Author: Uri Blass
Date: 02:30:49 02/21/03
Go up one level in this thread
On February 21, 2003 at 04:51:20, Janos Keinrath wrote: >On February 21, 2003 at 04:34:13, Uri Blass wrote: > >>On February 21, 2003 at 04:30:52, Janos Keinrath wrote: >> >>>On February 21, 2003 at 04:16:28, David Rasmussen wrote: >>> >>>>On February 21, 2003 at 03:49:48, Janos Keinrath wrote: >>>> >>>>>[D] 4k2r/1pp1bppp/p3bn2/2p3B1/4P3/2N2P2/PPP1N1PP/3K3R b k - >>>>> >>>>>In a blitz game here my program played 1...Ng4? . >>>>> >>>>>Only after few moves 2.Bxe7 Nf2+ 3.Ke1 Nxh1 4.Bxc5 b6 5.Be3 b5 6.a3 >>>>>sees the program the knight is lost. >>>>> >>>>>How much time need your program to avoid the loosing move? >>>> >>>>Well, I don't believe Ng4 in itself is the losing move. Nf2+ is. Chezzz avoids >>>>Nf2+ altogether, probably because of it's trapped piece code: >>> >>>After Ng4, Nf2+ is force. (2...Kxe7?? 3.fxg4 +-) >> >>What about 2...Ne3+? >> >>Uri > >You are right, I missed this move. >After 2...Ne3+ 3.Kd2 Nc4+ 4.Kc1 Kxe7 white stands better, but not winning. > >Most program here wants to play Nf2+, because they don't see the knight >will be captured. The game was a 5min blitz. Movei does not need to see that the knight is captured in order to avoid Nxh1. It knows by piece square table that a big part of the knight(that is clearly more than a pawn) is lost if the knight is at h1. If Ng4 is a positional mistake than avoiding it is a bigger problem and movei cannot see it(it can see a negative score for Ng4 -0.07/12,-0.14/13 but finds nothing better in a few minutes). 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.