Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Pondering and diminishing return ?

Author: Mogens Larsen

Date: 07:36:21 05/25/00

Go up one level in this thread


On May 25, 2000 at 10:24:12, pete wrote:

>From what I have observed correct pondering tends to be above 50 % percent in
>computer-computer matches most of the time . I have explained this to myself
>that all the programs despite their obvious differences do very similar things
>and so often come to similar results .
>
>I haven't seen a clear trend yet that this number goes down in Blitz games . Has
>there been any published analysis of this question somewhere ?

No, I don't think so. I just assumed that the ability to ponder the correct
moves differs if the program has 10 seconds or 2 minutes. That might be
incorrect, but I really don't think so. Because there's two uncertainties
involved; program A's ability to find a reasonable move with less time and
program B's ability to calculate that same move with less time. With more time
the probability of a match increases, because the difference in approach to
search evens out over time.

>In fact it seems to me that when there is _no_ diminishing return or only a very
>tiny one a logical conclusion is that pondering will be of similar usefulness at
>any time control .

No, that's not a logical conclusion IMO. Would pondering a wrong move be more or
less hurtful to the search itself if the programs has 10 seconds or 2 minutes?

Best wishes...
Mogens



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.