Why it is stupid to resign against an engine

[hgm]

For the Dutch open computer-chess championship, Joker had to play Ktulu, which about 500 Elo stronger than Joker. But in computer Ches one should never despair, as there is no limit to the stupidityof the bugs an engine can have:

Code: Select all

[Event "Computer Chess Game"]
[Site "SCHAAK_PC"]
[Date "2009.10.17"]
[Round "-"]
[White "joker_P2.exe"]
[Black "Ktulu"]
[Result "1/2-1/2"]
[TimeControl "4800"]
[Annotator "1. +0.12"]

1. e4 {+0.12/14} c6 2. d4 {+0.33/14} d5 3. e5 {+0.29/14} Bf5 4. Be2
{+0.45/14} e6 5. h4 {+0.22/14} Be7 6. Nf3 {+0.17/14} Nh6 7. Bxh6 {+0.45/13}
gxh6 8. Nbd2 {+0.32/14} Nd7 9. h5 {+0.31/15} Rg8 10. O-O {+0.19/15} Qb6 11.
Nb3 {+0.36/15} O-O-O 12. a4 {+0.25/14} a5 13. Qd2 {+0.23/15} Qb4 14. Qxb4
{+0.44/15} axb4 15. Ne1 {+0.36/16} c5 16. dxc5 {+0.67/15} Nxe5 17. a5
{+0.47/15} Kc7 18. Nd3 {+0.32/13} Nc6 19. Ra4 {+0.34/14} Bf6 20. Rb1
{+0.23/14} Be4 21. Bf1 {+0.27/15} Rg5 22. a6 {+0.07/14} bxa6 23. Rxa6
{+0.41/14} Rb8 24. Ne1 {+0.38/14} Rxh5 25. Be2 {+0.27/14} Rg5 26. Bd3
{+0.19/14} h5 27. Na5 {+0.20/14} Nxa5 28. Rxa5 {+0.73/15} Bd4 29. Kf1
{+0.43/15} h4 30. Ra7+ {-1.27/15} Rb7 31. Rxb7+ {-1.58/17} Kxb7 32. Bxe4
{-1.87/18} dxe4 33. c3 {-2.04/18} Bxc5 34. cxb4 {-1.98/19} Bxb4 35. Nc2
{-1.74/17} Bd2 36. Rd1 {-1.75/17} Rd5 37. Ke2 {-1.80/17} Bf4 38. b3
{-1.90/18} Rxd1 39. Kxd1 {-2.19/19} Kc6 40. Ke2 {-2.77/19 0.1} f5 41. Kf1
{-2.91/18} Kc5 42. Ke2 {-3.15/18} h5 43. Ke1 {-3.13/18} Bd6 44. Ke2
{-3.29/20} f4 45. Kf1 {-3.36/20} Kd5 46. Kg1 {-4.16/21} Bc5 47. Kf1
{-4.04/21} e5 48. Ke2 {-4.23/22} f3+ 49. gxf3 {-3.15/22} h3 50. fxe4+
{-5.37/23} Kxe4 51. Kf1 {-7.41/21} h2 52. Kg2 {-9.16/24} Kd3 53. Ne1+
{-10.89/24} Ke2 54. Nc2 {-10.94/25} e4 55. Kxh2 {-10.93/1} Kd2 56. b4
{-13.42/20} Bd6+ 57. Kg1 {-14.19/1} Kxc2 58. b5 {-14.19/18} Bc5 59. Kf1
{-14.65/18} h4 60. Ke2 {-14.37/17} Kc3 61. Kf1 {-14.37/16} Kd2 62. Kg2
{-23.05/17} h3+ 63. Kxh3 {-14.46/17} Bxf2 64. Kg2 {-100.11/17} Bb6 65. Kh3
{-100.09/17} e3 66. Kh4 {-100.08/16} e2 67. Kg5 {-100.07/13} e1=Q 68. Kf6
{-100.06/12} Bd8+ 69. Kf5 {-100.06/11} Qe8 70. Kf4 {-100.05/10} Qxb5 71.
Kf3 {-100.04/8} Qb4 72. Kg3 {-100.04/8} Qe4 73. Kh3 {-100.03/6} Qf3+ 74.
Kh2 {-100.03/6} Kc3 75. Kg1 {-100.03/6} Kb2 76. Kh2 {-100.03/6} Kc2 77. Kg1
{-100.03/6} Kb2 78. Kh2 {-100.03/6} Ba5 79. Kg1 {-100.03/6} Bb6+ 80. Kh2
{-100.02/4 0.1} Qf1 81. Kg3 {-100.04/8} Qf5 82. Kg2 {-100.03/6} Qg4+ 83.
Kh2 {-100.02/4} Kc2 84. Kh1 {-100.01/2} Kb2 85. Kh2 {-100.02/4} Kc1 86. Kh1
{-100.01/2} Kc2 87. Kh2 {-100.02/4} Bc7+ 88. Kh1 {-100.02/4} Qe2 89. Kg1
{-100.02/4} Kb2 90. Kh1 {-100.01/2} Kc1 91. Kg1 {-100.02/4} Kd2 92. Kh1
{-100.01/2} Kc2 93. Kg1 {-100.02/4} Bd8 94. Kh1 {-100.02/4} Bf6 95. Kg1
{-100.02/4} Bd4+ 96. Kh1 {-100.02/4} Qf3+ 97. Kh2 {-100.02/4} Bf6 98. Kg1
{-100.03/6} Qe2 99. Kh1 {-100.02/4} Bc3 100. Kg1 {-100.02/4} Kd1 101. Kh1
{-100.02/4} Be5 102. Kg1 {-100.02/4} Bh2+ 103. Kh1 {-100.03/6} Bc7 104. Kg1
{-100.02/4} Kc1 105. Kh1 {-100.01/2} Kb2 106. Kg1 {-100.02/4} Be5 107. Kh1
{-100.01/2} Kc2 108. Kg1 {-100.02/4} Kd2 109. Kh1 {-100.01/2} Qf3+ 110. Kg1
{-100.02/4} Bd4+ 111. Kh2 {-100.02/4} Bf6 112. Kg1 {-100.02/4} Qh3 113. Kf2
{-100.01/2} Kc2 114. Ke2 {-100.03/6} Bd4 115. Ke1 {-100.02/4} Qh2 116. Kf1
{-100.01/2} Qg3 117. Ke2 {-100.01/2} Qg2+ 118. Ke1 {-100.01/2} Bf2+ 119.
Ke2 {-100.02/4} Bc5+ 120. Ke1 {-100.01/2 0.1} Kb2
{Draw by fifty move rule} 1/2-1/2

