Computer Chess Club Archives


Search

Terms

Messages

Subject: I'm an A player--I already knew Fritz (6, 7) was not always positional

Author: Stephen A. Boak

Date: 20:05:19 10/08/02


I'm an A player (1800-1900 USCF).  Although I acknowledge I could not beat Fritz
6 or 7 (what I have used a lot), perhaps never even draw, I have nevertheless
remarked many times to myself and my chess friends that the suggested best moves
of Fritz (per overnight game analysis--my favorite use of the program) looked
odd.

They often were anti-positional, something we humans would look at with a very
strange feeling.

It's just that it throws in some real klinkers--on a regular basis.

Admittedly the program is a great searcher, and great in tactics.  Many of the
moves it makes look fine (or even great) positionally.  I learn a lot from
examining the Fritz analysis lines.

I always ask--why does this suggested move work best?  What is so good about it?
 I can often learn some tactical ideas.  I can often learn some positional ideas
from watching Fritz's recommendations.  All too often, however, I have to
say--that moves looks goofy, like a computer move.  Fritz is showing its colors
(that it is not human).

Sometimes two moves (in multi-variation analysis) are scored nearly the same.
Sometimes the anti-positional move scores a bit higher than the 'normal' move we
humans would applaud.  This always looks odd and makes me wonder.

Based on my use of Fritz 7, it seems to jump around (up/down) in the eval
scores, as it analyzes successive White & Black played moves.  Seemingly more
than its predecessor, Fritz 6.  To me, I say the scoring is not 'as smooth' as
Fritz 6.  Anyone else detect this difference?

When I use the program to analyze my games, looking for suggested moves that
improve my position by perhaps .2 to .3 pawns (I deem these positional
improvements, since they do not immediately or noticably win any material), I am
disheartened when the analysis of the next move suddenly drops!

Sure there are always horizon effects, but I expect that analysis that goes to
quescence will produce similar scores, move after move, except in the rare event
when the win of material is finally seen (not something that happens on each &
every move).

Are these signs of some null move problems?

Often I use Fritz to analyze & show the top several moves.  After I play another
move (perhaps as actually played in the game), not in the top suggested Fritz
move list, the score may jump and be equal or higher than the prior suggested
top 3-5 moves!  It seems that Fritz pruned out the move as a candidate, but once
it is forced on the board it indeed sees that is it a fine move (as good as what
Fritz recommended).

This happens very often, in my experiences.

Any comments by other users of Fritz 6 or 7?

Thanks in advance,
--Steve










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.