Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Are computer ugly looking moves better moves?

Author: Telmo Escobar

Date: 22:54:03 06/30/02

Go up one level in this thread


On June 30, 2002 at 04:22:06, Mike S. wrote:

>
>Another thing are the "typical computer moves" (not necessarily ugly, but
>unsually very surprising). An example:
>
>Nimzo35 P90 - Kallisto183 486/50 [C40]
>SSDF / Bertil Eklund, 1997
>
>1.e4 e5 2.Nf3 f5 3.Nxe5 Nc6 4.Qh5+ g6 5.Nxg6 Nf6 6.Qh3 Nd4 7.Nxh8 Nxc2+ 8.Kd1
>Nxa1 9.e5 Bg7 10.exf6 Qxf6 11.Nc3 Bxh8 12.Nd5 Qe5 13.Bc4 c6 14.Qh5+ Kd8 15.Qh4+
>Ke8
>[D]r1b1k2b/pp1p3p/2p5/3Nqp2/2B4Q/8/PP1P1PPP/n1BK3R w q - 0 16
>16.Nc7+!! A typical computer move.
>

 Alas, how can you call this a computer move? 16.Nc7+ is the first move I think
about!
 This example makes me suspect the very idea of "computer move" is based upon a
misunderstanding.




 16...Qxc7 17.Qh5+ Kd8 18.Qg5+ Ke8 19.Qg8+ Ke7
>20.Qf7+ Kd6 21.Qf8+ Ke5 22.Qe7+ Kd4 23.d3 Nc2 24.Be3+ Nxe3+ 25.Qxe3# 1-0
>
>>Maybe so, because Fritz is clearly comeing up with 2600-2700 rated moves,
>>on the average. Are these ugly moves also rated 2600-2700?
>
>Moves themselves aren't elo-rated, but since they are a part of the program's
>play and success, I'd say yes.
>
>>Do these moves point to
>>holes in human cognition that will eventually allow computers to eclipse human
>>beings in strength?
>
>Obviusly very "unusual" moves are easier to miss by a human. OTOH, any
>combination will start with an unusual move, by definition. So I think the
>problem isn't too big. But it may partially explain why computers are so strong
>in blitz, when the human must rely more on known patterns. I think, especially
>under blitz conditions typical and/or ugly computer moves have an anti-human
>effect (i.e. force you to calculate much more than the "beautiful" move...).
>
>Regards,
>M.Scheidl



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.