging on g4 ... yet black does not take the rook and plays Qe7? Such strange moves continue several times with both sides not realizing a rook is hanging. Also both sides do not see a bishop hanging and tons of laughable one move blunders that even a 500 ELO player would not do. There is even a queen hanging at some point and queen sacs for basically no compensation.

It seems something is not right from both sides. I don't think it is a Rybka bug because my Rybka sees it. It could be a bug in the PC GUI or a wrong score.

[Dr.Wael Deeb]

Something seriously wrong in this game. The score simply does not seem correct. On move 38 white plays Ba2? leaving a full rook hanging on g4 ... yet black does not take the rook and plays Qe7? Such strange moves continue several times with both sides not realizing a rook is hanging.

It seems something is not right from both sides. I don't think it is a Rybka bug because my Rybka sees it. It could be a bug in the PC GUI.

Playchess engine room says it all Majd,no need to comment
Buggy GUI regards,
Dr.D

[Guenther]

In Playchess engine room, Phil Harris2 played someone who used Rybka 3 Dynamic but probably on weaker hardware.

Now Phil Harris's Rybka made two serious errors. It could probably draw the game hadn't it been for these two bad moves. Here's the game:

[Event "5m + 0s, rated"]
[Site "?"]
[Date "2009.04.29"]
[Round "?"]
[White "Phil Harris2, Rybka 3"]
[Black "Jefk, Rybka 3aqua"]
[Result "0-1"]
[ECO "B90"]
[WhiteElo "2708"]
[BlackElo "2656"]
[PlyCount "181"]
[EventDate "2009.04.29"]
[EventType "blitz"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. f3 e5 7. Nb3 Be6 8. Be3 Be7 9. Qd2 O-O 10. O-O-O Qc7 11. g4 Nbd7 12. Kb1 {0.00/0 0} b5 13. a3 Nb6 14. g5 Nh5 15. Na5 Rab8 16. Nd5 Bxd5 17. exd5 Nxd5 18. Qxd5 Qxa5 19. Bd3 g6 20. Be4 Qc7 21. Rhe1 a5 22. Rg1 Rfc8 23. Rd2 Bf8 24. h4 Qe7 25. Rdd1 Rc7 26. Qb3 Qd7 27. Rg4 Ng7 28. Bd5 Rcc8 29. Bd2 Qc7 30. Qd3 Nh5 31. Be3 Kh8 32. Ba2 Bg7 33. Qd2 Rd8 34. Bd5 Re8 {-0.20/15} 35. Ba2 Rec8 36. Qd3 Rd8 37. Rg2 Rd7 {(Qd7) -0.21/14 4} 38. Bd5 {0.00/15 5} Re7{(Kg8) -0.22/14 9} 39. Be4 Re6 40. c3 {0.00/16 8} f5 {(Kg8) -0.21/14 7} 41. Bd5 {0.00/16 6} Ree8 {-0.25/15 0} 42. Rc2 {0.00/16 Qd7{-0.24/16 43. Bd2 {0.00/17 14} Rec8 {(Rbc8) -0.25/15 5} 44. Ba2 {0.00/16} Rd8 {-0.25/15} 45. Bd5 {0.00/17} Qe7 {(Bf8) -0.25/15} 46. Bc6 {0.00/16} Qe6 {(Qc7) -0.25/14} 47. Bd5 {0.00/17} Qd7 {-0.25/16} 48. Rh1 {0.00/17} Bf8 {(Ng3) -0.26/16} 49. Re1 {0.00/16} f4 {(Rdc8) -0.47/14} 50. Rh1 {-0.12/14} Ng3 {-0.39/14} 51. Rh2 {-0.23/16} Rdc8 {-0.41/16} 52. Be1 {-0.22/17} Rc5 {(Qf5) -0.39/15} 53. h5 {-0.23/15} Nxh5 {-0.41/15} 54.Rcg2 {-0.27/15} Be7 {(Rc7) -0.45/14} 55. Be4 {-0.19/15} Qe6 {(Kg7) -0.42/13} 56. Bf2 {-0.27/15} Rc7 {-0.42/16} 57. Bd5 {-0.29/16} Qc8 {(Qd7) -0.42/15} 58. Rg1 {-0.26/14} Qe8 {(Qd8) -0.42/13} 59. Rxh5 {0.00/13} gxh5 {-0.16/14} 60. g6 {0.00/15} hxg6 {-0.15/14} 61. Rxg6 {0.00/16} Qf8 {-0.20/15} 62. Bd4 {0.17/16} Bh4 {0.00/10} 63. Rg8+ {0.19/15} Qxg8 {0.00/9} 64. Bxg8 {0.19/16} Rxg8 {0.00/14} 65. Bb6 {0.19/18} Rd7 {(Rf7) 0.00/15} 66. Bxa5 {0.16/16} Rg1+ {0.00/15} 67. Ka2 {0.31/18} Kg7 {0.00/14} 68. Qf5 {0.31/17} Rf7 {0.06/15} 69. Qxh5 {0.31/18} Be7 {0.13/16} 70. Bb6 {0.31/18} Rg3 {0.20/15} 71. Bf2 {0.36/18} Rg6 {0.20/16} 72. Qh3 {0.36/18} d5 {0.21/15} 73. Qh1 {0.36/18} Bf6 {0.22/14} 74. Kb3 {0.35/17} Rg5 {0.26/14} 75. Qh2 {0.35/16} Rg6 {0.16/11} 76. Bc5 {0.35/16} Be7 {0.21/13} 77. Qf2 {0.09/16} e4 {0.05/13} 78. Qd4+ {0.09/16} Bf6 {0.00/13} 79. Qxd5 {0.09/17} e3 {0.00/15} 80. Qd1 {0.00/17} Rg2 {(Rg5) 0.00/14} 81. Qf1 {0.01/17} Rd2 {0.00/14} 82. Qh3 {0.08/17} Rd5 {(Bg5) 0.00/14} 83. Bb4 {0.00/15} Re5 {(Rg5) 0.00/13} 84. Qg4+ {0.00/12} Kh7 {(Bg5) 0.00/12} 85.Qxf4 {0.00/14} e2 {0.00/14} 86. c4 {0.00/14} bxc4+{0.00/14} 87. Ka2 {0.21/15} Kg6 {0.07/13} 88. Qg3+ {0.15/16} Rg5{0.00/13} 89. Qe1 {0.27/17} Rg2 {(Re5) -0.80/11} 90. Qh1 {1.34/13} e1=N {(Rg5) -9.47/11} 91. Bxe1 {-10.85/12} 0-1

