Author: Uri Blass
Date: 11:52:56 09/10/01
Go up one level in this thread
On September 10, 2001 at 13:30:30, Robert Hyatt wrote: >On September 09, 2001 at 11:47:45, Uri Blass wrote: > >>On September 09, 2001 at 11:40:25, Uri Blass wrote: >> >>>On September 09, 2001 at 11:33:37, Günther Simon wrote: >>> >>>>>>Please take a look again at the former post of K.Burcham and you >>>>>>would see that it took Shredder 5 (in the position when F6 grabbed >>>>>>at g4) 41! minutes on a 1Gig Cpu to avoid it here. >>>>> >>>>>The point is about stronger programs like Deep Fritz and not about shredder5. >>>>> >>>>>Uri >>>> >>>> >>>>huh? Shredder won recently the SingleCPU WM... >>>> >>>>Günther >>> >>> >>>It is totally unimportant. >>>The version of shredder that was used was not shredder5. >>>Shredder5 is even not the latest commercial version >>> >>>Deep Fritz is the best commercial program and not shredder. >>>Deep Fritz also does not like 8.h3 on p800 >>> >>>[D]r2qkb1r/pp2pp2/2n3pn/1B5p/6b1/2N2N2/PPPP1PPP/R1BQ1RK1 w kq - 0 1 >>> >>>Analysis by Deep Fritz: >>> >>>8.Bxc6+ bxc6 >>> ± (1.22) Depth: 1/3 00:00:00 >>>8.d3! >>> +- (1.75) Depth: 1/7 00:00:00 >>>8.d3 Bg7 >>> +- (1.47) Depth: 2/9 00:00:00 >>>8.d4! >>> +- (1.53) Depth: 2/13 00:00:00 >>>8.h3! >>> +- (1.56) Depth: 2/13 00:00:00 >>>8.h3! Bf5 >>> +- (1.63) Depth: 2/13 00:00:00 >>>8.h3 Bxf3 9.Qxf3 Qd6 >>> +- (1.78) Depth: 3/9 00:00:00 >>>8.h3 Bf5 9.d3 Bg7 >>> +- (1.59) Depth: 4/15 00:00:00 1kN >>>8.d4! >>> +- (1.63) Depth: 4/20 00:00:00 5kN >>>8.d4! a6 >>> +- (1.66) Depth: 4/20 00:00:00 9kN >>>8.d4 a6 9.Ba4 Bg7 10.Bf4 >>> +- (1.56) Depth: 5/20 00:00:00 21kN >>>8.h3! >>> +- (1.59) Depth: 5/20 00:00:00 23kN >>>8.h3! Bd7 9.d4 Bg7 10.Bf4 >>> +- (1.72) Depth: 5/20 00:00:00 26kN >>>8.h3 Bd7 9.d4 Bg7 10.d5 Bxc3 11.bxc3 >>> +- (1.66) Depth: 6/19 00:00:00 36kN >>>8.h3 Bd7 9.d4 Bg7 10.Bf4 Nf5 11.d5 >>> +- (1.66) Depth: 7/21 00:00:00 72kN >>>8.h3 Bd7 9.d4 Bg7 10.d5 Nb4 11.Bc4 0-0 >>> +- (1.59) Depth: 8/24 00:00:00 206kN >>>8.d4! >>> +- (1.63) Depth: 8/24 00:00:00 488kN >>>8.d4! a6 9.Bxc6+ bxc6 10.h3 Bxf3 11.Qxf3 Rc8 12.Bxh6 >>> +- (1.66) Depth: 8/25 00:00:01 628kN >>>8.d4 a6 9.Bxc6+ bxc6 10.h3 Bf5 11.Ne5 Rc8 >>> +- (1.56) Depth: 9/30 00:00:02 1401kN >>>8.h3! >>> +- (1.59) Depth: 9/30 00:00:02 1500kN >>>8.h3! Bd7 9.d4 Bg7 10.Bf4 Nf5 11.Nd5 0-0 12.c3 >>> +- (1.69) Depth: 9/30 00:00:02 1626kN >>>8.h3 Bd7 9.d4 Bg7 10.Bf4 Nf5 11.Bxc6 Bxc6 12.d5 Bd7 13.Re1 >>> +- (1.59) Depth: 10/30 00:00:03 2178kN >>>8.h3 Bxf3 9.Qxf3 Rc8 10.Ne4 a6 11.Re1 Bg7 12.Nf6+ Kf8 13.Bxc6 bxc6 >>> ± (1.38) Depth: 11/30 00:00:10 6301kN >>>8.d4! >>> +- (1.41) Depth: 11/31 00:00:22 14206kN >>>8.d4! a6 9.Bxc6+ bxc6 10.h3 Bxf3 11.Qxf3 Rc8 >>> +- (1.50) Depth: 11/33 00:00:27 17762kN >>>8.d4 a6 9.Bxc6+ bxc6 10.h3 Bxf3 11.Qxf3 Rc8 12.Rd1 Nf5 13.d5 cxd5 >>> +- (1.53) Depth: 12/33 00:00:55 35629kN >>>8.d4 a6 9.Bxc6+ bxc6 10.h3 Bxf3 11.Qxf3 Qd7 12.Bxh6 Rxh6 13.Rfd1 Rb8 >>> +- (1.47) Depth: 13/36 00:02:05 74546kN >>> >>>(Blass, Tel-aviv 09.09.2001) >> >>Sorry >> >>Deep Fritz later considers 8.h3 as the best move so I cannot say that it avoid >>8.h3 but at least it avoids 9.hxg4 >> >>Uri > > >If it avoids hxg4 it does so because it sees the loss. However, bring it to >ICC and let a couple of experts on this trojan horse attack try it. They are >very good at constructing positions where the program _can't_ see the loss >in any reasonable time limit. They did it to me several times as I was working >on the anti-trojan code currently in Crafty. If the program relies on search to >avoid taking the piece, it _will_ lose games to this attack when they are set >up right. The program has to know that it simply can't take, period, and once >it ignores the piece, things might work out fine for it. I simply does not like this solution because this solution means that the program is going to avoid capturing the piece when capturing the piece is winning. I believe that in part of the cases when programs lose the mistake is not capturing g4 but later moves. I prefer to see programs solve this problem at tournament time control by special extensions for typical important black moves(hxg4,Qh4,Bd6,g3 if white plays f3). Uri
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.