spaf@gatech.CSNET (Gene Spafford) (08/13/85)
Well, I have received a great deal of mail (well into the
hundreds, I believe) as a result of my articles about deleting
groups.
I am glad that enough people are willing to take a few moments to
write something, but a mite distressed at the flood of mail
I received (and that I could not possibly answer individually).
I was also very pleased with the general tone of the majority
of the responses. Normally, when I suggest something of
this type, I get lots of nasty mail.
I posted the original suggestion about deleting some groups to solicit
some feedback and to investigate the possibility of cleaning up some
"deadwood" groups. I also wanted to remind people that the groups
existed and possibly stir up some postings. The comments I received
ranged from long arguments to simple pleas. I received many requests
to keep the following groups: net.notes, net.bugs.v7, net.rec.birds,
net.lang.apl, net.rec.scuba, net.rec.skydive, net.math.symbolic,
net.decus, and net.games.go (in fact, for most of those groups, I
received more letters than there have been articles posted in the last
6 months...interpret that as you will). I got only one or two requests
to keep net.theater, net.usoft, and net.std. I got no requests asking
to keep net.micro.432. Therefore, I will be sending out "rmgroup"
messages on those 4. If interest can be shown by more than 2 or
3 people in the future, we can always recreate those groups.
I got a number of similar comments and questions which I'd like to
summarize (and respond to) here. I also got some comments specific
to certain groups, and I'll paraphrase those here.
Why delete low volume groups?
Currently, the only criteria we have for deleting a group is if
it falls inactive. It isn't sensible to delete a very active
group (although it might be better for the net if we did delete
some high-volume groups like net.flame and net.religion; but
that's a different story). The groups I mentioned haven't
had much activity recently and therefore appeared to be good
candidates.
But what harm does it do to leave them?
It is important that we continue to make some attempt to
trim the net. There is an incredible number of groups
currently in existence, and just leaving "dead" groups around
is wasteful.
We require a demonstrated need to create new groups. We should
likewise require some demonstrated need to keep groups around.
Admittedly, the activity of a group is not a perfect measure,
but it is the only one I can apply right now.
Furthermore, the argument can be made that we should attempt
to keep the name space of newsgroups as small as practical so
that the average user will be able to better determine the
appropriate groups for posting. Again, I think this is a bigger
issue and gets into the whole problem with the naming hierarchy
for news, so I won't go into it here.
The groups you targeted have a seasonal audience.
Maybe so, but that audience is extremely small. None of those
groups has seen much activity in the past few years (or since
their creation, whichever came last). If a flood of articles
on any of those groups would occur once semester-oriented schools
come back into session, we can recreate the groups. Again,
let's demonstrate a REAL need for them. Have any of you ever
heard of "zero-based" budgeting?
Those groups have a higher signal/noise ratio than the other groups --
they should be judged differently.
Perhaps so. I'm not going to pretend that I am in position
(or that ANYONE is) to judge the true value of any of these
groups. I would agree that something like net.lang.apl might
seem to be worth more than a net.flame (for example), but that
is just my particular orientation. In the context of the net,
it is really difficult to make such a judgment *in relation to
the entire net readership.*
If these groups work best with a small readership, and since
they seem to enjoy such a small posting audience, it might be
better to have them as moderated groups or mailing lists...or so I
might argue. Comments?
Just who are you anyway? What gives you the authority to suggest
deleting MY favorite group?
I'm a net user, just like the rest of you. I have been on the net
for almost 3 years, and I help to administer the news on about 8
sites around here. I also happen to maintain the "semi-official"
list of newsgroups. Those things don't grant me any special
authority. They do, however, give me a little bit of a different
perspective than many net users -- not necessarily a better
perspective, just a different one.
Anyone can suggest that a newsgroup be deleted. And anyone who
knows the format can actually put together a "rmgroup" message and
distribute it. I happen to believe it is better to follow accepted
procedure and get public comment. No one else has indicated an
interest in deleting groups recently. If the course of action I
decide is not accepted by the majority of administrators and
readers of the net, I'm sure it will be undone.
Net.notes should remain around, but the name is far from ideal.
It should be named "net.news.notes".
This suggestion came from at least 5 people. I agree.
Does someone want to formally suggest it and tally votes, etc?
Net.theater is a new group -- it should be given more time.
I only got 4 response from people who wanted the group kept, and
two of those people admitted that they only used the group
with a distribution of less than "net". The group just doesn't
seem to be living up to its intended purpose. If there is
a very active "theater" group in some distribution area,
it would be better to create a local group for their use.
