Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder8 - Hiarcs9 tc05: 25-35, +16 =18 -26

Author: George Tsavdaris

Date: 10:11:56 02/16/04

Go up one level in this thread


On February 16, 2004 at 10:04:12, Thorsten Czub wrote:

>On February 16, 2004 at 08:22:56, Chesster Fritz wrote:
>
>>Forget Fritz 8....Tweak Shredder 8 to combat your new settings...I'm certain you
>>could find a setting to wack the crap out of HIARCS 9!
>
>first i want to see if the new good style (tc05) is not losing against fritz8.
>if it is good against both, it is ok for me.
>
>I don't think that i am capable to tune shredder8. I guess Shredder is often
>optimized very perfectly by stefan and his team. they are doing a good job.
>thats ONE reason why they win so often on tournaments and championships.
>
>But IMO there is something WRONG with the program. a bug, a feature. or maybe
>this is due to new ideas of stefan. it seems shredder8 plays different and has
>strange effects people fail to reproduce. you can read about that in almost all
>forums.

 I don't know what bug you are talking about, but Shredder 8 makes the most
powerful analysis i have ever seen by a program. It understands ~35% of the
positions MUCH better and much quicker from Chess Tiger 15.0 and from others.
Only Deep Fritz's 8 analysis is comparable with Shredder's 8. And as we have
seen until now, no program can stand head to head with Shredder 8.

 One small bug i noticed is the crappy variations is giving for some endgame
positions, especially when it finds a mate. One other annoying feature is the
high score it gives for some positions where one player has an advantage, just
like Shredder 7/7.04. In many positions where Chess Tiger 15.0, Fritz 8, Ruffian
2.0 and Junior 8 give a +0.55 to +0.85, Shredder 8 gives a +2.10 just like it
happened with it's predecessors. But if a big number of these evaluations of
Shredder 7.04/7 is just untrue, the most of the high score evaluations of
Shredder 8 are representing the truth about the position. And there is no reason
if a position is better for one side, to give it +0.80 instead of +2.20 or +3.80
or more. So it will be good if all programmers agreed to a rule when a score of
0.00 means a draw and a score of 10.00 a 100% win
and the scores between them, to correspond in a linear way to the probability of
winning. Anyway some -not so rare- times even these high evaluations of Shredder
8 is wrong. One almost general rule is when the evaluation goes up to +3.2 then
Shredder 8 is right. Of course there are some endgame positions that this rule
is wrong. But in general Shredder 8 is a brilliant piece of software and of
course all the people have to buy it.



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.