Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder 7 didnt find it at depth 17/50

Author: Uri Blass

Date: 05:30:04 01/03/03

Go up one level in this thread


On January 03, 2003 at 07:37:12, Drexel,Michael wrote:

>On January 03, 2003 at 01:12:42, Uri Blass wrote:
>
>>On January 02, 2003 at 18:36:35, Drexel,Michael wrote:
>>
>>>On January 02, 2003 at 17:52:50, Uri Blass wrote:
>>>
>>>>On January 02, 2003 at 16:39:27, Drexel,Michael wrote:
>>>>
>>>>>On January 02, 2003 at 16:16:05, Dieter Buerssner wrote:
>>>>>
>>>>>>On January 02, 2003 at 15:53:44, Drexel,Michael wrote:
>>>>>>
>>>>>>>Hello Dieter,
>>>>>>[...]
>>>>>>>at which depth do you think it would find the mate? 30/30?
>>>>>>
>>>>>>Sorry, I am the wrong person to answer this question. How should I know?
>>>>>>
>>>>>>>this 'behaviour' has obviously something to do with its superior playing
>>>>>>>strength.
>>>>>>
>>>>>>?
>>>>>>
>>>>>>Regards,
>>>>>>Dieter
>>>>>
>>>>>just wanted to say that this kind of tests (with unreal or unlikely situations)
>>>>>are not important, regarding playing strength.
>>>>>
>>>>>Michael
>>>>
>>>>I disagree.
>>>>
>>>>I believe that knowing to solve the position is an advantage from playing
>>>>strength point of view.
>>>>
>>>>It may be unimportant if white had another win but it is not the case here.
>>>>
>>>
>>>In this special case it is an advantage but how often would such a type of
>>>position occur in the middle game? Shredder 7 and Fritz 8 (Junior and Chesstiger
>>>15 results would be of interest too) are not
>>>able to find the best move in reasonable time.
>>>Im not a chessprogrammer but it seems clear to me that 'comercials' are able
>>>to get deeper into the search, by excluding more variations then most freeware
>>>programs do. hence they often fail to find mates in such positions.
>>
>>I reject this explanation.
>>Most freeware also do not find it.
>
>I should have said most of the 'strongest' freeware programs.
>
>>
>>I also use pruning but for me pruning is not the main problem because even if
>>you give movei the position after Qxg6+ Kxg6 movei cannot find that white is
>>better in a reasonable time.
>>
>>pruning is important but good pruning rules should say not prune Qxg6 in that
>>position.
>>
>>I can give logical reasons.
>>1)white has inferior position without Qxg6+.
>>2)Qxg6+ lead to some attack for white.
>>
>>conclusion
>>Qxg6+ should be extended becuse it is the only chance.
>
>Stefan Mayer-Kahlens program (engine) do not extend Qxg6+ as far as I can
>understand.
>would Shredder 7 otherwise evaluate 1.Qxg6+ only 2 seconds (1.h5 about 10
>minutes) at depth 17?
>I simply cant understand why. Doesnt Stefan know what he is doing?.
>I dont think so.
>If I would write a chessprogram, the first thing I would probably do (after the
>basics) is:
>try to make some extensions in case 'King is marching over the board in a middle
>game situation'. my program would evaluate 1.Qxg6+ very very long in this
>position.
>
>so my conclusion is: with a good opening book in mind (and shredder has good
>tournament book)it is not that important to have such extensions.

I agree that it is not very important to have extensions when the king runs to
the middle of the board but I still think that it is possible to earn some
rating points from it.

I do not think that it is something that has to be with opening books.
I believe that in most cases the search depth is simply big enough to see cases
when a sacrifice leads to win.

In most cases it is better to prune moves that lose a lot of material but this
position is an exception(one of the reasons is the fact that white has bad
position without Qxg6+ so it is better to extend the move that seems to be a
hope to save the game).

Yace find it thanks to extensions that are not check extensions and
I do not have the relevant extensions.

I believe that I may earn also from other extensions and I am going to implement
them in the next version.

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.