Thoughts on the new-group creation process (was Re: holography)
Werner Uhrig
> In <23...@moocow.uucp> bro...@moocow.uucp (Kent C. Brodie) writes:
> Kent> DOES the field of holography (both from a general, hobbyist, and
> Kent> scientific view) warrant a newsgroup? I sure as heck think so.
>
> I don't. I think the topic has much merit for discussion, but I am
> unaware of sufficient, current traffic on it to justify a group. (If
> I am wrong about this, I would probably change my opinion.)
the question is really: how do I "locate" the netters who would
show interest in a discussion? I venture the opinion that:
"certainly not by posting to comp.misc or news.groups!"
given that all new groups catch my attention as I get to see the
first message (which mostly leads me to "unsubscribe") I find this
the most efficient way (for me) to keep track of new groups and
their purpose (hopefully, the first message explains that purpose :-)
Now, were it not for that (not so) mythical hard limit of total
number of news-groups, I'd see no reason why there should not be
a move to a liberalization (making it easier and faster) of the
rules for new-group creation, coupled, hopefully, with a "sunset-rule"
which leads to the removal of news-groups (personally, I find
mailing-lists of a small group of people who are *REALLY* interested
in a topic to be quite satisfying - and I would like to see people
encouraged to move news-groups with low-activity to that format)
of course, there is also the "nuisance factor" to consider which
is represented by having to unsubscribe to "many" new groups -
but given that I always get those presented AFTER I have viewed
all the groups that I subscribe to, it is easy for me to q(uit)
reading news as soon as I get to see the first new group - and I deal
with those on the weekend, or whenever I care to pay attention to
the new groups.
Regarding the question raised by Chuq of "shortening the voting
period", I am all for it: 2 weeks would be fine with me, 3 weeks
seems more than long enough. So what if a few people happen not to
get to vote? it should balance all out. and what if a vote fails
after 2 weeks that would have made it after 3 or 4 by the narrowest
of margins? I'm quite satisfied that within 2 weeks a representative
and sufficiently large group of netters get to vote, even if that does
not include me (or my mother-in-law in the Venezuelan bush - she may
feel different, of course :-)
Then there is the question of "pre-vote discussion period" and
"post-vote validation period" - both of which have aspects that
are caused by not having a somewhat consistent and centralized
group of people that are interested in and volunteer to "moderate"
the procedure. Here is how it could/should happen:
when a netter gets "the itch" for a new group, s/he POSTS an article
to news.announce.newgroups, which due to the nature of the group,
gets forwarded as mail to the moderator(s) address. this address
should be a redistribution-list to several volunteers who are willing
to advise regarding the new-group creation:
1) how to best describe the intended purpose of the group
or, what groups or private mailing-lists already do exist
for that (or related) purpose ...
2) how to separate the question of "what to name the group" from
"proposing the creation of a group for a specific purpose."
this advice should be given in regards to both educating and
explaining the purpose/intention of the top-level naming
hierarchy of news-groups and the limits of distribution implied
by the name ; furthermore. this advice should be given in the
spirit of maintaining the intended purpose of separating groups
into families of comp.*, sci.*, rec.*, talk.*, etc.
3) sometimes it seems advisable to start a pre-vote discussion on
the question of purpose and name for a group, sometimes not.
the news.announce.groups moderator(s) should be allowed to
require (or waive) such a discussion-period before posting a
call for avote in the moderated group.
4) last, the news.announce.groups moderator defines the address(es)
where votes are to be sent. This provision should be used by
creating an alias on a (hopefully central machine, such as UUNET
but any other will do) where all votes are kept for verification
purposes and from where they are forwarded to the person who
does the counting. Initially, this can be the address of the
person proposing the vote, as currently done, but I am willing
to start a tradition of separating the vote-counting from the
proposing by taking on the next (few) vote-tallying by providing
an address for votes to be sent to.
This should take care of the need of any post-vote delays (given
that I can be trusted :-) and that I can make all votes available
to others with FTP-access...
Suffice it to say, that one (I) might put a little peer-pressure
on the person proposing the last group to take on the vote-count
for the next! :-)
--
-----------> PREFERED RETURN-ADDRESS FOLLOWS <--------------
(ARPA) wer...@rascal.ics.utexas.edu (Internet: 128.83.144.1)
(UUCP) ..!utastro!werner or ..!uunet!rascal.ics.utexas.edu!werner
Chuq Von Rospach
>> unaware of sufficient, current traffic on it to justify a group. (If
>> I am wrong about this, I would probably change my opinion.)
This is a holdover from ancient days. While I was at one time a strong
supporter of the rule (primarily because it's difficult/impossible to
*delete* newsgroups) I now think it's a specious argument. The net is simply
too large these days for us to expect that all of the people who are
interested in Lithuanian Sheepdogs to easily find each other in some random
newsgroup to 'prove' they deserve a group.
It is also a rule that is inconsistently enforced. If 'we' (the generic we)
don't think 'you' (the generic you) deserve the newsgroup, we pull up this
strawman as a way of stopping the creation process. If, on the other hand,
'we' happen to think the idea is a good idea anyway, 'we' sort of ignore it.
