Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: draw-score behaviour is in DeepFritz T28 too...

Author: Mogens Larsen

Date: 00:36:52 12/12/00

Go up one level in this thread


On December 11, 2000 at 20:27:43, Christophe Theron wrote:

>Excuse me Mogens, but I believe I have some knowledge about what's going on in a
>chess program (mine for example), and I can assure you that programs DO plan.

Well, I'm sorry to say that programs don't plan. You can guide it towards a
certain type of position that it knows how to handle by changing certain
parameters, eg. putting pressure on the opponents king until it feels
confidentabout the attack.

But it's clear in the cases where the attack fails that the program won't have a
clue what to do next. It'll most likely attack even more and end up losing. The
same bewilderment will happen in quiet or very closed positions. There will be
numerous examples of simple planning that a program won't be able to solve.

That would include your program as well, so the answer is still no. Ordinary
calculation based on increased values (bigger carrot) and dedicated knowledge
(longer stick) isn't planning.

>But I can assure you that if humans are planning, then programs (some of them)
>do it as well.

Some programs in some cases will appear to be planning. Not a general trend by
far.

Mogens.

PS: And in case you say "Yes, they do" again, I say "No, they don't" :o).



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.