Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Junior's long lines: a witness speaks

Author: Thorsten Czub

Date: 04:12:31 12/25/97

Go up one level in this thread


On December 25, 1997 at 03:48:05, Christophe Theron wrote:
>This board is interesting. Things of interest happens here, and real
>computer chess stuffs are discussed.

I completely agree here !

>So I just go back to computer chess and, yes, I saw Junior in Paris. One
>day, it was doing a kind of self-play. I suppose Amir was testing the K6
>computer. The rate of play was roughly 1s/move. I was very surprised to
>see very long lines on the screen and stood there for a while. In only
>1s Junior was able to compute 10 or 12 plies lines. And the last moves
>usually were not captures.

Exactly.. And this behaviour was in the program since 1995 Paderborn
championship. In the game of 1995 CSTal-Junior I operated CSTal and
watched the screen a long time. The only difference between Junior 1995
and junior 1997 was, junior 1995 was NOT always overprotecting, had
problems with pins and king-attacks and, if we sum up it: was less
intelligent concerning chess-knowledge.

So this behaviour with enormous long lines and high NPS was in the
program for years.

It is - from my point of view - a different approach. I guess any of you
programmers should study junior and try to find out what they have done,
because they do something different !!

>I don't like to bring unverified news, so maybe Amir could confirm what
>I say. I often read postings from him on CCC, so he's likely to read
>this tomorrow!

IF amir tells us the complete truth ! In my talks with him in Paris I
had the feeling he spreads FOG arround his program....

>But Junior's algorithm
>is smarter. The only program I know with a similar behaviour is
>Genius...
>

Right. The very long main-lines remembers very much on the way old
Genius versions produced 8 plies main-line with only 2 plies of their
slow component.

We did some experiments in Paris on the hotel room to find out about
Genius
and Vincent told something about 1/3 plies and all this stuff.
I am sure this secret will be cracked ....



>    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.