spaf@gatech.CSNET (Gene Spafford) (10/21/85)
We have some problems about creating new groups. Specifically, people
don't seem to be inclined to follow the established procedures for
creating new groups. If this is due to the fact that people don't
agree with those procedures, then we need to discuss changing them.
First, let me outline the procedures the net has been using for the
last three years (that is almost as long as I have been on the net -- I
don't know what was done before then, when the net was young and
considerably smaller):
1) FIRST a need for a group must be established. This is done either
by showing that a mailing list has sufficient volume and widespread
readership that it makes sense to turn it into a newsgroup, -OR- the
volume of postings on a particular topic within some existing
newsgroup(s) gets so large for so long it makes sense to spin off a
separate newsgroup for the topic. Creating a newsgroup just because
the topic is interesting is *NOT* something we have had as part of this
procedure.
2) NEXT a discussion is held in the net.news.group newsgroup, as well
as any other groups related to the proposed new group. This discussion
should cover the need for the new group, the proposed charter of the
new group, alternative names, and so on.
3) ALL objections, suggestions, and comments generated by #2 should be
addressed and resolved in some form of near-consensus manner.
4) FINALLY, a vote is taken to see if the group is a good idea and
really wanted. We have been asking for a minimum of 40 to 50 "yes"
votes above and beyond the number of "no" votes before creating a
group. Ideally, 100 or more such votes should be recorded.
=====
The argument is often raised "But what harm does it do to create (keep)
another newsgroup, especially if the volume is small?" The answer is:
"Plenty." The volume of postings is so large now that at least 4
backbone sites are no longer carrying or forwarding all the newsgroups,
and more are considering cutbacks. Each new newsgroup established
brings in new postings, especially if there was no previous need for it
(as happened with net.bizarre). So, on a net-wide basis, newsgroups
add to the volume problems, even if they contribute a very small amount
to the overall volume. With 1200+ sites worldwide, even small (local)
volume adds up to a large amount netwide.
New newsgroups tend to encourage people to think up other newsgroups --
adding to the problem ("Gee, there's a net.micro.ibm-pc.256K, but no
net.micro.ibm-pc.512k.with.harddisk.and.mouse for MY system."). More
newsgroups add to the difficulties new users have in figuring out where
to post their articles -- often resulting in their posting to
inappropriate places and to too many groups (how many newsgroups might
be appropriate for an article about a bug in a C compiler running under
Xenix on a PC-AT? I can think of at least seven.). And much of the
software out there has built-in limits as to the number and size of
newsgroups that can be handled. Some systems have already had problems
because the number of groups exceded the compiled-in minimums.
Complicating the situation, we currently have no established procedure
for deciding when to remove a group. Candidates for removal run from
groups with (probably) under 25 readers net-wide, such as
net.rec.birds, to netwide garbage pails like net.flame. How do we
decided what is worth keeping and what to delete? Who decides? And
when?
In summary, more newsgroups do not appear to be a "good thing,"
especially not until we trim the current net volume and establish a way
of removing groups that shouldn't be kept anymore.
======
Proposals have been made recently for groups like "net.personals",
"net.os" and "net.docs". Although some people think these are good
ideas in some way or another, there has yet to be any *demonstrated
volume* of postings on these topics in any newsgroup. Thus, there is
no real need for separate newsgroups on those topics. (There are other
objections to "net.personals" and "net.docs", but that should be
enough.)
Net.bizarre was created without going through the above procedures, and
that's one of the reasons we killed it.
Now there is also a problem with "net.internat". The group was created
without #1 or #3 being done, and out of order. I missed the "newgroup"
message when the group first started, or I would have acted sooner. I
also have been in correspondance with the person starting the group,
and hoped that he would post an explanation somewhere, but he hasn't
(yet). There was no volume of postings on the topics presented in
net.internat anywhere else on the net before the group was created.
Furthermore, the discussion in net.news.group saw a few people ask why
the group shouldn't be made a second-level group under net.unix (e.g.,
net.unix.internat), if it should be created at all. This was never
addressed by anyone in favor of creating the group -- the group was
just created.
For these reasons, I have not included net.internat in my list of
active lists. It is also why I am sending out a "rmgroup" for it
sometime tomorrow. If we won't allow a net.bizarre to exist because it
was created outside of established procedure, we cannot allow
net.internat to be created either. The purposes and nature of the
groups are different, but the point remains the same. Arguments about
merit of a newsgroup really aren't appropriate here -- if the net is to
be judged based on the volume (=popularity?) of newsgroups, then Usenet
is not primarily for technical or theoretical discussions. Rather, the
net is a forum for cranks and flamers with some technical content
thrown in now and again.
=====
Perhaps it is time we explored defining a new set of rules for creating
and retaining newsgroups. Volume of postings is not the best criterion
for measuring the utility of a group, but it certainly is the easiest
to measure. We probably need to come up with a new procedure for
deciding whether to create a new group, and when to delete an old one.
If you have any thoughts on these topics, post them as a followup to
this article, only post them to "net.news".
Until some new procedures are established, I will continue to maintain
the list of groups by the established procedure. That means that groups
created outside the procedures get "rmgrouped", without regard to
subjective "value," as the only thing "fair" to everyone.
----
Gene Spafford
The Clouds Project, School of ICS, Georgia Tech, Atlanta GA 30332
CSNet: Spaf @ GATech ARPA: Spaf%GATech.CSNet @ CSNet-Relay.ARPA
uucp: ...!{akgua,decvax,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spaf
--
Gene "sometime in 1986" Spafford
The Clouds Project, School of ICS, Georgia Tech, Atlanta GA 30332
CSNet: Spaf @ GATech ARPA: Spaf%GATech.CSNet @ CSNet-Relay.ARPA
uucp: ...!{akgua,decvax,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spaf