Author: Robert Hyatt
Date: 13:57:49 01/22/99
Go up one level in this thread
On January 22, 1999 at 10:44:45, Christopher R. Dorr wrote:
>Hello!
>
>I was watching a veryy interesting game between Ban and Crafty on ICC the other
>day. The moves are:
>
>[Event "ICC 15 3"]
>[Site "Internet Chess Club"]
>[Date "1999.01.20"]
>[Round "-"]
>[White "*ban"]
>[Black "*crafty"]
>[Result "1-0"]
>[WhiteElo "2624"]
>[BlackElo "2594"]
>[Opening "QGD"]
>[ECO "D52"]
>[NIC "QP.08"]
>[Time "18:16:21"]
>[TimeControl "900+3"]
>
>1. d4 d5 2. c4 e6 3. Nf3 Nf6 4. Nc3 Nbd7 5. Bg5 c6 6. e3 Bb4 7. cxd5 cxd5 8. Qc2
>O-O 9. Bd3 h6 10. Bh4 Qc7 11. Rc1 Bd6 12. Bg3 e5 13. dxe5 Nxe5 14. Nxe5 Bxe5 15.
>Nxd5 Qa5+ 16. Nc3 Bxc3+ 17. Qxc3 Qxa2 18. Bc4 Qa4 19. Be5 Nh5 20. Be2 Nf6 21.
>Bxf6 gxf6 22. Qxf6 Qb4+ 23. Kf1 Be6 24. Rd1 Rfc8 25. Qxh6 Qxb2 26. Qg5+ Kf8 27.
>Qh4 Rc1 28. g3 Rc2 29. Qh5 Rac8 30. Qh6+ Ke8 31. Rd4 Rxe2 32. Qh8+ Ke7 33. Qh4+
>f6 34. Qh7+ Bf7 35. Re4+ Kd6 36. Rd4+ Qxd4 37. exd4 Bc4 38. Qh6 Kd7 39. h4 Rce8
>40. Rh2 Bd5 41. Qg7+ R2e7 42. Qg4+ Kc6 43. Qd1 Re1+ 44. Qxe1 Bc4+ 45. Kg2 Rxe1
>46. h5 Bd5+ 47. Kh3 Rh1 48. Rxh1 Bxh1 49. g4 a5 50. f4 Be4 51. f5 a4 52. h6 Bxf5
>53. gxf5 a3 54. h7 a2 55. h8=Q a1=Q 56. Qxf6+ Kb5 57. Qd6 Qf1+ 58. Kg4 Qg2+ 59.
>Kh5 Qf3+ 60. Kg6 Qg4+ 61. Kf6 Kc4 62. Qc5+ Kd3 63. Qb5+ Kc3 64. Qxb7 Qh4+ 65.
>Ke6 Qg4 66. Qc6+ Kd2 67. d5 Qg8+ 68. Kf6 Qh8+ 69. Kg5 Qg7+ 70. Qg6 Qe7+ 71. Qf6
>Qe3+ 72. Kg6 Qe8+ 73. Kg7 Qd7+ 74. Qf7 Qd6 75. f6 Kc3 76. Qe6 Qg3+ 77. Kf7 Qc7+
>78. Qe7 Qc4 79. Qd6 Kb3 80. Qb6+ Ka2 81. Qc6 Qb3 82. Qa6+ Kb1 83. Qf1+ Kb2 84.
>Qe2+ Kb1 85. Ke6 Qb8 86. Qe4+ Kc1 87. Qc4+ Kb2 88. f7 Qf8 89. Qd4+ Kc1 90. d6
>Qh6+ {Black resigns} 1-0
>
>This game left me somewhat puzzled. I'm going to analyze it on my MII/266 at
>home, but here at work I only have a P75 to look at it, and that doesn't help
>much.
>
>I had three primary questions about this:
>
>1. Crafty's play between moves 12-24 seems distinctly 'UnCrafty-like'. I've
>played/watch probably 1000 Crafty games, and this series of moves seems very
>peculiar. Crafty seemed to allow a total destruction of it's kingside, by moves
>only a few ply deep. What was going on? Bob - do you keep the log files on ICC
>games, or do they disappear?
I keep them to look at. In this (and several games around this time) my king
safety was 'broken'. It has always been 'too touchy'... IE any minor change
made huge changes. I have been redoing this completely to get rid of this,
to try to develop the idea that the first few king safety penalties are much
more important that the last few. But in doing this, for a while, i actually
had it disabled without knowing. I wouldn't say it is back to 'perfect' yet
but it isn't doing that nearly so much. I watched it lose several games where
it _let_ the opponent play something like Bxf6 where it had to respond gxf6,
when it could have defended the knight and avoided that.
>
>2. I'm not 100% sure, but the Crafty's queen sac with 36...Qxd4 seems remarkable
>and sound. Any idea how deep it was searching, and what it's evaluation was?
no idea now as once I found king safety was compromised I fixed the problem and
the deleted the games. But in such a game I would suspect 10-12 plies as a
normal search in that 15 3 time control.
>
>3. How could a computer that could come up with the above-mentioned profound
>sacrifice play 49...a5? This seems to lose almost by force. And at fairly
>shallow depths, especially for an ending. Kd6 appears to win (to me, anyway). It
>seems amazing to me that it could lose such a simple-appearing ending. There may
>well have been some other stuff soing on...maybe Kd6 lost too. Is there any way
>to get hold of the logs for this game?
>
>Or alternatively, does anyone have a very fast comp that might be able to
>explain this?
>
there are lots of possible reasons... the most common is that during the day,
_anything_ can happen. IE I do all sorts of _other_ things on this machien,
and can occasionally "smash" crafty's performance badly. At night this is not
common, but during the day anything can happen...
Whether this happened or not, or whether it was just a mis-evaluation or not,
would be guessing, but I will look at the position again to see...
>Thanks,
>
>Chris Dorr
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.