Author: Robert Hyatt
Date: 11:30:39 11/21/98
Go up one level in this thread
On November 21, 1998 at 13:36:46, Bruce Moreland wrote: > >On November 21, 1998 at 13:03:07, Robert Hyatt wrote: > >>I think their search is difficult to understand. IE I'll point back to the >>position I posted last year on r.g.c.c about the c5 move in a game against >>Cray Blitz, in Orlando at the 88 or 89 ACM event. They played c5 after >>failing high to +2.x, the game went *10* full moves further before *we* >>failed low to -2.x... I was looking right at their output and they had >>this incredibly long PV showing that the bishop was going to be lost. They >>saw it 20 full plies before we did. Lots of micros tried this position last >>year, and almost all would play c5 (as we expected that reply ourselves in >>the real game). But *none* had any clue that it was winning material.. even >>when they went far into the variation... > >[Event "ACM 1991"] >[Site ""] >[Date ""] >[Round ""] >[White "Cray Blitz"] >[WhiteElo ""] >[Black "Deep Thought II"] >[BlackElo ""] >[Result "0-1"] > >1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Nc6 6. f4 e5 >7. Nxc6 bxc6 8. fxe5 Ng4 9. Be2 Nxe5 10. Be3 Be7 11. O-O Be6 >12. Qd4 O-O 13. Rad1 f6 14. b3 Qe8 15. Na4 Qg6 16. Bf4 Rf7 17. Qe3 >Raf8 18. Qxa7 Qxe4 19. Bd3 Qb4 20. Qe3 Ra8 21. c3 Qb7 22. Rf2 >Qa7 23. Qxa7 Rxa7 24. Be3 Ra5 25. Bb6 Ra8 26. Bc2 Bf8 27. Re1 >c5 28. Be4 Ra6 29. Rb1 f5 30. Bc2 Rb7 31. Bd8 g6 32. Re1 c4 >33. Rb1 Bd7 34. Nb2 Ra8 35. Bg5 Rxa2 36. b4 Bb5 37. Re2 Bg7 >38. Nd1 Ra6 39. Bd2 Nd3 40. Ne3 Ra2 41. Bxd3 cxd3 42. Rf2 Rxd2 >43. Rxd2 Bxc3 44. Nf1 Bxd2 45. Nxd2 Re7 46. Nf3 h6 47. Rb2 Re4 >48. Kf2 g5 49. g3 f4 50. gxf4 Rxf4 51. Kg3 h5 52. Nd2 h4+ 53. Kg2 >Bc6+ 54. Kg1 Rg4+ 55. Kf2 Rg2+ 56. Ke3 Bb5 57. Ra2 Rxh2 58. Ra5 >Re2+ 59. Kd4 h3 60. Rxb5 Rg2 61. Rb8+ Kg7 62. Rb7+ Kg6 63. Rd7 0-1 > >r4bk1/5rpp/1Bppbp2/4n3/N7/1PP5/P1B2RPP/4R1K1 b - - 7 27 > >The goal is to search this position and achieve a score of approximately +2. It >is possible to find the move for other reasons, perhaps a sniff of danger, but >Bob says that Deep Thought saw to the end of this. > >>The stuff they do with singular extensions and threats really shines in some >>positions... and probably costs them dearly in others... But they have the >>horsepower to pay the price when it doesn't work, and then they kill us when >>it does.. >> >>In Cape May (94 ACM) we ran with the full singular-extension algorithm >>enabled and promptly lost the first game we played, because we searched out >>the bottom of our 60 ply limit, and we didn't detect that... > >It sounds like Deep Thought used a search that wasn't anything like a normal >null-move search. I'm willing to believe that it can find some shots, if for no >other reason (and I am sure there are other reasons) that there will be >positions that anything that is different than everything else will do better >on. > >bruce The best example of this comes up in games between "the king" and Crafty. At times the king looks brilliant, finding deep mates in a fraction of a second, and killing me by doing so. But at other times, it looks silly because it is probing checks so deeply it overlooks an 8 ply plan that gets a rook stuck right in its face... because it bogged down at depth=6 chasing checks to kingdom come and back. But in the case of deep blue, they get away with both... they run stuff so deeply that they find tactical shots we miss, yet their basic 10-11 ply search (with no null-move errors of any kind) is more than enough to prevent us from finding those "deeper" positional plans that they would overlook. Very frustrating machine to play against, from a computer's perspective. :)
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.