Author: blass uri
Date: 17:32:28 12/11/99
Go up one level in this thread
On December 11, 1999 at 19:18:06, Dadi Jonsson wrote:
>The Allwermann incident has been discussed here recently. On a related note I
>have been assigned to the task of determining if one of the players in the
>Icelandic Internet Chess Championship used a chess program. The problem of
>determining this is very interesting, but of course it is not obvious how one
>can *prove* or *refute* such a theory. (Any ideas are welcome!)
>
>I have gone through many games from the tournament using several different
>methods with the help of Fritz 6. The following game was quite interesting with
>regard to Fritz's evaluation:
>
>[Event "Icelandic Internet Championship"]
>[Site "ICE, Reykjavik, Iceland"]
>[Date "1999.11.07"]
>[Round "8"]
>[White "Player1"]
>[Black "Player2"]
>[Result "1-0"]
>
>1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 e5 5. Nb5 a6 6. Nd6+ Bxd6 7. Qxd6 Qf6 8.
>Qd1 Qg6 9. Nc3 Nge7 10. g3 d5 11. exd5 Nb4 12. Bd3 Bf5 13. d6 Bxd3 14. cxd3
>Nxd3+ 15. Ke2 Nxc1+ 16. Rxc1 Nc6 17. d7+ Kd8 18. Re1 Nd4+ 19. Kf1 Qe6 20. Kg1
>Nc6 21. Ne4 Qxd7 22. Qh5 Kc7 23. Qxe5+ Kb6 24. Qc5+ Kc7 25. Red1 Qe6 26. Ng5
>Qf6 27. Ne4 Qe6 28. Nd6 Rad8 29. Qa5+ Kb8 30. Rxc6 Rxd6 31. Rcxd6 Qe7 32. Rd7
>Qe4 33. Qc7+ Ka7 34. Qc5+ Kb8 35. Qd6+ Ka7 36. Qc5+ Kb8 37. Rd8+ Rxd8 38. Rxd8+
>1-0
>
>White played 36. Qc5. Fritz's Blundercheck came up with the following
>"improvement":
>
>{0.00 36.Rd4 Qe1+ 37.Kg2 Qa5 38.Rxb7+ Kxb7 39.Rb4+ Qxb4 40.Qxb4+ Ka8 6.78}
>
>This means that Fritz saw 36. Qc5 as leading to an equal position, but scored
>36. Rd4 as 6.78 in White's favour.
>
>What is interesting about this is that the move played in the game, 36. Qc5,
>actually forces mate in three! Something one would expect the best chess
>programs to discover in a split second.
>
>I decided to investigate this further with both the Friz 5.32 and the Fritz 6
>engines, and neither one of the spotted that 36 Qc5 mates (ply 12).
>
>On the other hand, when I pasted the EPD posion into Fritz it had no problem
>finding the mate:
>
>7r/kp1R1ppp/p2Q4/8/4q3/6P1/PP3P1P/3R2K1 w - -
>
>(wKg1,Qd6,Rd1,d7,Pa2,b2,f2,g3,h2/bKa7,Qe4,Rh8,Pa6,b7,f7,g7,h7)
>
>This looks like a bug in Fritz 6.
36.Qc5+ leads to repetition of the position and most programs assume that a
repetition is a draw.
white could win by 35.Rd8+ faster with no repetition.
I believe that this is no bug in Fritz6 because a bug is something that the
programmers did not mean and in this case they meant that fritz will evaluate
repetition as a draw.
Uri
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.