Author: Christophe Theron
Date: 10:21:00 10/21/99
Go up one level in this thread
On October 21, 1999 at 10:33:16, Peter Pilarinos wrote:
>My personal opinion is that it would be EXTREMELY difficult to make further
>progress based on the present Genius program structure. I remember Richard Lang
>saying this in one interview that I read about him some years back. His programs
>are very unique in their structure. It was this type of programming tequnique
>that allowed the Genius programs to be so poweful on relatively slow computers.
>Now that the computers are so much quicker, the advantages of this type of
>search structure are somwhat dim. I believe that if Mr. Lang wanted to make
>futher noticable progress, he would have to re-write his program from scratch,
>using different, more contemporary methods. He is obviously a very talented man,
>and I am sure he could do this if he wished.
>
>Pete P
I'm sure you are going to think that the following is very arrogant, but I
really think what he said is BULLSHIT.
Lang's program are not unique in their structure. He wants you to believe so,
but it's not true. He is using alpha-beta, hash tables, pruning techniques,
static piece square tables, dynamic part in the evaluation and so on.
What Genius misses is some modern pruning techniques that work better at higher
depths. Because Genius is killed by search depth, which is ironic as he was
killing everybody by search depth several years ago. probably also relying less
on static evaluation made at the root would be needed.
The truth is that he has probably lost the will to work on Genius. What he has
does not need to be rewritten. He should keep almost everything, because he has
powerful algorithms. He just has to add several modern techniques, which are BTW
publicly known.
In this case however, Genius would become a program similar to many others, with
just a little bit of originality.
But who am I to pretend he is wrong... :)
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.