Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Play-the-game test Introduction/FAQ

Author: Thorsten Czub

Date: 03:34:21 12/14/97

Go up one level in this thread


>So you admitted that "intelligent" programs can be fast searchers.
>Why not admit that fast searchers like Fritz5 or Nimzo98 can play
>"intelligent" - whatever this means...?

If they WOULD i would admit it. But when I have to play games with
Fritz5 (and I have to do it manually !!!) it is very often a torture !
If I could it blind, without seeing the game, it would be better for my
health-condition. Therefore I am really looking forward for an
autoplayer for Fritz5...

>
>I laughed when I saw Fritz2 play great combinations and positional
>blunders in the next move. Later since Fritz4 I began to see some
>"intelligent" positional play from the program here and there, and now
>Fritz5 makes hardly less "intelligent" proposals in GM game anylysis
>than other top programs - if you take the average rate of good move
>proposals in well annotated Grandmaster games. Maybe it's similar with
>Nimzo98?

I remember Fritz5 play a beautiful combination in Den Haag against
Karpov.
Karsten was the operator. When Fritz played this combination any of us
was breathless. Only Karpov looked astonished to the board. And than you
could see a smile on his face. YES, the combination worked, but the
END-POSITION WAS A TECHNICALLY LOST ENDGAME.
Thats the problem with Fritz. It finds something using the SEARCH but
has no idea about HOW to evaluate the END-POSITION.

>
>So the old pattern in my eyes has already have become increasingly
>obsolete since two or three years and will probably become even more
>untrue in the future.

I cannot subscribe to the above.

>It's not the question which way you use to do something proper as long
>as you succeed to do it proper. You can like one way more than the
>other. But you have to have respect good results and acknowledge

Good results. BUT WHERE ???
Only on Moritz' machine ??
In sweden ?
Where ?
I don't believe the results on the ChessBase site. Sorry.
My results correlate brilliant with Paris Fritz-results but NOT with the
advertising-results of fritz5.


>improvements - this can be no matter of some principal dualistic
>ideology. And by the way, *any* of the "intelligent" programs will not
>have an easy advantage in a long match against Fritz5 with big hash. Of
>course you know that from testgames as well as I do.

As long as I have NO FRITZ5 autoplayer I have no idea about this !
From the small samples I have played I would say:
NO ! I don'T believe this fairytale.
If Fritz5 is so strong, why don't they sell it with autoplayer that the
ssdf-guys can easily produce some results ? They are feared to do this.
Because WHEN the opponents and ssdf-guys have Fritz5 autoplayer, anybody
would SEEEEEEEE how strong Fritz is. :-)


>If not, I guess(!) the SSDF-list may cause some surprise for you some
>day.

Show me the games, show me the data and I will comment.
As long as this is not true, I will not be surprised.

>Fritz5 plays some kinds of positions better than others, but this is
>still nearly as much true for any other top program.

Right.

>None of them has a
>totally balanced overall performance, each has weaknesses and amoung
>them positional ones...

Right.


>All the best from Dirk
Thanks Dirk.

>
>P.S.
>By the way, didn't you notice how much Fritz really improved a lot if
>you take a look at the P a w n  s t r  u c t u r e s? :-)))))
>(Sorry folks, this is an allusion only Thorsten will understand)

ahem - yes . Especially the pawn-structure of Fritz5 is very impressive
Dirk. If you tell it, it must be true - smile ! :-)



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.