Author: Robert Hyatt
Date: 07:44:54 11/22/98
Go up one level in this thread
On November 22, 1998 at 02:31:47, blass uri wrote: > >On November 21, 1998 at 19:12:59, Robert Hyatt wrote: > >>On November 21, 1998 at 15:55:11, blass uri wrote: >> >>> >>>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. >>> >>>I do not see something singular in this position >>>white can play many moves without being -2.xx and even without being-1.xx >> >> >>better look again. First, there is *definitely* something singular if black >>plays c5. White has to move the bishop. (remember it is black to move in >>the above game and has to find the move c5 with a score of +2.x for black). >>If you don't do something right *now* after c5, the bishop is trapped. But >>it is in fact *very* deep. Deep enough that CB didn't find it... > >I do not see how the bishop is trapped after other moves like 28.b4 and I see >nothing forced in the position(even at move 34 of the game Junior evaluates the >position as -0.xx with another move Rff1) >> What junior thinks doesn't matter. Cray Blitz thought exactly the same thing and it searched a good bit deeper than Junior. That was the whole point of this exercise... this is apparently too deep for any of them, including anything I've ever done, but it wasn't too deep for deep thought... >> >>> >>>If there are singular extensions then there must be many extensions that begin >>>far from the root. >> >> >>as I said, perhaps you haven't read Hsu's paper on singular extensions and >>understand what this algorithm is. But from the initial position, if black >>plays c5, then white has to walk a tightrope to avoid immediately losing the >>bishop, and at the end it falls anyway... > >I do not see something singular for white. >Junior5 is going to play 29.h3 after 27...c5 28.Be4 Ra6 with evaluation of -0.24 >after an hour on pentium200MMX Again, because you don't "see something singular" doesn't mean there is nothing singular. A "singular move" is nothing more than a move that is about 1/2 of a pawn (or more) better than all other moves at that point in the tree. I don't remember the exact singular margin they used, it is in the ICCA Journal, but that's at my office... But singular extensions happen in lots of places where you wouldn't think they happen... > >Genius3 also play 29.h3 after 1 hour and fifty minutes on pentium100 with >evaluation of -0.54 50 minutes on a P100 has *nothing* to do with the search dt did... and they don't do the same kind of search by any comparison either... > >I see that white has many options not to lose the bishop so I do not understand >how singular extensions can help because I do not see one move that is clearly >better than the others at *any* point in the tree? Not just at the root? Singular extensions apply 20 plies deep as well as at the root... *every* chess position has some singular moves when searched... >> >> >> >> >>> >>>If I were white then I would play 28.b4(Junior5 changes its mind from Be4 to b4 >>>and now to Rd2 without a losing evaluation for white) >>> >>>I want to see a tree that prove Junior5 or Fritz5 that the evaluation of the >>>position is less than -1 (I mean the evaluation at the leaves after search at >>>tournament time control is less than -1, and the evaluation after every move >>>of white that go out of the tree is less than -1) >>> >>>Uri >> >> >>I'm not sure what you mean "I want to see a tree ..."... I *saw* the output, >>I was on the *wrong* end of all this... I saw their score, I watched our score, >>thinking their score was the result of a bug. It wasn't... >It may be because of a different evaluation function or because of a bug > >I do not believe that it is because of a tree before I see a part of the tree >that convince my Junior5 that black has more than a pawn advantage. > >I mean that in every leaf of the tree the evaluation(after search of 3 minutes) >is more than a pawn advantage for black and white cannot do a move to get out of >the tree with better evaluation. > >Uri I'll try to ship this to you. Do you have a train track close to your house? because the output might well have about 10 *million* pages of output. :) I don't have their "output" for this move. As I said before, we sat at the same table playing this game at the 88 ACM event (I think). I saw their output, they saw ours. We both saw them fail high with a score > 2.0, while we were reasonably happy with our score... until the roof fell in about 10 moves later... and their eval didn't vary by much for the entire sequence... So I can't give you their output, since I don't have it (they were using a laptop to display their stuff). I can only tell you what actually happened in the game. Another such game was against *socrates in cape may in 1994. They had a big + eval about 10 moves before *socrates failed low. Everyone was assuming they had a "bug" or that their eval was skewed based on king safety. It wasn't. It was skewed based on something it saw *very* deep in the tree. Mike Valvo discussed this game again the next night and said "last night, Deep Thought gave *Socrates a chess lesson they won't *ever* forget, by seeing something so deep only they knew the game was over, and they knew it so far ahead of *socrates that it was difficult to believe." (this from a *socrates that was running on a big parallel machine so it wasn't a "pc" program at all that year, either.)
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.