Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Killer line against computers

Author: Robert Hyatt

Date: 07:23:45 11/14/98

Go up one level in this thread


On November 14, 1998 at 09:42:50, Chris Janeke wrote:

>The following line was published on one of the rec. chess groups about a year
>ago.   Although the line is definitely not sound against accurate counterplay,
>I've had succees with it in 5 minute blitz games against heavy weights such as
>Crafty 15.20, Phalanx 18, Gnuchess, Gromit, Exchess, Ychess, Rebel Decade2.0,
>Mirage, Rookie, Zchess  .. the list goes on (but neither Chessmaster 6000 nor
>Fritz 5 appear to fall for it).
>
>Here is my win against Crafty 15.20 (Pentium 133Mhz, 32 Meg Ram and using the
>5,7 meg book bin file for Crafty - which is available at the crafty download
>site - and no tablebases).
>
>1 e4 e5 2 Nf3 Nc6 3 Bc4 Nf6 4 Nc3 Nxe4 5 Nd5 Bc5 6 d3 Nxf2 7 Qe2 Nxh1 8 Ng5 0-0
>9 Nxc7 Qxc7 10 Qh5 Bf2 11 Kf1 h6 12 Bxf7+ Kh8 13 Qg6 Ng3+ 14 hxg3 hxg5 15 Qh5#
>
>A I recall, the critical moment is after 9 Nxc7 where if  black does not take
>the knight, white's game falls apart.   Of course, after 9 ..Qxc7, black can no
>longer prevent the mate.  It seems however that many chess programs are just too
>greedy  to explore the consequences of taking the knight fully.



Also you are making too many generalizations based on playing on a very slow
computer, at a fast time control, which invites this kind of mis-analysis...
To see what I mean, I'd be happy to set crafty's book (on ICC) so that it
will only play this line and let you play a series of games, you having white
in each one...  I suspect that my faster hardware will turn this into a blood-
bath, but am not certain of course.  But I'm willing to try it.

This is a common problem in that I have tuned what I do based on a processor
of the speed of the P6/200...  the P5/133 is about 2.5 times slower than a
single P6, which means that Crafty is going to suffer from quite a few null-move
search failures...  Thorsten saw this many times when he was using a similar
machine...  My assumptions simply don't work on slower hardware.  But on a good
piece of hardware I don't think it will fall for that..



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.