In last year's championship, it already turned out that Ktulu's ability to checkmate an opponent was not very robust, when it took 99 moves (i.e. 199 ply) to win KQQK againt Joker. It could finally only do it by sacrificing one of its Queens on move 49, forcing Joker to take it, and then took another 50 moves to perform the mate in KQK.

But of course such a trick does not work in KQBK, so I had actually good hopes to draw, this time!

[slobo]

For the Dutch open computer-chess championship, Joker had to play Ktulu, which about 500 Elo stronger than Joker. But in computer Ches one should never despair, as there is no limit to the stupidityof the bugs an engine can have:


In last year's championship, it already turned out that Ktulu's ability to checkmate an opponent was not very robust, when it took 99 moves (i.e. 199 ply) to win KQQK againt Joker. It could finally only do it by sacrificing one of its Queens on move 49, forcing Joker to take it, and then took another 50 moves to perform the mate in KQK.

But of course such a trick does not work in KQBK, so I had actually good hopes to draw, this time!

By simple not allowing Joker to resign to improved your engine.
Very economic solution! I liked it.

[George Tsavdaris]

[quote="hgm"]For the Dutch open computer-chess championship, Joker had to play Ktulu, which about 500 Elo stronger than Joker. But in computer Ches one should never despair, as there is no limit to the stupidityof the bugs an engine can have:

Code: Select all

[Event "Computer Chess Game"]
[Site "SCHAAK_PC"]
[Date "2009.10.17"]
[Round "-"]
[White "joker_P2.exe"]
[Black "Ktulu"]
[Result "1/2-1/2"]
[TimeControl "4800"]
[Annotator "1. +0.12"]


But of course such a trick does not work in KQBK, so I had actually good hopes to draw, this time! :lol:  :lol:  :lol:[/quote]

 :lol:  :lol: 
So Ktulu couldn't checkmate Joker from e.g:
[d]8/8/5b2/8/8/5q2/3k4/6K1 b - - 0 112

 That is a clearly stupid bug. I wonder if there are other types of bugs that they would have been discovered in CEGT or CCRL if no resign option was enabled, and what would be the ratings then....

[Michael Diosi]

Hi,

Just take care people here don't like if you report bugs. Harvey will ask you whether you have this engine or not and you will be banned from the ChessBase server for saying anything. Ray will tell you this is one in a million poitions and he can't care less. Ruxy will ignore once omre all the bug reports and get personal. They will delete Arena instead


Now imagine all those Iranian nuclear powerplants where the guy is working now...

Michael
http://www.playwitharena.com