Author: Christophe Theron
Date: 16:28:28 10/08/01
Go up one level in this thread
On October 08, 2001 at 15:19:39, Francesco Di Tolla wrote:
>>>>This goes against the already old urban legend saying that "more knowledged
>>>>programs" (programs with slower evaluations) will be superior on faster
>>>>computers (which is clearly denied year after year).
>>>
>>>Then why not using a pure "bean counter"? This should beat any program just
>>>because of the extra fraction of a ply it can reach.
>>>
>>>regards
>>>Franz
>>
>>
>>
>>I don't understand how you can come to this conclusion using what I said...
>>
>>I'm not saying that knowledge is useless or that more knowledge cannot beat more
>>depth, I'm just saying that I believe that there is also dimishing returns from
>>improved knowledge at deeper ply depths.
>
>It's called "reductio ad absurdum": pushing your idea to the extreme side an
>evaluation function wich just barely counts pieces could ouperform an evaluation
>function that spends time to "look" at the position, thanks to extra fraction of
>a ply or some plys.
>
>Something nobody would think.
>
>This does not imply that the better the evaluation function (at any CPU time
>cost) ALWAYS the better results.
>
>There is a balance at any speed between what "is worth" spending in evaluating
>and what you "get" more from extra plys.
>
>The more CPU time you have the more it seems worth, to me, to inwest in
>evaluation.
>
>Seems so linear.
>
>regards
>Franz
Thanks for explaining this to me, Franz, but I think I already have a pretty
good idea of this concept of compromising between evaluation accuracy and speed.
However, this still has nothing to do with what I said earlier...
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.