Author: Robert Hyatt
Date: 20:30:00 10/21/05
Go up one level in this thread
On October 20, 2005 at 12:44:55, John Merlino wrote: >On October 20, 2005 at 11:46:34, Robert Hyatt wrote: > >>Here is the PGN for a blitz game Crafty played on ICC (ignore the pgn headers, I >>fabricated them when annotating the game, I believe the game was a 5 0 blitz). >> >>[Event "?"] >>[Site "?"] >>[Date "2005.10.18"] >>[Round "?"] >>[White "Crafty 20.1"] >>[WhiteElo ""] >>[Black "unknown"] >>[BlackElo ""] >>[Result "*"] >> >>1. d4 Nf6 2. c4 e6 3. Nc3 Bb4 4. Bg5 c5 5. d5 h6 6. Bh4 Bxc3+ >>7. bxc3 d6 8. e3 e5 9. Bd3 e4 10. Bc2 Qe7 11. Bxf6 Qxf6 12. Ne2 >>Qe5 13. Qb1 Nd7 14. Bxe4 O-O 15. Bd3 Nf6 16. Qc2 a6 17. a4 Bg4 >>18. Rb1 b5 19. axb5 axb5 20. Rxb5 Bxe2 21. Kxe2 Rfe8 22. h3 >>Nh5 23. Kd2 Qg5 24. Rg1 Qf6 25. Rf1 Qe5 26. Rb7 Nf6 27. Rbb1 >>Qh2 28. Rg1 Nd7 29. Rbf1 Ne5 30. Ke2 Nxd3 31. Kxd3 Qe5 32. Kd2 >>Qf6 33. f4 Ra3 34. Re1 Qd8 35. Re2 Qa5 36. Rb1 Ra4 37. Kd3 f6 >>38. Rb5 Qa6 39. Qb3 Ra3 40. Qb2 Ra4 41. e4 Kh7 42. Qc2 Kg8 43. Qb3 >>Ra3 44. Qb2 Ra4 45. Qc2 Kh8 46. Qb3 Ra3 47. Qb2 Ra4 48. Qb3 >>Ra3 49. Qb2 Ra4 50. h4 h5 51. f5 Kh7 52. Qb1 Ra1 53. Rb6 Qa5 >>54. Qb3 Ra3 55. Rb5 Qa6 56. Qc2 Ra1 57. Rb1 Kh8 58. Ree1 Ra2 >>59. Qd1 Kg8 60. Qxh5 Ra8 61. Qg4 Ra4 62. e5 dxe5 63. Qe4 Ra3 >>64. Rb5 Qa4 65. d6 Ra2 66. Rc1 Rf2 67. Rxc5 Qa2 68. Qe1 Rd8 >>69. Ke3 Rxg2 70. Rd1 Rh2 71. Rd2 Rh3+ 72. Ke4 Qa8+ 73. Rdd5 >>Rxd6 74. Qb1 Rxd5 75. Rxd5 Qc8 76. Qb5 Rxh4+ 77. Ke3 Qxf5 78. Rd8+ >>Kh7 79. Qb3 Qf4+ 80. Kd3 Rh2 81. Rh8+ Kxh8 82. Qd1 e4+ 83. Kd4 >>Qd6+ 84. Ke3 Qxd1 85. Kxe4 Rf2 86. c5 * >> >> >>The interesting stuff starts around move 65. White played d6 with an eval of >>+3.20. Let your program search there to see what it likes. The longer you >>search, the more the score should drop. >> >>If it doesn't follow the game another move and try again. Crafty's scores >>looked like this: >> >>move 66: +3.15, 67: +2.41, 68: +1.59, 69: +.62, 70: 0.00, 71: -.39, 72: 0.00, >>73: 0.00, 74: -1.33, 75: -2.76, 76: -2.75, 77: -12.0, 78: -13.66, etc. >> >>I first thought I had some sort of bug, but after playing back through the >>position, this is less a bug and more just missing deep tactics. Would be >>interesting to see what kinds of scores at different depths other programs see >>starting at move 66... > >At move 65, CM9_R1 on a P4-2.4 likes d6 only between 1 and 11 seconds -- after >that it switches to Rc1, and eventually Reb1 with a good score: > That was actually the real point here. This was a very fast game, and by the time Crafty reached this point, it was using a couple of seconds per move, which is what made this interesting. Once the search time stretches out, the position is less interesting, but