Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Stephen Ham et al ... correspondence chess and computers

Author: Bob Durrett

Date: 08:32:31 04/20/04

Go up one level in this thread


On April 19, 2004 at 16:17:37, Mark Ryan wrote:

>Hi Stephen and others:
>
>Which chess engines are most likely to be used by correspondence chess players?
>Which engines are used to check for tactical shots, which for positional play,
>which for the endgame, which for opening theory?

There was a time when I played a lot of correspondence chess but have given that
up.  Back then, in international tournaments, we were told that we in the USA
should use computers "because the Europeans all did it and there is no point in
handicapping oneself."  [When in Rome, do as the Romans do.]  Chess engines were
not nearly as good then as they are today and I found them to be useful only for
tactical blunder-checking, used after selection of the move to play.

Today, the engines are much better and there are more of them.

If I were to get back into correspondence chess [unlikely], I would probably
find new ways to use chess engines.  One feature of chess engines, which becomes
evident during post-mortem analyses, is that certain positions get an incorrect
evaluation from the engine but if subsequent moves are entered into the chess
computer [based on human insight and planning] it sometimes happens that the
engines begin to see things differently.  [This is due to the horizon effect.]

The logical implication of the above observation is that engines can help to
check out human-generated plans.

Chess strategy books written by GMs [and the top players of the past] repeatedly
[and consistently] say that one should learn how to evaluate positions to: (1)
determine what plans are available to each side, (2) assess [by analysis]the
liklihood that the plan could be realized, and (3) assess the benefits [and
negatives] to be derived from accomplishment of the plan's objectives.  These
three things need to be done for both sides.  Essentially, one obtains a feel
for the prospects for each side and the findings serve as a guide as to how to
proceed.  Note that for some positions, no clearly discernable plan can be found
and for some other positions there may be two or more seemingly reasonable plans
which could be pursued, consistent with the demands of the position.

If a correspondence player wishes to just let the engine select the moves for
him, he might as well give up chess entirely.  On the other hand, there may be
ways to improve one's general advancement in understanding chess if the engine
is used properly to help one understand the game better.  This might be achieved
in correspondence chess by letting the engine provide feedback during the
analysis process in much the same way as is done during post-mortem analyses.

For example, one could formulate a plan and try to analyze without the aid of
the engine to get the best insights one could on one's own.  Then the lines
could be checked with an engine and the inevitable errors in thinking
identified.  Then the human would "go back to the drawing board" again without
the aid of the engine and try to do a better job the second time around.
Several iterations of this process might be needed.

The danger is that one might get into the bad habit of doing superficial
analyses [without the engine] with the knowledge that the engine would likely
find the errors.  This form of mental laziness would serve the human poorly in
the long run.

Shredder, Fritz, and Junior would make good partners.  Be sure to use tablebases
and consult good opening books.

Bob D.




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.