[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: popularity contest and queso

On Sun, Mar 11, 2001 at 01:58:37PM -0500, Jacob Kuntz wrote:

> > Any identifying information, such as your IP address, hostname, etc, is
> > deleted as the first step during processing.  There are so many ways to
> > cheat on popcon anyway that it would pointless to try to eliminate a subset
> > of them.  We just assume that it would be silly to cheat, because it's not
> > worth anything to anyone.
> heh. when i discovered that joe wasn't on cd1 of potato, i seriously
> considered cheating. :)

I know that exact feeling (I vote for joe regularly :)).  I think what we
really need is some adjustments to the scripts that interpret the results.

The simplest change would be to adjust the scripts so that editors starting
with "jo" get a higher weight, but in order to be less obvious we need to
invent something that sounds plausible but still does what we want.

We could give smaller packages a higher preference even if they get fewer
votes -- that is, we want to maximize the CD score, so really what we're
interested in is the popularity of a package per kilobyte, not just the

For example, to use some COMPLETELY RANDOM numbers, 40% of people might use
Gnome, but gnome is really big and you might have to take out KDE (for
example) to make it all fit.  That might make 30% of people unhappy, if 30%
of people use KDE.  On the other hand, if only 20% of people use joe, we can
immediately make those 20% of people happier by including it, without
needing to remove KDE, so the net benefit is actually greater than the
benefit of including gnome.

Somebody who knows more about statistics ought to write this one, though :)

Have fun,


Reply to: