X-Message-Number: 26179
Date: Sat, 14 May 2005 03:49:12 -0400
From: "Kevin Q. Brown" <>
Subject: Re: Cryonet Rating Results

In the previous Cryonet message Flavonoid asks several questions
of the CryoNet "moderator" concerning the rating system.  I don't
really do moderation, but I do administrate CryoNet and program
the software running it, which should be close enough.

Concerning the algorithm for determining whether or not a
poster's messages will be excluded from daily digests
(but not the archives), there is an online explanation,
but it takes a few steps to reach.  Start at:

     http://www.cryonet.org/cgi-bin/rate.cgi

and proceed to the page that prompts you with the various
rating options.  Then click on the "Get Reputation" hyperlink, which
takes you to a page with the link "Reputation Computation Algorithm".
Alternatively, just go directly to:

     http://www.cryonet.org/cgi-bin/reputation.cgi?help=y

to get the quick overview.  Although that page does not give a detailed
example, the idea is simple; one's reputation score is computed from a
weighted average of the ratings assigned by CryoNet readers, where the
weights are the values listed on that page.  In particular:

   "A poster with a reputation less than one third currently
   is considered a spammer."

That is what triggers exclusion of one's messages from the daily digests
(but not the archives).  No intervention from me is required.

If someone has posted many messages, then it takes several poor
postings to drop his/her reputation below a third, but someone who
posts infrequently, and whose postings are not well appreciated,
could reach that level quickly.

You can check your own rating anytime from the page:
     http://www.cryonet.org/cgi-bin/reputation.cgi
and, if you want, examine the ratings for each of your postings in detail.
If you want to find somebody else's rating, though, then you will
be disappointed, since the software is designed to reveal only your
own rating to you, not anyone else's.

It's not a perfect system, and I haven't completed all the Bayesian
fine-tuning of weights or automated aging of ratings originally envisioned,
but at least it seems workable, and, except for the time and effort
CryoNet readers invest to enter ratings for messages, it's fully automated.

     Kevin Q. Brown
     

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26179