Author: Slater Wold
Date: 22:37:32 07/23/02
Go up one level in this thread
On July 24, 2002 at 00:56:28, Dann Corbit wrote: >On July 24, 2002 at 00:17:05, Robert Hyatt wrote: > >>On July 23, 2002 at 23:45:30, Miguel A. Ballicora wrote: >> >>>On July 23, 2002 at 21:12:05, Robert Hyatt wrote: >>> >>>>On July 23, 2002 at 19:41:40, Slater Wold wrote: >>>> >>>>>On July 23, 2002 at 18:32:27, Robert Hyatt wrote: >>>>> >>>>>>A few years ago, Komputer Korner had a "correspondence kup" match with Crafty, >>>>>>chessmaster and some other programs, playing a correspondence tournament. In >>>>>>that match, Crafty found a wild sacrifice to play against Chessmaster, I think >>>>>>it was Nf6. After searching for 24 hours or whatever the time limit was, Crafty >>>>>>saw a draw score. And it saw a draw score for the next couple of moves. >>>>>> >>>>>>Right after Crafty played that move, Hsu sent me an email saying "brilliant >>>>>>move". I told him Crafty only saw a draw. He said "let it keep searching... >>>>>>DB Jr saw +2.5 very quickly." Sure enough, a few moves later Crafty realized >>>>>>it was not just drawing, it was winning easily. >>>>>> >>>>>>I will try to find the game. If anybody has it, perhaps _that_ position will >>>>>>be a good one to see how long it takes today's programs to find the move and >>>>>>find it is winning, compared to DB1 back in 1996... >>>>>> >>>>>>I will see what I can find... >>>>>> >>>>>>I think the event was called "the komputer kup korrespondence tournament" or >>>>>>something similar... >>>>> >>>>>The fastest, and highest score I get in 10 minutes is from, BIG SURPRISE, DJ7. >>>>> >>>>>Deep Junior 7 - W,S >>>>>r1b1r1k1/1q3ppp/ppn5/2bNp3/P4B2/5Q1P/BP3PP1/R2R2K1 w - - 0 1 >>>>> >>>>>Analysis by Deep Junior 7: >>>>> >>>>>1.Bh2 e4 2.Qf4 >>>>> ³ (-0.41) Depth: 3 00:00:00 >>>>>1.Bd2 e4 2.Qg3 Kh8 3.Qf4 >>>>> ³ (-0.30) Depth: 6 00:00:00 5kN >>>>>1.Bd2 e4 2.Qg3 Kh8 3.Qf4 >>>>> ³ (-0.30) Depth: 6 00:00:00 5kN >>>>>1.Bd2 e4 2.Qg3 Kh8 3.Qf4 >>>>> ³ (-0.30) Depth: 6 00:00:00 5kN >>>>>1.Bd2 e4 2.Qg3 Kh8 3.Qf4 >>>>> ³ (-0.30) Depth: 6 00:00:00 5kN >>>>>1.Bg5 Nd4 2.Qe4 Qc6 3.Rac1 >>>>> ³ (-0.27) Depth: 6 00:00:00 19kN >>>>>1.Bg5 Be6 2.Bb1 Rac8 3.Qd3 Bxd5 >>>>> ³ (-0.45) Depth: 9 00:00:00 228kN >>>>>1.Bd2 Be6 2.Qg3 Kh8 3.Qd3 Rad8 >>>>> ³ (-0.43) Depth: 9 00:00:00 233kN >>>>>1.Bh2 Bd4 2.Rac1 e4 3.Qb3 Bf5 >>>>> ³ (-0.34) Depth: 9 00:00:00 338kN >>>>>1.Bg3 Bd4 2.Nc3 Be6 3.Bxe6 Rxe6 >>>>> = (-0.16) Depth: 9 00:00:00 501kN >>>>>1.Bh6 Re6 2.Be3 Nd4 3.Qh5 g6 4.Qh6 Nf5 5.Qg5 Nxe3 >>>>> = (-0.06) Depth: 9 00:00:00 819kN >>>>>1.Bh6 Re6 2.Be3 Nd4 3.Bxd4 Bxd4 4.Rac1 e4 5.Qe2 Bc5 6.Qc4 >>>>> = (-0.17) Depth: 12 00:00:01 2649kN >>>>>1.Nf6+! >>>>> = (0.24) Depth: 12 00:00:02 5827kN >>>>>1.Nf6+! >>>>> ² (0.54) Depth: 15 00:00:07 18172kN >>>>>1.Nf6+! gxf6 2.Bd5 Bd7 3.Bh6 Bf8 4.Qxf6 Re6 5.Bxe6 Bxe6 6.Rd3 Ne7 7.Bxf8 Rxf8 >>>>>8.Rg3+ >>>>> ² (0.56) Depth: 15 00:00:17 38549kN >>>>>1.Nf6+ gxf6 2.Bh6 Bf8 3.Bd5 Bd7 4.Qxf6 >>>>> ² (0.58) Depth: 17 00:01:28 199636kN >>>>>1.Nf6+ gxf6 2.Bh6 Bf8 3.Bd5 Bd7 4.Bxf8 Kxf8 5.Qxf6 Re6 6.Bxe6 Bxe6 7.Ra3 >>>>> ² (0.60) Depth: 18 00:04:04 560089kN >>>>>1.Nf6+ gxf6 2.Bh6 Bf8 3.Bd5 Bd7 4.Bxf8 Kxf8 5.Qxf6 Re6 6.Bxe6 Bxe6 7.Ra3 >>>>> ² (0.60) Depth: 18 00:07:13 999089kN >>>>> >>>>>(W, 23.07.2002) >>>> >>>> >>>>Unfortunately