Author: Christophe Theron
Date: 15:10:18 10/06/98
Go up one level in this thread
On October 06, 1998 at 16:56:02, Robert Hyatt wrote:
>On October 06, 1998 at 13:16:50, Christophe Theron wrote:
>
>>On October 05, 1998 at 17:10:07, Cristian Zaslo wrote:
>>
>>>Hi everybody!
>>>One day I made a comparison between some ZX Spectrum chess programs (Colossus
>>>4.0, Superchess 3.5) and some medium - level PC programs (K-Chess, Now, Gnu
>>>Chess) and found that these ZX programs use a pretty strange searching algorithm
>>>(for me) , as follows :
>>> 1. It seems to me that entire searching tree is built round the PV which is
>>>often longer than current depth and it may contain positional moves (no
>>>captures or checks) everywhere including in the Q-search part.
>>> 2. The other branches are cut-of very quickly, sometimes just from the root,
>>>and so, can overlook. good moves.
>>>
>>>I now that many programs (mine too) grow branches this way:
>>> (Depth + Extension) + Q-Search
>>>I d like to now a little bit more about the search-engine (cut-off techniques)
>>>used by these old fashioned programs and if this approach still (no)works on
>>>any PC programs.
>>>
>>>Much obliged to you,
>>>Cristian Zaslo
>>
>>Maybe we can try to answer this question.
>>
>>But first, can you please tell us who is the author of these programs? Or which
>>company if the author's name is not mentionned.
>>
>>If it is Richard Lang, you have something like an old version of what is today
>>called "Genius". To my best knowledge, it is the only program that shows this
>>kind of behaviour (very deep quiet moves in the PV).
>>
>>This approach is known to be very good on very slow processors, but much less
>>effective of today's fast computers.
>>
>>
>> Christophe
>
>
>
>actually, I don't know of any program that can "roll up" genius tactically
>today, still. It is *very* strong. Just has a suspect opening book by today's
>standards... but definitely not a pushover. You should try it on a PII/450...
Oh yes... I think I didn't express my idea correctly in my previous message.
You are right, Bob, and I agree with you. In fact I have several times expressed
here on CCC what I think about Lang's program: this program has, in my opinion,
the best search algorithm ever written for chess. Nothing less.
What I really meant is that it used to be an overwhelming advantage on slow
processors, but the advantage toady is smaller. Positional understanding is now
the real issue (once your search is good enough), and Genius simply stands in
the top pack. Not 1 mile above the rest as it used to be.
However Genius search is still unequaled in the endgame. Why??? I really don't
know. Playing correctly an endgame against Genius is always a very very hard
task. I would not advice all programmers to try their program in an endgame
against Genius. Some of them might well give up chess programming after that! :)
Christophe
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.