As you can see 89. Qe1 and 90. Qh1 are fatal mistakes by white because after 89. Qe1 black can play Rg2 followed by Rd7 and ...Rd1. I'm a weak player and even I could see that mistake. 90. Qh1 is also a big error as black can promote to a knight which protects the rock and threatens mate after ...Rxb2.

Old programs like Shredder 10 and Toga II 1.4.2JD could see that these two moves are a mistake. I hope these bugs will be corrected in Rybka 4.

The game is unusable and even Chessbase itself cannot read it,
because several closing brackets are missing.
Even without this it looks like a known problem with _not_ calculating
underpromotions _in the search_ at least for move 90.
(I cannot say anything for moves before, because the game is unreadable
as mentioned above)
Moreover generally saying you shouldn't trust any such 'server' games,
because no one knows how crappy something is set up by a hobbyist, or
if someone overclocked to death...

Guenther

[AdminX]

Something seriously wrong in this game. The score simply does not seem correct. It is not 2 bad moves, it is a bunch of complete blunders on both sides which continue throughout the game. Nothing to do with Rybka or bugs in Rybka. On move 38 white plays Ba2? leaving a full rook hanging on g4 ... yet black does not take the rook and plays Qe7? Such strange moves continue several times with both sides not realizing a rook is hanging. Also both sides do not see a bishop hanging and tons of laughable one move blunders that even a 500 ELO player would not do.

It seems something is not right from both sides. I don't think it is a Rybka bug because my Rybka sees it. It could be a bug in the PC GUI or a wrong score.

Looks like a bad PGN file paste here. If I paste this PGN in Chessbase GUI it only goes to move 54.

**Correction any GUI, it will not work.**

[Uri]

Here, I'll post the game again...

