Author: Uri Blass
Date: 14:05:57 05/29/03
Go up one level in this thread
On May 29, 2003 at 13:49:42, stuart taylor wrote:
>On May 29, 2003 at 10:36:33, Stephen Ham wrote:
>
>>Dear Gentlemen,
>>
>>I too noted the erratic evaluation behavior in Shredder 7 when I reviewed it for
>>ChessCafe.com. That review goes into more detail and can be found here:
>>
>>http://www.chesscafe.com/text/review365.pdf
>>
>>I see the same behavior in Shredder 7.04 too. Nonetheless, I'm truly impressed
>>with Shredder's ability to play chess at a very high level...in spite of its
>>evaluation function, which I find unreliable. In short, Shredder somehow
>>generally finds better moves than any other engine I've seen, but it's
>>assessment of the position is not to be trusted.
>
>Stephen, How can it be that the evaluation is less good than its actual playing?
>Perhaps because its evaluation is only done much slower?
>S.Taylor
static evaluation is not everything.
The problem in chess is to choose better moves and
sometimes you need to choose correctly between
2 option that seem equal based on static evaluation.
Here is an example(the game that I posted is copied from another post but I know
that after 2000 posts things disappear so I prefered to give it again and not to
give a link for an old post).
give program the following game without the moves
41...Rh2 and after it.
Ask them to analyze the position at move 41 for black.
Most programs will show positive score for Rh2 and the score will go down to a
draw score very quickly but they will not change their mind very fast because
there is no alternative that is better than a draw.
Shredder is different.
For shredder not all the draws are the same.
Rh2 is negative draw because white force repetition
and maybe can do more than it when the alternative is a positive draw
because black force repetition.
Shredder does not need to see that 41...Rh2 is losing to change it's mind to
the positive draw.
Other programs need to do it.
[Event "ICS rated standard match"]
[Site "204.178.125.65"]
[Date "2003.05.23"]
[Round "-"]
[White "TombRaider"]
[Black "PostModernist"]
[Result "1-0"]
[WhiteElo "2433"]
[BlackElo "2456"]
[TimeControl "1800+10"]
1. Nf3 Nf6 2. d4 d5 3. c4 e6 4. Nc3 dxc4 5. e4 Bb4 6. Bg5 c5 7. e5 h6 8.
exf6 hxg5 9. fxg7 Rg8 10. dxc5 Qf6 11. Rc1 Bxc5 12. Bxc4 g4 13. Bb5+ Bd7
14. Ne4 Bb4+ 15. Nfd2 Qxg7 16. Rc8+ Ke7 17. Rxg8 Qxg8 18. Bxd7 Nxd7 19. O-O
Qg6 20. Qb3 Bxd2 21. Qa3+ Ke8 22. Nxd2 Qc2 23. Nb3 b6 24. Qa4 Rd8 25. Qxg4
Qxb2 26. Qa4 a5 27. h3 Ke7 28. Re1 Nf6 29. Qc4 Rd5 30. a4 Kf8 31. g3 Nd7
32. Re3 Rd1+ 33. Kg2 Qb1 34. Nd4 Rg1+ 35. Kh2 Rh1+ 36. Kg2 Qg1+ 37. Kf3
Qd1+ 38. Kg2 Nc5 39. Nf3 Nxa4 40. Ng5 Qg1+ 41. Kf3 Rh2 42. Nxe6+ fxe6 43.
Qc8+ Kg7 44. Qd7+ Kf8 45. Kg4 Rxf2 46. Rxe6 Rf7
{PostModernist resigns} 1-0
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.