Author: Peter Berger
Date: 10:05:12 09/24/01
Go up one level in this thread
On September 23, 2001 at 14:25:58, Michel Langeveld wrote: >Ok I start with ECM.1134: > >You are probably right in this: > >2r2rk1/1p1q1ppp/1p3n2/3p1N2/4n3/1N3Q2/PPP2PPP/R2R2K1 w - - bm Rxd5; id >"ECM.1134"; > >It is no problem to find Rxd5 here - but is it really the best move ? > >0:00:00.2 ( 7/13) 46412 0.19 1.Nbd4 Rfe8 2.Re1 Ra8 3.c3 (Mat=0,50=0) >0:00:00.3 ( 7/13) 84545 0.20 1.Rxd5 (Mat=100,50=0) >0:00:00.4 ( 7/15) 107045 0.69 1.Rxd5 Ng5 2.Rxd7 (Mat=1100,50=0) >0:00:00.6 ( 8/17) 184103 0.50 1.Rxd5 Qe8 2.Nd6 Nxd6 3.Rxd6 Rxc2 4.Rxb6 >Qe2 (Mat=100,50=1) >0:00:01.6 ( 9/21) 414101 0.67 1.Rxd5 Ng5 2.Rxd7 >0:00:03.5 (10/24) 1006908 0.45 1.Rxd5 Qe8 2.Nd6 Nxd6 3.Rxd6 Qe5 4.Rxb6 >(Mat=200,50=0) >0:00:07.9 (11/27) 2212733 0.48 1.Rxd5 Qe8 2.Nd6 >0:00:16.8 (12/28) 5107430 0.29 1.Rxd5 Qe8 2.Nd6 >0:00:55.6 (12/33) 15894314 0.30 1.Nbd4 (Mat=0,50=1) >0:01:16.1 (12/33) 21718210 0.31 1.c3 (Mat=0,50=0) > >I tried with a few programs and my own brain: Rxd5, Nbd4 and also Ne3 to some >extent all look very promising. >My impression is that this testposition is still correct but I can't prove. >If I had to play this position myself after looking at some of the lines I'd >definitely prefer 1.Nbd4 :). > > >Score after Nbd4: > >13 4:14 0.73 1. ... g6 2. Nh6+ Kg7 3. Qe3 Rfe8 f3 Nc3 5. Nhf5+ Kg8 6. >Qh6 Ne2+ 7. Kf1 Nh5 8. Re1 Nef4 9. g4 gxf5 10. gxh5 > >Score with Rxd5: >17 48:36 0.99 1. Rxd5 Qe8 2. Re1 Nxd5 3. Rxe4 Qc6 4. Nbd4 Qf6 5. Rg4 Ne7 >6. Rxg7+ Kh8 7. Qh5 h6 8. Nxe7 Kxg7 9. Ndf5+ Kh7 10. Nxc8 Rxc8 > >So whites score is really different. I am not sure which side you are on ;-) . The Crafty analysis seems to confirm that it isn't clear at all that Rxd5 is the best move. In general I don't think you can prove right or wrong in these difficult positions by just letting your engine search forever ( I know you know this but it's a message board,isn't it :)?). The lines have to be played out and you have to look at positions deeper in the tree when the program has a chance to get a stable evaluation that can be trusted - often you even need some human interference ( look at Miguel's 4.fxe6!( ECM.1622 ?)- you won't get that soon with Crafty et al) . I am extremely short of time during this week so I can't do any analysis myself but maybe it would be a good idea to slowly move forward in the trees ? I think ECM.1134 remains unclear so far although I am convinced Rxd5 will survive :). Cheers, pete
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.