the score is not yet right. Let it compute until it sees a >>>>big material win. That will give us a good time to compare to DB Jr... >>> >>>What was DB's time? >>> >>>Miguel >> >> >>It found it almost instantly. But it didn't get +2.5 until just under the two >>minute mark. I will try to post the analysis as soon as I find the email, if I >>still have it... Crafty used to find this fairly quickly, but with the wrong >>score. If a computer feels black is winning it might find Nf6+ more quickly >>by finding the repetition first, which is what happened to me in the KKUP >>tourney... I think we reached depth=16 on this position. >> >>It is also possible that the Crafty analysis and the DB Jr analysis is still >>lying around in DejaNews as I remember posting both. I first posted Crafty's >>analysis (I was not running crafty in any of the games but I set up the >>position when KK posted "Crafty unleashes a stunning sac against ChessMaster". >>I set up the position and ran for 8 hours on my P6/200 and posted the analysis >>with the note "it sees a repetition, not a big win." I later (after hearing >>from Hsu) posted his analysis as well). > >=================================================================== >KKUP's original article: >=================================================================== > >Message 3 in thread >From: Komputer Korner (korner@netcom.ca) >Subject: Korrespondence Kup update >Newsgroups: rec.games.chess.computer >View this article only >Date: 1997/02/27 > > > >You can catch all the action with 6 of the games having diagrams at >http://www.gambitsoft.com/kup.htm >After I had reported/predicted Crafty's first win against MacChess in >the Korrespondence Kup in another post, meanwhile in another of Crafty's >games, this time against CM5000, Crafty was sacking a piece!!! It turns >out that after a 25 ply combination!!!!, Crafty will end up with a Rook >and 2 pawns against B+N. Each side will have another Q,R, and 3 pawns. >Did Crafty see all this when it went in for the piece sac? I have to >conclude YES, and while this may be routine for the likes of Deep Blue, >it goes a long way to explain how the level of correspondence play has >dramatically improved over the last couple of years, not that I would >suggest that correspondence players would use computers.!!!!! Anyway, >will Crafty win its 2nd game seeing that the CM5000 king is stuck in the >centre with nowhere to hide? >-- >Komputer Korner > >The inkompetent komputer. > >=================================================================== >And your follow-on: >=================================================================== >Search Result 7 >From: Robert Hyatt (hyatt@crafty.cis.uab.edu) >Subject: Re: Korrespondence Kup update >Newsgroups: rec.games.chess.computer >View: Complete Thread (32 articles) | Original Format >Date: 1997/02/28 > > >Chris Whittington (chrisw@demon.co.uk) wrote: > >: -- >: http://www.demon.co.uk/oxford-soft >: Robert Hyatt <hyatt@crafty.cis.uab.edu> wrote in article >: <5f53hf$b7t@juniper.cis.uab.edu>... >: > Komputer Korner (korner@netcom.ca) wrote: >: > : You can catch all the action with 6 of the games having diagrams at >: > : http://www.gambitsoft.com/kup.htm >: > : After I had reported/predicted Crafty's