[Event "5m + 0s, rated"]
[Site "?"]
[Date "2009.04.29"]
[Round "?"]
[White "Phil Harris2, Rybka 3"]
[Black "Jefk, Rybka 3aqua"]
[Result "0-1"]
[ECO "B90"]
[WhiteElo "2708"]
[BlackElo "2656"]
[PlyCount "181"]
[EventDate "2009.04.29"]
[EventType "blitz"]

1. e4 {0.00/0 0} c5 {0.00/0 0} 2. Nf3 {0.00/0 0} d6 {0.00/0 0} 3. d4 {0.00/0 0}
cxd4 {0.00/0 0} 4. Nxd4 {0.00/0 0} Nf6 {0.00/0 0} 5. Nc3 {0.00/0 0} a6 {
0.00/0 0} 6. f3 {0.00/0 0} e5 {0.00/0 0} 7. Nb3 {0.00/0 0} Be6 {0.00/0 0} 8.
Be3 {0.00/0 0} Be7 {0.00/0 0} 9. Qd2 {0.00/0 0} O-O {0.00/0 0} 10. O-O-O {
0.00/0 0} Qc7 {0.00/0 0} 11. g4 {0.00/0 0} Nbd7 {0.00/0 0} 12. Kb1 {0.00/0 0}
b5 {0.00/0 0} 13. a3 {0.00/0 0} Nb6 {0.00/0 0} 14. g5 {0.00/0 0} Nh5 {0.00/0 0}
15. Na5 {0.00/0 0} Rab8 {0.00/0 0} 16. Nd5 {0.00/0 0} Bxd5 {0.00/0 0} 17. exd5
{0.00/0 0} Nxd5 {0.00/0 0} 18. Qxd5 {0.00/0 0} Qxa5 {0.00/0 0} 19. Bd3 {
0.00/0 0} g6 {0.00/0 0} 20. Be4 {0.00/0 0} Qc7 {0.00/0 0} 21. Rhe1 {0.00/0 0}
a5 {0.00/0 0} 22. Rg1 {0.00/0 0} Rfc8 {0.00/0 0} 23. Rd2 {0.00/0 0} Bf8 {
0.00/0 0} 24. h4 {0.00/0 0} Qe7 {0.00/0 0} 25. Rdd1 {0.00/0 0} Rc7 {0.00/0 0}
26. Qb3 {0.00/0 0} Qd7 {0.00/0 0} 27. Rg4 {0.00/0 0} Ng7 {0.00/0 0} 28. Bd5 {
0.00/0 0} Rcc8 {0.00/0 0} 29. Bd2 {0.00/0 0} Qc7 {0.00/0 0} 30. Qd3 {0.00/0 0}
Nh5 {0.00/0 0} 31. Be3 {0.00/0 0} Kh8 {-0.27/12 4} 32. Ba2 {0.00/0 0} Bg7 {
-0.21/15 33} 33. Qd2 {0.00/15 5} Rd8 {-0.22/15 5} 34. Bd5 {0.00/0 0} Re8 {
-0.20/15 8} 35. Ba2 {0.00/0 0} Rec8 {-0.24/13 4} 36. Qd3 {0.00/16 7} Rd8 {
-0.22/15 7} 37. Rg2 {0.00/16 2} Rd7 {(Qd7) -0.21/14 4} 38. Bd5 {0.00/15 5} Re7
{(Kg8) -0.22/14 9} 39. Be4 {0.00/15 6} Re6 {-0.22/15 0} 40. c3 {0.00/16 8} f5 {
(Kg8) -0.21/14 7} 41. Bd5 {0.00/16 6} Ree8 {-0.25/15 0} 42. Rc2 {0.00/16 9} Qd7
{-0.24/16 0} 43. Bd2 {0.00/17 14} Rec8 {(Rbc8) -0.25/15 5} 44. Ba2 {0.00/16 7}
Rd8 {-0.25/15 4} 45. Bd5 {0.00/17 3} Qe7 {(Bf8) -0.25/15 2} 46. Bc6 {0.00/16 3}
Qe6 {(Qc7) -0.25/14 2} 47. Bd5 {0.00/17 4} Qd7 {-0.25/16 2} 48. Rh1 {0.00/17 1}
Bf8 {(Ng3) -0.26/16 30} 49. Re1 {0.00/16 3} f4 {(Rdc8) -0.47/14 3} 50. Rh1 {
-0.12/14 6} Ng3 {-0.39/14 9} 51. Rh2 {-0.23/16 0} Rdc8 {-0.41/16 18} 52. Be1 {
-0.22/17 0} Rc5 {(Qf5) -0.39/15 5} 53. h5 {-0.23/15 10} Nxh5 {-0.41/15 0} 54.
Rcg2 {-0.27/15 9} Be7 {(Rc7) -0.45/14 0} 55. Be4 {-0.19/15 9} Qe6 {
(Kg7) -0.42/13 8} 56. Bf2 {-0.27/15 10} Rc7 {-0.42/16 0} 57. Bd5 {-0.29/16 21}