I got mail from officials of formal organizations related to each of
net.lang.apl, net.games.go, and net.decus asking me to keep them. A
couple of people from the Decus organization promised that if they
didn't get more active they'd delete the group themseleves in 6
months.
At least 10 people wrote and urged me to trash all the mentioned groups,
and while I was at it take net.flame, net.general, net.abortion,
net.politics, net.origins, net.religion, and net.jokes (not all of
these were suggested by each). I didn't count these responses
carefully, nor did I (or do I) solicit such mailings. Someone
else can tally such urgings.
I got the following mail from reid@Glacier (Brian Reid), and I think
he made some interesting points. I'd like to excerpt portions of
that mail here (with Brian's permission):
> The net serves three fundamentally orthogonal needs:
> 1) It is a communications vehicle by which people with certain
> interests and needs can communicate information with one another.
> 2) It is a social vehicle by which people develop a sense of
> electronic community.
> 3) It is a soapbox that satisfies a hunger for attention, a desire
> to have one's words be published. A vanity press, if you will.
>
> Need #1 is fulfilled by the hobbyist groups (net.rec.ski, net.bicycle,
> net.audio, etc.) and by the technical exchange groups (net.bugs,
> net.math.symbolic, net.lang.c, etc.)
>
> Need #2 is fulfilled by the personality/social club groups (net.jokes,
> net.sport.football, net.singles, net.motss, etc.)
>
> Need #3 is fulfilled by the soapbox groups (net.flame, net.politics,
> net.religion, etc.)
>
> It is important that you realize that these are three fundamentally
> different kinds of netnews groups, populated by fundamentally different
> kinds of readers.
>
> Type-1 groups (information exchange) will have many more readers than
> writers. For example, there are a large number of readers of net.movies
> at Stanford, but almost none of them ever posts anything. Nearly every
> user on Glacier reads net.lang.mod2, but we rarely post anything.
> Someone who was measuring newsgroup worth by looking at writer count
> instead of reader count would get a very distorted view of the worth of
> these things.
>
> Type-2 groups (club membership) seem to have the property that a huge
> fraction of the people who read them also post to them; that's part of
> the fun. Practically everybody who reads net.singles posts to it sooner
> or later, though some people subscribe for a while, read it for a
> while, then unsubscribe without ever posting anything. Some groups kind
> of hover at the boundary between information exchange and club
> membership; net.bicycle is a good example of this.
>
> Type-3 groups (vanity press) have the unusual property that, at least
> around here, they have more writers than readers. I know plenty of
> people who have posted to net.flame but who do not subscribe to it.
> In the limiting case, these vanity-press groups divide themselves up
> into time-division-multiplexed subgroups. The people in net.flame who
> are talking about women's use of toilet paper do not read the messages
> about what we should do with terrorists, and the people who read the
> messages about terrorists do not read the messages about toilet paper.
>
> The net was founded for the purpose of supporting type-1 groups, and
> then we discovered that type-2 was fun, too. However, type-3 groups
> fill a deeply-felt need in certain people, and the temptation was too
> great to resist. As either Mark or Chuq pointed out, the type-3 groups
> were founded to keep type-3 material out of type-1 and type-2 groups,
> by asking that the people who want to stand on the soapbox provide an
> audience for the others who want to stand on it too.
Perhaps it is time we discussed setting up 3 sets of rules for
creation and deletion of newsgroups?
My thanks to everyone who responded by mail or in postings.