first win against MacChess in >: > : the Korrespondence Kup in another post, meanwhile in another of Crafty's >: > : games, this time against CM5000, Crafty was sacking a piece!!! It turns >: > : out that after a 25 ply combination!!!!, Crafty will end up with a Rook >: > : and 2 pawns against B+N. Each side will have another Q,R, and 3 pawns. >: > : Did Crafty see all this when it went in for the piece sac? I have to >: > : conclude YES, and while this may be routine for the likes of Deep Blue, >: > : it goes a long way to explain how the level of correspondence play has >: > : dramatically improved over the last couple of years, not that I would >: > : suggest that correspondence players would use computers.!!!!! Anyway, >: > : will Crafty win its 2nd game seeing that the CM5000 king is stuck in the >: > : centre with nowhere to hide? >: > : -- >: > : Komputer Korner >: > : The inkompetent komputer. >: > >: > Dammit Whittington, you should have kept your infectious sacs out of my >: > damn program. :) >: How did that happen ? I thought when I ripped off your code verbatim to >: write my program, I'ld get some of your ideas :) >: Didn't realise that Crafty was able to reverse obtain algorithms via this >: process :) >: Clever program that Crafty. But then I found the clue in the source: >: if (crafty clone) >: dial hyatt >: transmit new ideas >: Chris Whittington > > >however you will notice that you are now on Crafty's noplay list on all >servers. :) No more of your evaluation contaminating my code.... :) > >In reality, Jason Deines and I analyzed this lastnight. It turns out that >the explanation is much less clever, and probably most programs will play >that Nf6+ sac. Crafty had played a pretty poor opening in this game, as >white, and the eval was -.6 the last time I checked. I think material was >even, but the position was not great, but also not lost. At depth=10, it >found Nf6+ led to a perpetual, which was 0.000, and played it. I believe, >at just under one hour per move in this tournay (to normalize against a 486/66) >it reached depth=13 or 14 here, but we still had an eval of draw when we checked >this out last nite... > >So, no magic tactical crush when it played that. The eval is certainly going >up, to about +1 at present, but what's going to happen is anyone's guess. I'd >rather be white myself, but there's no mate in sight. > >Therefore, "just" another case of deep/fast search finding a cheap way out of a >poor position, via a repetition. And then some luck thrown in that the >repetition turned out to not be the best option there was after the first move >was played OTB... > >I could not find Hsu's analysis. I searched *everywhere*. All I could find was: "One point that Hsu's going to shoot me about... in the Crafty vs CM5000 game in the Kup, Hsu said that Crafty was going to like the outcome of that sac *far* more than we knew at the time (we'd barely gotten to the +2 point a few moves later). He used "baby blue" and found a most impressive variation that I hope he'll post here one day... to show just what his "baby" machine can do. And he wasn't giving his "baby blue" machine an hour to search, either, which is frightening for several reasons..."
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.