Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Short German resumee about Shredder 6...

Author: Jeroen van Dorp

Date: 07:11:54 01/17/02

Go up one level in this thread


Thanks for the review; it's very interesting.

I noticed that the opening book of S6 is enhanced in comparison to the S5 book.
Maybe the wide variety of the book options - even in tournament setting - could
be responsible for it. A serious tree weeding could be an option to improve
play, or - as you state- a better learning function. Better to have a lean and
mean book than a big book with all possible less moves, which have to be weeded
out by a learning function. And if that function doesn't work well...


I have examples of endgame play in which Shredder is clearly better than other
engines. So the claim that it can be stronger in the endgame can be uphold
AFAIC. One example is where Shredder finds a winning sequence which other
engines don't find. That could be better tactical calculation as well. I don't
know, but I guess so, and that's why the subject of King safety popped up in my
mind after reading it in your review - in conjunction with your remark about the
results against Fritz 7.

King safety very often becomes an issue in the late middle game and the endgame,
when pawn formations are being attacked and eventually distorted.
Shredder is said to play solid. It might well mean a somewhat *careful*
approach, but doesn't necesarily mean "good understaning of king safety".
When I play against Shredder, I can uphold an even evaluation quite long,
because it doesn't attack so agressively like other chess engines do. Evenually
I perish, but it gives me an idea that with "just shuffling" I could hold
against Shredder.So not necessarily "good understanding of positional play" but
just "solid play".

But I think the new kid on the block Fritz 7 is different in this respect and
has done something (for the time being :))to the hierarchy of the strong chess
engines here.
I'm very impressed with some playing features of the new Fritz 7a. Mind you -
I'm not saying that Fritz 7a is "the ultimate and strongest ever" or claims that
pop up from time to time about newcomers, but at some aspects Fritz 7(a) is a
lot stronger than Fritz 6.
The endgame play has indeed improved, but most of all understanding of king
safety by Fritz7.


There's a weird story to that. I recently played a game against someone online,
who is well known to use Fritz6 for playing his games. It's not allowed, I think
it's annoying, and I ususally try to avoid him in a tournament I join, but I
decided not to care, because at correspondence chess you can take your time to
outwit a chess program, based on knowledge of it's weaknesses. I have played him
before and lost all of course, until now I decided to play against Fritz 6,
instead of just looking "at the best moves". A bit of anticomputer strategy, if
you like.
Surprisingly I won the game, mainly because a sloppy king safety problem of my
opponent (or better: Fritz6) that popped up, and which I could use.

After the game I was very interested in the analysis of Fritz 7a of the game,
and found out that Fritz7a differed substantially from the opinion of Fritz6,
and agreed with some manoeuvers I made in the game to undermine my opponent's
king safety. His Fritz6 walked in with eyes wide open, while Fritz7 analysed it
much more correctly in my opinion, and agreed almost everywhere with me.

After I played the game I recalled a remark by Kramnik about Fritz 7. He told
ChessBase that he was impressed because of improved understanding of king safety
and (IIRC) of pawn structures.
When I read that my first thought was: he means Fritz6 was lousy in that
respect, like most programs, and Fritz 7 is less bad. But OTOH if a guy like
Kramnik tells that - and with my modest CC experience as well- he might be right
with his assessment that Fritz 7 has much better understanding about king safety
than other programs.

Because S6 performs well against all the others, it could well be that it's not
a specific Shredder problem, but a *positive* Fritz 7 issue.

Just a thought.

Good luck with the research.

J.



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.