Author: Mark Schreiber
Date: 18:37:35 01/07/01
Go up one level in this thread
On January 07, 2001 at 10:07:01, Uri Blass wrote: >On January 07, 2001 at 09:14:37, Mark Schreiber wrote: > >>On January 07, 2001 at 07:41:41, Uri Blass wrote: >> >>>On January 07, 2001 at 07:24:38, Mark Schreiber wrote: >>> >>>>On January 05, 2001 at 19:30:02, Enrique Irazoqui wrote: >>>> >>>>>On January 05, 2001 at 16:22:34, Mark Schreiber wrote: >>>>> >>>>>>On January 05, 2001 at 12:57:19, Ernst Walet wrote: >>>>>> >>>>>>>Still, according to me, you cannot completely compare the two matches, as in >>>>>>>Dortmund Junior played each opponent just once (as far as I know), while Rebel >>>>>>>plays the same opponent six times. >>>>>>> >>>>>>>Ernst. >>>>>>> >>>>>> >>>>>>You’re right, we can not compare the 2 matches. The Super-GM Dortmund was harder >>>>>>for Junior. It had 9 rounds instead of 6 rounds Rebel is playing. So there was >>>>>>more opportunity for the humans to find and learn Junior’s mistakes. There were >>>>>>more humans looking for Junior’s mistakes. Also Dortmund had much stronger >>>>>>players. I don’t think v/d Wiel would have any chance with the 8 processor >>>>>>Junior. >>>>>>Mark >>>>> >>>>>Van der Wiel never lost to a computer in a slow game. I don't know about Junior, >>>>>but Fritz on a multiprocessor machine lost to van der Wiel half a year ago in an >>>>>official game. If you look at it you will realize what he does to programs. >>>>> >>>>>[Event "ch-NED"] >>>>>[Site "Rotterdam NED"] >>>>>[Date "2000.05.19"] >>>>>[Round "11"] >>>>>[White "Van der Wiel, J."] >>>>>[Black "Fritz SSS"] >>>>>[Result "1-0"] >>>>>[ECO "D00"] >>>>>[WhiteElo "2558"] >>>>>[PlyCount "91"] >>>>>[EventDate "2000.05.07"] >>>>> >>>>>1. d4 d5 2. c3 Nf6 3. Bg5 Ne4 4. Bf4 g5 5. Bc1 h6 6. e3 Bg7 7. Bd3 Nd7 8. c4 >>>>>Ndf6 9. f3 Nd6 10. c5 Nf5 11. Ne2 g4 12. f4 Qd7 13. Nbc3 Qe6 14. Qd2 Bd7 15. b4 >>>>>h5 16. a4 O-O-O 17. Kd1 h4 18. b5 Kb8 19. Rb1 h3 20. g3 Be8 21. a5 Ka8 22. Ke1 >>>>>Bd7 23. Kf2 a6 24. Qc2 Rb8 25. Bd2 axb5 26. Nxb5 Bxb5 27. Rxb5 Ne4+ 28. Bxe4 >>>>>Qxe4 29. Qxe4 dxe4 30. Nc3 e6 31. Nxe4 Ne7 32. Ng5 Rhf8 33. Rhb1 Ka7 34. a6 >>>>>bxa6 35. Rxb8 Rxb8 36. Rxb8 Kxb8 37. Nxf7 Kc8 38. Ng5 Kd7 39. Ke2 Nf5 40. Ne4 >>>>>Kc6 41. Nf2 Nh6 42. Ba5 Bf6 43. Kd3 Kd7 44. e4 Bg7 45. Kc4 Kc6 46. Bd2 1-0 >>>>> >>>>>Enrique >>>>> >>>> >>>>All of Van der Wiel games except the one against Fritz sss were played in 1997 >>>>and earlier. That’s 4 years ago. A long time for chess programs. That’s 164 SSDF >>>>points. >>> >>>We can compare Van der Wiel results to the results of other players 4 years ago. >>> >>>It will be interesting to calculate the performance of the players who played 4 >>>years ago against computers(you can give every program rating that is equal to >>>it's performance against humans for the calculation). >>> >>>We can get an estimate for van der Wiel's rating against computers by this >>>calculation. >>> >> >>I am not sure I understand you. If we get Van der Wiel performance against >>computers for 1997 how will that tell how Van der Wiel will perform against >>today’s programs on today’s computers. > >Suppose we find that Van der Wiel's performance against computers was 2750(we >calculate the performance by assuming that the rating of the programs is equal >to their performance against humans) > >Suppose also that Van der viel's rating at 1997 was 2600. > >We can use this information to guess that Van der Wiel's rating against >computers is 150 elo better than his rating against humans. > >If we evaluate computer programs like Deep Junior(8 processors) as 2700 against >humans and we also evaluate Van der Wiel as 2700 against computers then it means >that we can expect a draw in a 6 game match between Van der Wiel and Deep >Junior(8 processors) > >Uri That’s a lot of supposes. To test this we need to have a match between Van der Wiel and Deep Junior on 8 processors. Mark
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.