Author: Christophe Theron
Date: 13:53:45 10/19/00
Go up one level in this thread
On October 19, 2000 at 16:31:34, Ratko V Tomic wrote:
>> I don't need reverse engineering to understand what other programmers do.
>> It's clear enough when you play with a program for some time.
>
>I didn't mean that you necessarily disassembled their code and extracted their
>algorithms/ideas out of there. That's one way to reverse engineer some software,
>but not the only or necessarly the most effective. If you're doing similar
>things as someone else, and have developed a good intuition of the domain,
>observing outputs for carefully selected small sample of inputs often works
>better.
>
>But, you did mention on few occasions within the last year that you do know how
>Lang and Kittinger
I mentionned Lang, but not Kittinger.
I have no program from Kittinger, except an old version of MyChess, back from
the eighties.
I did not know Kittinger's program had long lines, similar to what Lang's
programs do.
> do their magic long lines in specific types of positions.
>Although you were somewhat dismissive about that technique (which you didn't
>actually describe but just stated that you knew how it is done), it seemed to me
>that the dismissive attitude decreased over time. Coupled with the arrival of
>the GT phenomenon (which reminds me most of arrival of Constelation), it occured
>to me that you may have experimented a bit further with what you have
>reconstructed as a simulator of the Lang's and Kittingers magic far-sight
>algorithm (and which actually may have been entirely your invention and it
>merely mimicked the behavior of their programs). Then a new possibility suddenly
>opened up for refining it in a substantial way, maybe a much more accurate
>algorithm to filter out the bad lines or some such, that the other two may have
>missed (or the opportunity may have not existed at all in their form of the
>algorithm). And then this new code you were toying with probably surprised you
>the most.
I still think that Genius long lines have almost nothing to do with its
strength. It's spectacular, but that's all.
My opinion has not changed, and actually the novelty in Gambit Tiger is not in
the search but in the evaluation.
I'm not saying that Tiger's search has not evolved. I have worked several months
in the past year on improving the search algorithms, and it is easy to see that
Chess Tiger 13.0 and Gambit Tiger 1.0 have much better searches than Chess Tiger
12.0.
But Chess Tiger 13.0 and Gambit Tiger 1.0 do not differ significantly in their
search algorithms. They differ mainly in the way they evaluate positions.
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.