Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: More bugs in Shredder9

Author: Komputer Korner

Date: 23:39:26 05/13/05

Go up one level in this thread


On May 14, 2005 at 01:26:50, Dann Corbit wrote:

>On May 13, 2005 at 22:39:23, Komputer Korner wrote:
>
>>On May 13, 2005 at 13:39:00, Dann Corbit wrote:
>>
>>>On May 12, 2005 at 21:46:14, Komputer Korner wrote:
>>>
>>>>If you hang Shredder9 any time either by making the mistake of pressing the X
>>>>button in the top right korner or else by pressing CTRL-Alt-Delete when
>>>>Shredder9 is calculating, Shredder9 refuses to shut down. It does not behave
>>>>like a well mannered Windows program is supposed to. In the above cases, you
>>>>have to reboot the computer because Shedder9 will keep creating clones of itself
>>>>to prevent it from being shut down abnormally.
>>>
>>>Chessbase?  UCI?
>>>I have never had any problem.  I use UCI exclusively.  Sometimes under the
>>>Shredder classic interface and sometimes under other interfaces.
>>>
>>>>Another small bug I found is that
>>>>it often will report a  mate in 150 when it has a winning advantage. I suppose
>>>>the programmer's  sense of humour is at work here, but these bugs should be
>>>>corrected.
>>>
>>>That is not a defect.  In fact, if it does checkmate in 150 moves or less, then
>>>it was surely correct.  Chess programs do not report minimum distance to mate
>>>like a chess solver.  They report "Mate in AT MOST <n>".
>>>
>>>Now, if later on Shredder says "Mate in 300" then there might be a bug because
>>>it should not lose a close mate and switch to a farther one.
>>>
>>You meant "Chess programs DO NOT report maximum distance to mate"
>
>Maximum distance to mate is about 6000 fullmoves.  It would not be a very
>interesting announcement.
>
>>However they
>>do. almost all of the best ones do including Fritz,Shredder..etc
>
>You are wrong.  They give their best estimate, which often improves (and that is
>for every single program that is not a mate solver).  Each and every
>professional program will announce a mate in x and later a mate in x-k and then
>later in x-k-l etc.
>
>>They don't bother with a nonsensical mate in 150 forecast.
>
>Several programs do this.  It simply means "I am going to win" and it is a very
>smart thing to do.  You will find that it is not an uncommon announcement among
>excellent programs.


Well I consider that it is silly. As for the exact mate anouncements, they all
seem accurate enough. I never have questioned these because they all seem
accurate. They don't have to be mate solver programs to be accurate almost all
of the time  in mate announcements if the evaluation hits the roof on a certain
ply.



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.