Author: Mats Winther
Date: 01:31:56 04/11/98
Sorry for not answering all of the responses to my previous messages. I
generally agree with Fernando's latest message (One more shot...) and
Thorsten's view is definitely worth consifering.
I checked that position prior to blacks -Na5? in game two in my previous
message (The Fritz5 discussion). I used Watch Mode with almost 22 Mb
hash. Initially the Fritz5 module thinks of -b4, but after a couple of
minutes he changes to -Na5 and stays there for a couple of minutes. Then
he goes back to -b4 as the best move. So you guys that cannot reproduce
-Na5 can check that these changes occur in Watch Mode.
The thing is that Fritz5 doesn't understand that Na5 ruins the position
since the pawn chain is wrecked. Fritz5 has to count all the way to the
pawn loss before he understands that the move is bad. Any human can see
that -Na5 is equal to resigning the game.
I have another example of the same error in the following game where I
let them follow a theoretical line a little longer. Time was
90´/30+60´20+30´(no Power Books) on 686/200+.
Perhaps Fritz5 cannot live up to its rating on my computer, although my
computer is rather strong. So the question (which you have debating, I
suppose) is then: (1)Is Fritz5 the leader of the 60Mb hash league or (2)
is Fritz5 the leader of the overall league?
Don't people have any opinion on the problem of the playing style?
I maintain that if I were to introduce a strong chessplayer to computer
chess and presented the "strongest" program to him (Fritz5) and he
played any of the referred games against the program, then he would
say to me afterwards: - Don't talk to me about computerchess anymore!
I think we have to be much more critical about the style of the
programs. I think that the programs must be programmed to be strong
against humans - not necessarily strong against computers. The Fritz5
team uses a different programming strategy than the others. I maintain
that this would damage the development of "strength against humans" if
the other programmers were to follow. I am convinced that Fritz5
deserves a lower rating against humans than against computers (but that
remains to be shown). One cannot hope for such strong computers that the
programmers can revoke the positional intelligence of the programs. Deep
Blue used a fast counting strategy too on a super computer. Up to the
previous match it played ridiculous chess. Some of the games of the
previous match against Kasparov were completely stupefying.
/Mats Winther
[Event "test game no 3"]
[Site "?"]
[Date "1998.??.??"]
[White "Hiarcs6"]
[Black "Fritz5"]
[Result "1-0"]
1. d4 d5 2. c4 dxc4 3. Nf3 Nf6 4. Nc3 a6 5. e4 b5 6. e5 Nd5 7. a4 Nxc3
8. bxc3 Qd5 9. Be2 Bb7 10. 0-0 e6 11. Ne1 Qd7 12. Bf3 Nc6 13. Nc2 Rb8
14. Qe2 Nd8 15. Bxb7 Nxb7 {theory ends} 16. Rd1
{
1. . r . . k b . r
2. . n p q . p p p
3. p . . . p . . .
4. . p . . P . . .
5. P . p P . . . .
6. . . P . . . . .
7. . . N . Q P P P
8. R . B R . . K .
A B C D E F G H
}
16. ...Na5? {Again! After this blacks pawn chain is ruined.} 17. axb5
Qxb5 18. Bg5 c6 19. f4 h6 20. Bh4 Rg8 21. Ne3 Rh8 22. f5 Nb3 23. Rab1
exf5 24. Nxf5 Qd5 25. Ne3 Qe4 26. Qf2 Rb7 27. Nxc4 Qd5 28. Qf5 Qe6
29. Qxe6+ fxe6 30. Nd2 g5 31. Be1 Na5 32. Ra1 Nb3 33. Nxb3 Rxb3
34. Rxa6 Kd7 35. Ra7+ Kc8 36. c4 Ra3 37. Rxa3 Bxa3 38. Ra1 Bf8
39. Ra8+ Kd7 40. Bb4 Bxb4 41. Rxh8 Bc3 42. Rh7+ Ke8 43. Rxh6 Bxd4+
44. Kh1 Kf7 45. Rf6+ Ke7 46. Rg6 Be3 47. Rg7+ Kf8 48. Rc7 c5 49. g4
Bd4
1-0
This page took 0.02 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.