Author: Christophe Theron
Date: 18:10:32 12/31/97
Go up one level in this thread
On December 31, 1997 at 08:40:09, Chris Whittington wrote:
>Play style (Tal or Tarrasch) has to be evaluation based. Search is only
>about what can be found, at what depth, and in what time; not about
>style of play.
>
>Chris Whittington
Yes, it is better if it is evaluation based.
But Thorsten said Genius algorithm put some bias in the way it plays.
I think that if you do the asymetric thing he describes you will get the
Genius style of play, in a certain sense.
You won't see your program attacking weak pawns if the evaluation has no
clue about that. But if you prune heavily on your side and use a
"satisfaction" method (prune, but keep on trying others moves if you are
not satisfied with your score), your program will behave the "do nothing
but do it well" way, and will defend very well if it is attacked.
And it will often miss good active lines, if they don't begin with an
obviously good or threatening move. But it will be somehow faster in
compensation...
The next time I'll see a program with this style of play, I'll first try
to see if it is asymetry!
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.