Qc8 {(Qd7) -0.42/15 0} 58. Rg1 {-0.26/14 10} Qe8 {(Qd8) -0.42/13 9} 59. Rxh5 {
0.00/13 2} gxh5 {-0.16/14 7} 60. g6 {0.00/15 0} hxg6 {-0.15/14 2} 61. Rxg6 {
0.00/16 0} Qf8 {-0.20/15 2} 62. Bd4 {0.17/16 13} Bh4 {0.00/10 4} 63. Rg8+ {
0.19/15 0} Qxg8 {0.00/9 0} 64. Bxg8 {0.19/16 3} Rxg8 {0.00/14 0} 65. Bb6 {
0.19/18 4} Rd7 {(Rf7) 0.00/15 5} 66. Bxa5 {0.16/16 2} Rg1+ {0.00/15 9} 67. Ka2
{0.31/18 0} Kg7 {0.00/14 1} 68. Qf5 {0.31/17 1} Rf7 {0.06/15 2} 69. Qxh5 {
0.31/18 1} Be7 {0.13/16 1} 70. Bb6 {0.31/18 1} Rg3 {0.20/15 2} 71. Bf2 {
0.36/18 2} Rg6 {0.20/16 3} 72. Qh3 {0.36/18 0} d5 {0.21/15 2} 73. Qh1 {
0.36/18 1} Bf6 {0.22/14 1} 74. Kb3 {0.35/17 4} Rg5 {0.26/14 0} 75. Qh2 {
0.35/16 13} Rg6 {0.16/11 1} 76. Bc5 {0.35/16 1} Be7 {0.21/13 1} 77. Qf2 {
0.09/16 9} e4 {0.05/13 2} 78. Qd4+ {0.09/16 1} Bf6 {0.00/13 1} 79. Qxd5 {
0.09/17 1} e3 {0.00/15 1} 80. Qd1 {0.00/17 4} Rg2 {(Rg5) 0.00/14 1} 81. Qf1 {
0.01/17 3} Rd2 {0.00/14 1} 82. Qh3 {0.08/17 2} Rd5 {(Bg5) 0.00/14 1} 83. Bb4 {
0.00/15 2} Re5 {(Rg5) 0.00/13 0} 84. Qg4+ {0.00/12 2} Kh7 {(Bg5) 0.00/12 2} 85.
Qxf4 {0.00/14 2} e2 {0.00/14 0} 86. c4 {0.00/14 2} bxc4+ {0.00/14 0} 87. Ka2 {
0.21/15 5} Kg6 {0.07/13 3} 88. Qg3+ {0.15/16 0} Rg5 {0.00/13 1} 89. Qe1 {
0.27/17 1} Rg2 {(Re5) -0.80/11 9} 90. Qh1 {1.34/13 1} e1=N {(Rg5) -9.47/11 1}
91. Bxe1 {-10.85/12 8} 0-1

[parrish]

...
Even without this it looks like a known problem with _not_ calculating
underpromotions _in the search_ at least for move 90.
...

Seems strange that Rybka 3 would have this underpromotion to a knight bug when the free Rybka 2.2n2 does not have that problem.

[M ANSARI]

This looks like a hash table bug which seems to also happen with wrong bishop. Some hash information simply is not seen until after the fact. I think this is already fixed in the new Rybka. Anyway this game was lost and no way to save it, so why not go out with a bang

[parrish]

This looks like a hash table bug which seems to also happen with wrong bishop. Some hash information simply is not seen until after the fact. I think this is already fixed in the new Rybka. Anyway this game was lost and no way to save it, so why not go out with a bang

Au contraire, without the bug, Rybka would've never moved 89. Qe1?, but would've drawn instead with 89. Qf2.

[Uri]

This looks like a hash table bug which seems to also happen with wrong bishop.

Ansari, what do you mean by "wrong bishop"?