Author: Dann Corbit
Date: 18:52:11 11/29/01
Go up one level in this thread
On November 25, 2001 at 04:49:50, Mogens Larsen wrote:
[snip]
>I referred to your method of avoiding modifcations, not CMFun in particular.
>However, allowing mulitiple versions of some engines for subjective reasons
>yields an unfair and skewed list. That is a fact. Whether it is uninteresting or
>not is up to the reader.
Just to clarify for people who are not keen on statistical stuff, imagine the
following (exaggerated) scenario.
We have two engines, x and y, of equal ability.
We enter x one million times, tweaking some parameter each time. The overall
strength changes a few ELO +/- each time, but the average is about equal.
We enter y one time, with all default parameters.
What are the odds of an 'x' clone winning? 999999/1000000 = 99.9999%
What are the odds of y winning? 0.0001% {y, equal in ability, has no real chance
to win}
The greater number of times a cloned program enters, the higher the probability
that one of the clones will win the contest.
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.