--
Gene "4 months and counting" Spafford
The Clouds Project, School of ICS, Georgia Tech, Atlanta GA 30332
CSNet: Spaf @ GATech ARPA: Spaf%GATech.CSNet @ CSNet-Relay.ARPA
uucp: ...!{akgua,allegra,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spafjjhnsn@ut-ngp.UTEXAS (J. Lee Johnson) (08/13/85)
Gene Spafford (spaf@gatech) writes: > (in fact, for most of those groups, I > received more letters than there have been articles posted in the last > 6 months...interpret that as you will). > If these groups work best with a small readership, and since > they seem to enjoy such a small posting audience, it might be > better to have them as moderated groups or mailing lists...or so I > might argue. Comments? But later in the same article <813@gatech.CSNET> he quotes Brian Reid (reid@glacier), who gives an excellent explanation of what's happening: >> Type-1 groups (information exchange) will have many more readers than >> writers. For example, there are a large number of readers of net.movies >> at Stanford, but almost none of them ever posts anything. Nearly every >> user on Glacier reads net.lang.mod2, but we rarely post anything. >> Someone who was measuring newsgroup worth by looking at writer count >> instead of reader count would get a very distorted view of the worth of >> these things. Consider net.rec.birds, a model newsgroup. It has a high signal/noise ratio and almost no abuses or rudeness. Net.rec.birds has an exact purpose that would not be appropriately accomplished in another group. There have been 35 articles in the last 2 weeks. This is just about perfect as far as I am concerned; after all, I have other interests and responsibilities. Of the 35 articles, only 3 were cross posted to other groups. Two of these were the net.news.group postings by spaf@gatech. The other was a summary article on binoculars (very appropriate). I wish net.unix, net.unix-wizards, and net.lang.c were more like this. I believe they are becoming Type-3 groups (vanity press). I applaud the efforts of people like Gene to keep USENET from collapsing under its own weight, but in this particular case I believe the approach was wrong. P.S. Although I have read almost every article posted to net.rec.birds, this article will be the first one I ever posted to it. -- James Lee Johnson, U.T. Computation Center, Austin, Texas 78712 ARPA: jjhnsn@ut-ngp UUCP: ihnp4!ut-ngp!jjhnsn allegra!ut-ngp!jjhnsn gatech!ut-ngp!jjhnsn seismo!ut-sally!jjhnsn harvard!ut-sally!jjhnsn
ray@othervax.UUCP (Raymond D. Dunn) (08/13/85)
References: >..flame...flame.....whats going on here.........flame...flame.... >..mumble..mumble....threats to resign...........mumble..mumble... >..bitch...bitch.....leave this group alone......bitch...bitch.... >..grumble.grumble...unilateral group removal....grumble.grumble.. >..etc.....etc.......etc..etc....................etc.....etc...... Half burner on: I can understand the outcry when a newsgroup becomes congested with irrelevant material, or its content threatens the very existence of the net because of cost or legal implications. I can also understand why those who have *chosen* to become 'net gurus' like to keep their house clean (as they see it), and that it is only reasonably to become upset when flamed at. However I believe that volunteers are rarely altruistic. Even though they may provide a valuable service relied upon by others, they do the job because they find it satisfying and self fulfilling in one way or another, and this is rarely based on other than internally generated motives (generalisation...generalisation...generalisation (what is experience if not that?)). Whether or not this is true in a specific case, volunteers should be no more immune to criticism than if they were doing the job professionally. Dealing with their public, with all the frustration and sometime nastiness that that implies, is an integral part of that job. 'Threats' to resign are neither professional nor facing up to the realities! So, compromise a little! Resist some of the things you personally are dying to see changed on the net but which dont' matter in the *least* to the scheme of things, are changes for their own sake, and will cause a minority to flame! Why start a long discussion as to whether net.xxx should be removed because it has *LITTLE* traffic. If it does, then it is an innocuous little beast and should be left sleeping! Else it may bite (and at the least cause major wailing come Easter when resurrection is attempted)! Let the gurus respond to the net rather than the net having to respond to the gurus! Keep up the good work guys, you're doing a great job! Ray Dunn. ..philabs!micomvax!othervax P.S. How about removing net.general? - everyone has been so intimidated that there is so little traffic on it these days it is obviously redundant - just think of all that space its directory entry is wasting on all those machines!
