Author: Albert Silver
Date: 06:11:30 07/15/04
Go up one level in this thread
On July 15, 2004 at 02:18:12, Jouni Uski wrote:
>[Event "ITT Copa CMSalud"]
>[Site "Vicente Lopez"]
>[Date "2004.07.09"]
>[Round "1"]
>[White "Shredder 8, COMP."]
>[Black "Valiente, Cristobal"]
>[Result "1/2-1/2"]
>[BlackElo "2323"]
>[PlyCount "118"]
>[EventDate "2004.07.09"]
>[SourceDate "2004.07.08"]
>
>1. e4 d6 2. d4 g6 3. Nf3 Bg7 4. Nc3 Nf6 5. Bf4 Nc6 6. Qd2 Bg4 7. O-O-O O-O 8.
>d5 Ne5 9. Be2 Bxf3 10. gxf3 c6 11. Kb1 cxd5 12. Nxd5 Nxd5 13. Qxd5 Qb6 14. Be3
>Qc7 15. Bd4 Rac8 16. c3 a6 17. f4 e6 18. Qb3 Nd7 19. Bxg7 Kxg7 20. f5 Nc5 21.
>Qc2 Rcd8 22. h4 Qc6 23. f3 d5 24. fxg6 fxg6 25. h5 dxe4 26. hxg6 hxg6 27. Rxd8
>Rxd8 28. Qc1 Rh8 29. Rxh8 Kxh8 30. Qh6+ Kg8 31. Qxg6+ Kf8 32. b4 Na4 33. Qf6+
>Ke8 34. Kc2 b5 35. Kb3 exf3 36. Qxf3 Qd5+ 37. Qxd5 exd5 38. Kc2 Ke7 39. Kd3 Kd6
>40. Bg4 Nb6 41. Kd4 Kc6 42. Bf3 Kd6 43. Bh1 Na4 44. Bg2 Nb6 45. Bf3 Na4 46. Bh5
>Nb6 47. Bf3 Na4 48. Bg2 Nb6 49. Bh3 Kc6 50. Bf1 Kd6 51. Bh3 Kc6 52. Be6 Kd6 53.
>Bf7 Na4 54. Bg6 Nb6 55. Bf5 Kc6 56. Be6 Kd6 57. Bf7 Na4 58. Bh5 Nb6 59. Be8 Na4
>1/2-1/2
>
>Is white really 2800 player?
>
>Jouni
You're not new here Jouni, so I know the following arguments are not new to your
ears either:
- the SSDF rating list is not a human rating list and has no human participation
so claiming the results of Shredder vs. various versions of Fritz, Junior,
Hiarcs, etc. In no way can be given as evidence of its results vs humans.
Remember that ratings are not a pure chess value, but just a comparative system
showing how well one has scored against others. The Elo system is used in plenty
of other games than chess. The 2800 SSDF rating reflects its dominance against
other programs on similar or worse hardware.
- Even if one is to accept that Shredder on a 3 GHz system plays at 2800 FIDE,
it is very *old* news that it's just not possible to say it will play like a
human 2800 player. The 2800 or 2700 rating (it's not terribly important since
we're still talking the very elite in the world) is its average performance when
playing. There's a very good reason stratagems such as anti-computer chess
exist: avoid playing on computer strengths, tactics, and steer towards closed
strategic positions. Give it an open tactical position. Is it 2800? No, it's
probably more like 3000+ *compared* to human FIDE rated players. Give it a
closed slow-maneuvering game with few tactical chances. Is it 2800? No, it's
anywhere between 2400 (sometimes worse) to 2500. Since variety is the spice of
life, and chess is never only one or the other, it most likely will average out
to 2700+. But it will never mean it plays like another human 2700 player.
None of this is new, and has been argued since the "are comps GMs yet?" to the
near 2700 rating of Chess Tiger on SSDF some years back. (you can run a CCC
search on the rating 2696 if memory serves).
Anyhow, it is doing quite well in both tournaments in Argentina, with 5 in 6 in
the GM event and 4 in 5 in the IM tournament.
Albert
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.