here, the score just tanks, even though playing what appears to be reasonable moves at this time limit... >Time Depth Score Positions Moves >0:00 1/4 2.10 12311 65.Rc1 Ra1 66.Rxa1 Qxa1 67.Qe2 Ra2 >0:00 1/5 2.00 19324 65.Rc1 Ra2 66.d6 Qa3 67.Qd5+ Kh7 > 68.Rf1 >0:00 1/6 1.98 46313 65.Rc1 Ra2 66.d6 Qa3 67.Qd5+ Kh7 > 68.Rc2 >0:00 1/7 1.86 111561 65.Rc1 Ra2 66.d6 Kh7 67.Rxc5 Qa3 > 68.Qe3 Rxg2 >0:01 1/7 1.97 230224 65.d6 Ra2 66.Rc1 Rd8 67.Qd5+ Kh7 > 68.g4 Qa3 69.Rc2 e4+ 70.Kd2 >0:02 1/8 1.90 511510 65.d6 Ra2 66.Rc1 Rd8 67.Qd5+ Kh7 > 68.Rxc5 Qd7 69.g4 Qxd6 70.Qxd6 > Rxd6+ 71.Ke3 >0:07 1/9 1.76 1510755 65.d6 Ra2 66.Rc1 Rf2 67.Ke3 Rf4 > 68.Qd5+ Kh7 69.Rxc5 Qa3 70.Rc2 Rxh4 >0:11 1/9 1.93 2350849 65.Rc1 Ra1 66.Rxa1 Qxa1 67.Qe2 > Re8 68.h5 e4+ 69.Kd2 e3+ 70.Kd3 Qa7 >0:17 1/10 1.83 3830527 65.Rc1 Ra1 66.Rxa1 Qxa1 67.Qe2 > Re8 68.Kc2 Ra8 69.Rb2 Re8 70.Rb7 e4 >0:36 1/10 2.18 7797126 65.Reb1 Ra2 66.R1b3 Re8 67.Rb7 > Re7 68.Rb8+ Re8 69.Rxe8+ Qxe8 70.d6 > Kf8 71.Rb7 Qa4 >1:06 1/11 2.36 14374356 65.Reb1 Ra1 66.Qg4 Rxb1 67.Rxb1 > Qa7 68.Qe2 Re8 69.Ke4 Ra8 70.g4 > Qc7 71.Kd3 >1:50 1/12 2.26 24381826 65.Reb1 Ra1 66.Qg4 Rxb1 67.Rxb1 > Qa7 68.Qe2 Re8 69.Qh5 Rd8 70.Qf3 > Rd7 71.Rb5 > >Here's what I get after forcing each pair of moves (analysis provided for moves >on which The King disagrees with the move played): > >65.d6 Ra2 (+1.48 at depth 11, +0.88 at 12 -- agrees with Rc1) > >66.Rc1 Rf2 (+1.5 at depth 11, +1.48 at 12 -- wants Ke3) >Time Depth Score Positions Moves >0:00 1/3 2.26 2703 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 Rxh4 >0:00 1/4 2.26 6605 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 Rxh4 >0:00 1/5 2.61 17824 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 > Rxh4 70.Rc7 >0:00 1/6 1.76 79263 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 > Qa3 70.Rc2 Rxh4 >0:00 1/7 1.93 155162 67.Ke3 Qa2 68.Qxa8+ Qxa8 69.Kxf2 > Qe4 70.Ra1 Qxf5+ 71.Kg1 Qd3 >0:01 1/8 1.81 346542 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 > Qa3 70.Rd1 Rxf5 71.Ke2 Rb8 >0:04 1/9 1.74 1017780 67.Ke3 Rf4 68.Qd5+ Kh7 69.Rxc5 > Qa3 70.Rd1 Qxc3+ 71.Qd3 Qb2 72.Rd2 > Qa1 >0:15 1/10 1.58 3036554 67.Ke3 Rf4 68.Qd5+ Kh7 69.g3 Rxc4 > 70.Rxc5 Rxc5 71.Qxc5 Qg4 72.Qd5 > Rd8 73.Kf2 Qxf5+ 74.Kg1 >0:30 1/11 1.50 6378080 67.Ke3 Rf4 68.Qd5+ Kh7 69.g3 Rxc4 > 70.Rxc5 Rxc5 71.Qxc5 Qg4 72.Qc6 > Rd8 73.Rg1 Qxf5 74.Rd1 >1:14 1/12 1.48 15876152 67.Ke3 Rf4 68.Qd5+ Kh7 69.g3 Rxc4 > 70.Rxc5 Rxc5 71.Qxc5 Rd8 72.Qb6 > Qd7 73.c4 Qxd6 74.Qxd6 Rxd6 75.c5 > Rc6 > >67.Rxc5 Qa2 (-0.3 at depth 11, agrees with Qe1) > >68.Qe1 Rd8 (-0.81 at depth 11, agrees with Ke3) > >...and it's pretty much all downhill from there. Let me know if you need more. > >jm
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.