Author: Graham Laight
Date: 16:49:41 01/05/01
Go up one level in this thread
On January 05, 2001 at 10:25:54, Uri Blass wrote: >On January 05, 2001 at 09:54:15, Robert Hyatt wrote: > >>On January 05, 2001 at 03:16:49, Ed Schröder wrote: >> >>>On January 04, 2001 at 19:19:12, Graham Laight wrote: >>> >>>>On January 04, 2001 at 17:43:30, Ed Schröder wrote: >>>> >>>>>On January 04, 2001 at 12:20:41, Ulrich Tuerke wrote: >>>>> >>>>>>On January 04, 2001 at 12:09:25, Enrique Irazoqui wrote: >>>>>> >>>>>>>[Event "?"] >>>>>>>[Site "?"] >>>>>>>[Date "2001.01.04"] >>>>>>>[Round "?"] >>>>>>>[White "Rebel Century 3"] >>>>>>>[Black "van der Wiel, John"] >>>>>>>[Result "0-1"] >>>>>>>[ECO "B15"] >>>>>>>[BlackElo "2531"] >>>>>>>[PlyCount "118"] >>>>>>>[EventDate "2001.01.02"] >>>>>>>[SourceDate "2001.01.04"] >>>>>>> >>>>>>>1. e4 c6 2. d4 d5 3. Nc3 a6 4. Be2 b5 5. a3 e6 6. Nf3 Nf6 7. e5 Nfd7 8. Bg5 Be7 >>>>>>>9. Bxe7 Qxe7 10. b4 a5 11. Na2 axb4 12. axb4 O-O 13. Bd3 f6 14. exf6 gxf6 15. >>>>>>>O-O Rf7 16. Re1 Nf8 17. Nh4 Qd8 18. Re3 Rg7 19. Rg3 Qe7 20. Qf3 Rxg3 21. Qxg3+ >>>>>>>Qg7 22. Qd6 Qd7 23. Qf4 Qg7 24. Nf3 Bd7 25. Qc7 Be8 26. Qd8 Bg6 27. h3 Qf7 28. >>>>>>>Bxg6 hxg6 29. Nh2 Kg7 30. Ng4 Nfd7 31. Qc7 e5 32. Qb7 Ra3 33. dxe5 fxe5 34. >>>>>>>Nxe5 Nxe5 35. Qxb8 Qf6 36. Re1 Rxa2 37. Qxe5 Qxe5 38. Rxe5 Rxc2 39. Re1 Kf6 40. >>>>>>>Kf1 d4 41. Re8 Rc4 42. Rc8 Ke7 43. Ke2 Kd6 44. Rd8+ Kc7 45. Rg8 Rxb4 46. Kd3 c5 >>>>>>>47. Rg7+ Kb6 48. Rxg6+ Ka5 49. h4 Rb3+ 50. Ke4 d3 51. Rd6 c4 52. g4 Rb2 53. g5 >>>>>>>Re2+ 54. Kf3 Re8 55. Rd4 b4 56. Rxc4 b3 57. Rc1 b2 58. Rd1 Kb4 59. Kf4 Kc3 0-1 >>>>>>> >>>>>>>Enrique >>>>>> >>>>>>A convincing demonstration that anti-computerchess played by a GM is still too >>>>>>much for todays programs. >>>>>>This opponent is indeed a very brave choice of the Rebel team. I think that >>>>>>there are many opponents with higher ELO, but easier to play for Rebel, - my >>>>>>deep respects to Ed for playing John van der Wiel. >>>>>>And btw a very interesting game ! >>>>>>Uli >>>>> >>>>>Thank Uli for the encouraging words. John v/d Wiel after the game said he >>>>>wasn't impressed by Rebel's play in game 1 and 2 but in game-3 he had a >>>>>very hard time and was impressed by Rebel's pressure. He also said that >>>>>39.Re1 was Rebel's only mistake (39.Re6! Kf7 40.Rd6! is probably a draw) >>>>>and thereafter white was lost. >>>>> >>>>>He also mentioned that during the middlegame white maybe could have played >>>>>Re1 sacrificing the kningt on a2 for a strong attack. I forgot about the >>>>>move number. I would be interested to know if anyone noticed this. >>>>> >>>>>2-1 for GM John v/d Wiel still 3 games to go starting next tuesday. >>>>> >>>>>During the 3 games so far I noticed the following pattern by John: >>>>> >>>>>. force the game only into strategic lines even if there (probably) is a >>>>>better move. >>>>> >>>>>What can one do? >>>>> >>>>>Ed >>>> >>>>If JVDW is "playing the player", then play him at his own game. >>>> >>>>There is an option which I'm 90% sure will win the next game (but has a 10% >>>>chance of making you look silly). >>>> >>>>Bob once mentioned that, long ago, due to some sort of error, the clocks didn't >>>>work properly on Cray Blitz. The computer played its moves much too quickly. >>>>What happened was that, while the computer undoubtedly played marginally worse, >>>>the human opponents played a lot worse, and did far more badly than the form >>>>guide would have predicted. >> >>Remember that those were _blitz_ games. I don't think this will work at all >>in a 40/2 game. You will just get crushed tactically on the moves you don't >>predict correctly (and therefore use a short search time). When's the last time a human beat a computer on tactics??? >Ed explained that Rebel used average time of 1:45 minutes per move instead of >2:45 minutes per move in the first game. > >The difference in elo is not big(it is probably not more than 40 elo against >computers) > >The difference if one program is 165/105=1.57... times slower is close to be 50 >elo assuming 70 elo for doubling but if you consider the fact that you have more >time to use when you ponder when the opponent has less time to ponder the >difference is smaller. > >I think that the chances to get crushed tactically against computers because of >this decision is small against computers and it is not clear if it is not a good >decision against humans. > >Uri At the moment, Rebel is being outplayed anyway. What we need to do is to find ways to bring down JDVW's standard of play. As you point out, cutting Rebel's thinking time could diminish its Elo rating by over a hundred, if you cut the time to move to 60 seconds (as I would). However, I think that the effect on the human opponent's Elo level would be much greater - they're not used to having moves consistently thrown back at them this quickly. The requirement to get a mistake from JVDW is also the reason why I suggested setting Rebel to play in a safe, bland style - to lull him into a false sense of security. According to Ed further back in this thread, JVDW is "playing the opponent" rather than "playing good chess". If this is so, it represents a high compliment - which Ed should do JVDW the honour of returning! -g
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.