Computer Chess Club Archives


Search

Terms

Messages

Subject: How, and how well do chess engines attack [& defend vs attacks]?

Author: Robert Henry Durrett

Date: 06:48:10 08/27/98


In the threads "CSTal Win95 needs only 62 seconds to reject Qxg7 , so what ? "
and "So What?", there is a discussion related to this topic.

In the past, there have been many comments in the chess literature to the effect
that it is easy to win against a chess engine simply by starting an attack.  The
engine, supposedly, cannot see far enough ahead, so gets whipped.

One would suppose, based on the above assumption, that chess engines would also
be poor attackers, also for the same reason, that they could not see far enough
ahead.

But, in the above threads, it appears that the engines being discussed [Crafty,
etc.] do find the positional preconditions for attack, [&, presumably, also
sense danger in the same way] and make moves based on the programmer's desire
for his engine to find an attack.

Presumably, if an engine senses that the preconditions for attack are present,
it's programming might cause it to look deeper into attack lines.

Similarly, if an engine senses that some but not all of the conditions for
attack are present, then it could be programmed to search for ways to cause the
remaining preconditions to become met.  [This is positional chess!!!]

For defense, the above two paragraphs apply except that preconditions for an
enemy attack are detected and the engine might be programmed to search for ways
to eliminate the preconditions [also positional chess!!!].

The above merely speculates on what engines might do.

So, the question is:   To what extent do existing chess engines do the above?
Are modern chess engines quite good at attack and defense?  If so, what are the
real reasons for that?

Just curious.



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.