Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question about Fritz's draw evaluations

Author: Uri Blass

Date: 15:38:38 12/10/00

Go up one level in this thread


On December 10, 2000 at 18:13:02, Thorsten Czub wrote:

>On December 10, 2000 at 18:00:30, Enrique Irazoqui wrote:
>
>>On December 10, 2000 at 17:33:14, Thorsten Czub wrote:
>>
>>[crap snipped)
>>
>>>that of course the reason there is no fritz7 on the market.
>
>that what you call crap and what you snip is in fact data.
>you can snip it, but you cannot desinform the people about it.
>
>>There is.
>
>no - there isn't.
>
>>It's out.
>
>too late.
>
>>No.
>
>in the position i quoted: it would. you would see this,
>if you would not have snipped it.
>
>impressive enrique. but the games are fast-time controls.
>
>if you would have played longer time control it would not be
>
>6 wins 6 losses 8 draws
>for DF but
>
>6 wins 12 losses 8 draws against DF.
>
>try with 40/120.
>
>then you can snip it again. and even then: the draw-score behaviour is still in.

My guess is that the result will be similiar at longer time control.


I know that Gambittiger has also significant evaluation problems.

I saw tournament time control game when GambitTiger could not win against
Gandalf inspite of evaluation of 6 pawns advantage.

The problem is that GambitTiger did not trade to an endgame because it was too
optimistic.

Here is the position when Gambit evaluated +6.12

[D]r6k/7p/p2qbp2/4pp1P/r7/P1pB1PP1/2N1R3/2K1R1Q1 b - - 0 1

GambitTiger played 30...Qxd3 (6.12 at depth 11) and after 31.Qb6 could not find
31...Qb5 and prefered 31...Qd7(4.34 at depth 14) and later drew the game.


GambitTiger is also a preprocessor and it is another evaluation problem.

Uri



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.