[Bug-gnupedia] A Strawman Proposal for a Moderation System

[Robert Fisher]

First, let me say that I think the concept of a comprehensive, GPL'ed
encyclopedia, will be an enormous gift to the intellectual heritage of the
world. We want the entries to be of the highest quality, and as many
posters have commented, a comprehensive moderation system will be
essential in order to achieve this goal. At the same time, most posters
(in addition to Stallman) have agreed that we do not wish to establish a
preisthood in charge of the GNUpedia (hereafer Gp), and that we do not
wish to artificially rank posters by the volume of their posts. 

I submit the following moderation system as a strawman proposal for
further discussion. With enough constructive criticism, we can create an
insanely great moderation system which will enable an insanely great
Gp. 

Principles
----------

The principles underlying any good moderation system are :

        (*) Initially equal rankings for all; status must be earned

        (*) Everyone is entitled to rank any article (once)

        (*) No serious articles are dropped from the archive; anyone is
              entitled to post any article at any time, and anyone
              is entitled to view any article at any time

(This last point allows for the possibility to drop spam and troll posts
from the archive, assuming a sufficient number of readers agree, since we 
do not wish store such material.)

System
------

We will first discuss a very simple system which enables these
principles. This system can be (and perhaps should be) further refined, as
I mention below. Right now, we just want to get the discussion rolling.

        (*) Each reader, after viewing an article, can choose to rank the
             article on a numerical scale. This scale should be fine
             enough to distinguish good articles from great articles, and
             also allow postings to be designated as spam or troll posts.
             However, it should also be coarse enough to allow readers to
             rapidly assign a ranking to any article.

        (*) All postings begin with rank 0. A rank 0 article is one
              of no particular merit, though is still a serious post.

        (*) For the purposes of this discussion, let's say the scale goes 
              from -M to N. (For instance, -3 to 5 -- we shouldn't have
              delve very deeply into just how bad an article is!).

        (*) The mean ranking of the article by all readers is used as the
              primary ranking of that article.

        (*) The mean ranking of all aricles written by an author is used
              as the primary ranking of the author.

        (*) Articles determined to be in the range -M to -M+1 by a 
              statistically significant portion of readers will be dropped
              from the archive as spam or troll posts.

Possible refinements on this system include

        (*) Weighting the rankings of articles by the ranking of the
              reader's author rank. Authors who have published
              authoritative articles in the past will have reviews that
              count for more. (A more refined system still would allow
              for authors to gain different rankings in different field;
              high-ranking science authors may not also be high-ranking
              authors in the humanities, though such a system would allow
              for that possibility.)

        (*) Weighting the rankings of article for their statistical 
              significance. The more reviews, the greater the significance
              of the article. This allows one to distinguish an article
              which has been determined to be of the highest rank by a 
              single reader from an article which has been determined
              to be of the highest rank by a thousand readers.


Comments and criticisms welcome.

Robert Fisher