Nan:2007-05-06-low-traffic-rfd

From: Jim Riley <jimrtex@pipeline.com>
Subject: Policy RFD: removing extremely-low-traffic unmoderated groups
Newsgroups: news.announce.newgroups, news.groups, news.groups.proposals
Followup-To: news.groups.proposals
Date: Sun, 06 May 2007 10:38:55 -0500
Organization: http://www.big-8.org/

                  POLICY REQUEST FOR DISCUSSION (RFD)
           removing extremely-low-traffic unmoderated groups

This is a formal Request For Discussion (RFD) to discuss a policy change 
in the Big-8 Usenet newsgroups.  For more information, see the proposed 
policy, listed below.


RATIONALE: 

A process for removing unused or little-used newsgroups can give
better definition to the process of creating new groups.  Without such
a process, the canonical list of newsgroups simply becomes a list of
newsgroups that were created according to whatever process was current
at the time, whether by a vote of potential users, by fiat of the
backbone cabal or Inet organizer, or by decision of the B8MB.

With a removal procedure, the list becomes one of groups that are
currently used.  New groups can be added on the belief that they will
also be used.  The proposed procedure provides a method for comparing
groups newly created by the B8MB against groups that are under
consideration for removal.

When Usenet was young, a news admin would notice that some groups were
empty, and propose their removal.  If there weren't too many
complaints or undue amounts of wailing, the groups would be removed.
At the time, retention times were short, perhaps three weeks, so the
above procedure meant that groups without any messages over the
previous three weeks might be considered for removal.

Later, when the group creation process was being codified, there was
discussion about a complementary process for group removal.  But a
system of Yes-No voting did not work as well for group removal as it
did for group creation.  A Yes vote could be considered to at least
nominally measure interest in participating in a proposed newsgroup,
while No votes were typically low enough in number to not derail too
many ordinary newsgroup creations.

But a Yes vote for a group removal doesn't measure interest or
disinterest in the group.  In effect, a Yes vote measured how many
people wanted to disregard any complaints or wailing from those who
wanted to keep the group and voted No.  On rare occasions, the group
creation process was used to remove groups, usually as part of a
hierarchy re-organization.  In those circumstances, a Yes vote might
be cast by those who favored other aspects of the re-organization and
would vote Yes on all items on the ballot.  In some cases, a Yes vote
was confusing, as when a Yes vote for a group meant the voter favored
removal, when ordinarily it meant they favored creation.

In 1997, Jani Patokallio proposed a two-step process for removing low
traffic groups.  The first step would identify low traffic groups, and
the second step would hold a CFV to determine whether the group would
be kept or not.  There would be no Yes or No votes, but only Keep
votes.  If 50 persons favored keeping a group, it would be kept.  In
e-mail discussion between Patokallio and Tale, Tale suggested that the
threshold for Keep votes be the same as for group creations, that is
100.  In other words, a low traffic group would have to re-establish
that it had the same level of support as a proposed new group had.

The process proposed in this RFD is similar to that proposed by Jani
Patokallio.  It would have a first step to identify low-traffic
groups, and a second step involving a vote to determine if a group
would be retained.  The threshold to keep a group is much lower.  As
the events of 2005 demonstrated, it is difficult to get 100 votes for
even straightforward group creations.  Instead, the threshold is
dynamic, with a requirement that at most only 10 persons speak up to
retain a group.

The system avoids making a determination of the worthiness of a
newsgroup, or even worse, the worthiness of its topic.  It simply
measures whether there is a modest level of interest in maintaining
the newsgroup.  This is consistent with the criteria that has been
used in the creation of almost all Big 8 newsgroups: "is there a
sufficient level of interest in the proposed newsgroup."


PROPOSED POLICY:

Removal would follow a two-step process.  The first step would
identify the lowest-traffic newsgroups.  The second step would
determine the least popular, as determined by a vote, among the lowest
traffic newsgroups.  These least popular groups would then be removed.

The process would be run annually, and be conducted on a
hierarchy-by-hierarchy basis.  This avoids comparing serious groups in
the comp.* and sci.* hierarchies with noisier groups in the rec.*,
soc.*, and talk.* hierarchies.  Only unmoderated groups would be
involved.

The lowest-traffic N% of newsgroups would be identified, based on
their non-cross-posted on-topic posts over the previous 12 months.
This identification process need not be perfect.  If there are groups
with lower traffic that are missed, they will likely be caught in
subsequent years.  If there are groups that are actually outside the
bottom N% in traffic that are included, participants will be able to
save the group by a vote.

The value of N for the first year would be 20%.  In subsequent years,
this number would be reduced if a relatively large share of the groups
are being retained due to the votes of their participants.

The B8MB may strike any groups that they choose from this list.  In
addition, in subsequent years, groups on the "do not bother" list (see
below) would be excluded.  New groups created by the B8MB that have
been in existence for at least 6 months would be added to the list.
This would allow confirmation that these new groups have at least as
much interest as groups that are being removed.

The vote would be conducted by STV, which requires only that the
voters rank the groups that they wish to have kept.  There would be no
obligation to vote for more than one group.  Under STV, there is no
way to vote against a group.  One can simply be more for keeping one
group than they are for keeping another group.

Groups that receive more than 50 first preference votes, would have
the votes scaled to 50 first preference votes.  This prevents large
numbers of voters who support one particular group from overwhelming
the process, but they would still save their group, and have some
influence over keeping some other groups.  The quota would be
determined based on having half the groups "elected" (ie, kept).  If
the initially calculated quota was greater than 10, it would be set to
10.

Counting would continue until either there are no continuing candidate
groups, or half of the groups have been eliminated (ie, determined to
be removed).

The above process ensures that at least half of the low traffic groups
will be kept, while also ensuring that any group that receives 10
first preference votes would be retained.

If a group received a high number of first preference votes, it would
be placed on a "do not bother" list.  A group that received 20 or more
first preference votes would be skipped the following year.  A group
that received 50 or more 1st preference votes would be skipped the
following two years, and a group with 100 or more 1st preference votes
would be skipped for three years.

After the votes were tallied, the B8MB may issue a reprieve for any of
the groups slated for removal.

The value of N% for the subsequent year would be determined in the
following  manner: The percentage of groups to be removed would be
calculated.  If this is less than 2.5% it would be set to 2.5%.  The
maximum possible value is 10% (ie, half of 20%).  This value would
then be doubled to become the new value of N%, so that the new value
of N% would range from 5% to 20%.


PROCEDURE:

For more information on the newsgroup creation process, please see:

  http://www.big-8.org/dokuwiki/doku.php?id=policies:creation
  [need something better here]

Those who wish to influence the development of this RFD and its final 
resolution should subscribe to news.groups.proposals and participate in the 
relevant threads in that newsgroup.  This is the best method of making sure 
that one's comments or criticisms are heard.

All discussion of active proposals should be posted to news.groups.proposals.
To this end, the 'Followup-To' header of this RFD has been set to this group.

We urge those who would be affected by the proposed policy to make a
comment to that effect in this thread; we ask proponents to keep a list
of such positive posts with the relevant message ID (e.g., Barney Fife,
<4JGdnb60fsMzHA7ZnZ2dnUVZ_rWdnZ2d@sysmatrix.net>).  Such lists of positive
feedback for the proposal may constitute good evidence that the group will be
well-used if it is created.




DISTRIBUTION:

This document has been posted to the following newsgroups:

  news.announce.newgroups
  news.groups.proposals
  news.groups


PROPONENT:

Jim Riley <jimrtex@pipeline.com>



CHANGE HISTORY:

2007-05-06     1st RFD