Author: Dave Gomboc
Date: 02:38:10 09/13/00
Go up one level in this thread
On September 13, 2000 at 00:24:18, Dave Gomboc wrote: A grandmaster who is writing a book brought the position after 19...Nc6 in Shirov-Kasparov 1997 to my attention: [D]r2qk2r/1p1b1pb1/p1npp3/8/3NPPp1/1P2B1Pp/P1P1N2P/R2Q1RK1 w kq - 0 1 (I've also placed the complete PGN score at the bottom of the message.) Kasparov, when annotating this position, had the following to say: "The future importance of the long diagonal(a8-h1) is not yet clear, but from my experience I can guarantee that the white king is potentially in a much worse position than his black colleague. Any further opening of the long diagonal or the appearance of the queen on the second rank will create a deadly threat on g2. In fact, the pawn on h3 can be seen as a material advantage for Black, because it is so important that you could value it has a whole piece. It not only helps the queen to create mating threats, but in most endgames, this pawn will also guarantee Black a winning edge because of the threats that Black can create against the h2-pawn, when the black h-pawn is very close to the promotion square."-GK I was asked to comment on how a computer program would assess this position. My reply was that strong programs would indeed assess this kingside structure as favourable for Black, and specifically that "The computer would deduct points for White's poor king safety and the colour complex weakness: White's pawns are far from the ideal "f2-g2-h2 in front of a king on g1", worse--Black has a solid grip on g2 and f3, and g2 is adjacent to White's king. To compound the problem further, White doesn't even have a light-squared bishop to try to limit the damage with. Still, depending on the program, the penalty terms may only kick into high gear when the program can see Black beginning to exploit these factors within its search horizon, by which time it may well be too late for the program to do anything about it." My goal in bringing up this position and commentary is two-fold: 1. The information is of high quality due to the top calibre of the players and annotator of the game. I believe it is valuable for developers to ask how their program assesses this position (specifically the kingside situation), reflect on whether that assessment is correct or not, and if not, what features might be introduced or modified in their program to cause it to give an estimate more in line with Kasparov's assessment. 2. I would like to verify that the comments I provided are well-founded (the book will soon go to print, but hasn't yet, so I can still modify them if they are not! ;) With regards to reason 2., while I think an interesting conversation here could arise, I would be happy to receive email (dave@cs.ualberta.ca) if you have some comments but would prefer not to post them on CCC for whatever reason. Thanks, Dave [Event "It (cat.19)"] [Site "Linares (Spain)"] [Date "1997.??.??"] [White "Shirov, A "] [Black "Kasparov, G "] [Result "0-1"] [ECO "B90"] [Round "10"] 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be3 Ng4 7. Bg5 h6 8. Bh4 g5 9. Bg3 Bg7 10. Be2 h5 11. Bxg4 Bxg4 12. f3 Bd7 13. O-O Nc6 14. Bf2 e6 15. Nce2 Ne5 16. b3 g4 17. f4 h4 18. Be3 h3 19. g3 Nc6 20. Qd3 O-O 21. Rad1 f5 22. c4 Qa5 23. Nc3 Rae8 24. Rfe1 e5 25. Nxc6 Bxc6 26. b4 Qa3 27. b5 exf4 28. Bxf4 axb5 29. cxb5 Qc5+ 30. Be3 Qxc3 31. bxc6 Qxc6 32. Qxd6 Qxe4 33. Qd5+ Qxd5 34. Rxd5 Bc3 35. Re2 Re4 36. Kf2 Rfe8 37. Rd3 Bf6 38. Red2 Rxe3 0-1
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.