kre@ucbvax.ARPA (Robert Elz) (08/16/85)
I would have thought that someone would have remarked on this before, but, apparently not. A while back someone said that if there was some fixed limit to the number of newsgroups, that would justify (in his opinion) removing dead groups, but otherwise he preferred keeping them. Well, big surprise people, there is just such a limit, in every version of news I have seen (up to, and including 2.10.3). The limit is on the number of characters in all the newsgroup names concatenated together (including separators). On ucbvax the active file is currently about 60% there already (of the limit in 2.10.3 - I have a vague memory that the limit was set smaller in older versions). True, this is a configurable number, sites can change it, but I bet almost no-one has. And the way its used, the net effect is that the smallest limit out there wins. Asking people to recompile news with a bigger number (even if its possible due to restricted memory sizes on some archaic processors) won't get us anywhere - from the dates of some of the news versions people are still using, the source must have gone mouldy & been discarded well before now. Its possible that this limitation might be removed in some future version of news, but getting people to upgrade is impossible. Thus, if we want to keep making new groups, we simply *must* get rid of some of the old ones. Robert Elz ucbvax!kre kre@monet.berkeley.edu
smb@ulysses.UUCP (Steven Bellovin) (08/16/85)
> I would have thought that someone would have remarked on this before, > but, apparently not. > > A while back someone said that if there was some fixed limit to the > number of newsgroups, that would justify (in his opinion) removing > dead groups, but otherwise he preferred keeping them. > > Well, big surprise people, there is just such a limit, in every version > of news I have seen (up to, and including 2.10.3). > > The limit is on the number of characters in all the newsgroup > names concatenated together (including separators). On ucbvax > the active file is currently about 60% there already (of the > limit in 2.10.3 - I have a vague memory that the limit was > set smaller in older versions). Just as a case in point, ulysses lost about 12 hours worth of news last night (and I lost a lot of time cleaning up after it) because net.music.guitar blew that limit... Which brings up a suggestion -- would it be that hard for rnews to try a bit harder to keep an article when things have blown up? Say, try saving it in /usr/spool/news/SAVE/dd.hh.pid (which should be unique for the life of the problem)? Maybe even a list of alternate directories to try, site- configurable, for when /usr/spool is out of space or i-nodes.
peter@baylor.UUCP (Peter da Silva) (08/16/85)
> P.S. How about removing net.general? - everyone has been so intimidated that > there is so little traffic on it these days it is obviously redundant - > just think of all that space its directory entry is wasting on all those > machines! We have several hundred K of these empty directories. I would like to see a few groups combined... like for example info-term (which is mostly requests for termcap entries) and net.unix, a place where such requests are more likely to be noticed. And what in the hell is net.cse about? -- Peter da Silva (the mad Australian werewolf) UUCP: ...!shell!neuro1!{hyd-ptd,baylor,datafac}!peter MCI: PDASILVA; CIS: 70216,1076
spaf@gatech.CSNET (Gene Spafford) (08/20/85)
In article <424@baylor.UUCP> peter@baylor.UUCP (Peter da Silva) writes: >.... And what in the hell is net.cse about? Well, checking the latest edition of the list of newsgroups reveals: net.cse Computer science education. I'm not sure that it couldn't be combined with net.college, though.... -- Gene "4 months and counting" Spafford The Clouds Project, School of ICS, Georgia Tech, Atlanta GA 30332 CSNet: Spaf @ GATech ARPA: Spaf%GATech.CSNet @ CSNet-Relay.ARPA uucp: ...!{akgua,allegra,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spaf
hansen@pegasus.UUCP (Tony L. Hansen) (08/27/85)
In article <424@baylor.UUCP> peter@baylor.UUCP (Peter da Silva) writes:
<
< We have several hundred K of these empty directories. I would like to see a few
< groups combined... like for example info-term (which is mostly requests for
< termcap entries) and net.unix, a place where such requests are more likely
< to be noticed. ....
It is precisely BECAUSE net.info-terms is a separate newsgroup from net.unix
that requests in there get noticed and responded to. If requests in there
were posted to net.unix, they would never be seen due to the swamped
situation in that other newsgroup. I've responded to many requests for
termcap or terminfo entries that I may have never seen if they had been
posted to net.unix.
Tony Hansen
pegasus!hansenwcs@ho95e.UUCP (x0705) (08/28/85)
> In article <424@baylor.UUCP> peter@baylor.UUCP (Peter da Silva) writes: > < > < We have several hundred K of these empty directories. I would like to see a few > < groups combined... like for example info-term (which is mostly requests for > < termcap entries) and net.unix, a place where such requests are more likely > < to be noticed. .... > > It is precisely BECAUSE net.info-terms is a separate newsgroup from net.unix > that requests in there get noticed and responded to. If requests in there > were posted to net.unix, they would never be seen due to the swamped > situation in that other newsgroup. I've responded to many requests for > termcap or terminfo entries that I may have never seen if they had been > posted to net.unix. > > Tony Hansen pegasus!hansen Also, there are ARPANET gateway issues; net.unix postings go to INFO-UNIX, and I think net.info terms postings go to INFO-TERMS? Traffic from the Arpa side to UNIX is easy to join, but the UNIX-Arpa direction requires two separate groups for the two mailing lists. -- ## Bill Stewart, AT&T Bell Labs, Holmdel NJ 1-201-949-0705 ihnp4!ho95c!wcs