Case in point on the latter, comp.unix.aux, which I was the coordinator on --
it had no visible traffic, being a completely new system, yet we sort of
glossed over that because there was a general feeling that the group ought
to exist for when the volume did show up -- to keep it from spilling out
into any of half a dozen other groups. I think that decision was correct,
and I think the group's been a success.
It's time to put this stupid strawman to rest once and for all. It's primary
purpose was to try to hold back or limit expansion of the network -- by
limiting groups to being better place-holders for things we were already
doing. It's done a wonderful job at that, based on the statistics we see
(that's a joke, son).
It's a subjective rule that is observed subjectively. Let's get rid of it.
If someone comes up with a good idea for a group and can find enough people
to vote it into service, let the group live or die on its own merits. I
don't see *any* rational basis for a "we have no volume, so it's not worthy"
argument. Time to let it die.
The *real* problem is not creation of specious groups, but the inability to
delete groups once they exist. I think it's time we seriously consider some
form of sunset law -- if a group dies, make it go away. I think it's better
for the net that we let groups be born and try to find an audience than make
them stillborn and not see to their potential.
My personal preference (which I've mentioned a few times over the years) is
to set a cap to the number of groups on USENET. Once you hit that cap, to
create a group requires the deletion of another group. That way we can clear
out the deadwood while still giving the people who like to argue such
thing wonderful fodder for flamewars... Instead of arguing about the merits
of rec.holography, we can argue whether rec.holography is better for the
network than comp.std.internet (to choose just one piece of deadwood).
Inspired flamewar folks could, for instance, argue that we should create
sci.fusion.lukewarm by deleting talk.religion.misc. So we can fix up some of
the process while not removing the 'fun' from the newsgroup wars!
> I'd see no reason why there should not be
> a move to a liberalization (making it easier and faster) of the
> rules for new-group creation, coupled, hopefully, with a "sunset-rule"
> which leads to the removal of news-groups (personally, I find
> mailing-lists of a small group of people who are *REALLY* interested
> in a topic to be quite satisfying - and I would like to see people
> encouraged to move news-groups with low-activity to that format)
Definitely. I would suggest that things with no volume for six months
through the backbone (it would be fairly easy to write addendums to what
uunet is doing to track) be put on a list of potential retirees that have to
justify their continued existence or disappear from the official lists.
I would also suggest that we consider looking at Brians Bottom 40 in
popularity to see if there are some groups that just don't deserve to
continue to exist -- especially (and I kno this won't be popular) in the
very low readership/high volume newsgroups.
--
Chuq Von Rospach <+> Editor,OtherRealms <+> Member SFWA/ASFA
ch...@apple.com <+> CI$: 73317,635 <+> [This is myself speaking. I am not Appl
Segmentation Fault. Core dumped.
Dave Sill
>[Arguing against a new group based on the lack of current volume on the
>topic] is a holdover from ancient days. While I was at one time a strong
>*delete* newsgroups) I now think it's a specious argument. The net is simply
>too large these days for us to expect that all of the people who are
>interested in Lithuanian Sheepdogs to easily find each other in some random
>newsgroup to 'prove' they deserve a group.
Which was exactly the purpose of the trial newsgroup and the Trial
Newsgroup Announcement in my "ill-conceived" new group creation
procedures. For those who missed it, ignored it, or only skimmed it,
rather than initiating a new group by starting with a Call for
Discussion, under my alternative procedures the interested party would
announce, for example, that a trial group for the discussion of
Lithuanian Sheepdogs was being formed. This announcement would go out
to the same groups that a Call for Discussion would, perhaps rec.pets,
soc.culture.lithunania, alt.sex.bestiality, etc. The trial newsgroup
would be conducted in an appropriate *.misc group, rec.misc in this
case, and messages would be marked as belonging to the group by
containing the name of the trial group within square brackets at the
end of the subject. E.g.,
Subject: Re: Breeding advice needed [rec.lithuanian-sheepdogs]
>It is also a rule that is inconsistently enforced. If 'we' (the generic we)
>don't think 'you' (the generic you) deserve the newsgroup, we pull up this
>strawman as a way of stopping the creation process.
I seem to remember precious little support when I argued that
doubter's NO votes were A Bad Thing. Since we can't really tell
whether a NO vote is for a valid reason, I thought it would be
sufficient to demonstrate that there *was* sufficient traffic for a
new group. And there have been cases where apple-pie sounding groups
have been created and turned out *not* to have sufficient need in the
first place.
>It's time to put this stupid strawman to rest once and for all. It's primary
>purpose was to try to hold back or limit expansion of the network -- by
>limiting groups to being better place-holders for things we were already
>doing. It's done a wonderful job at that, based on the statistics we see
>(that's a joke, son).
Whenever someone argues that "more groups" == "more traffic" I wonder
if they also feel that the number of new books published is a result
of new genres or recently discovered ranges of the Dewey Decimal
System. Newsgroups are a way to classify news articles; they don't
generate them. In an ideal net, which USENET doesn't even approach,
there would be one group for every major topic, ranging from the
general to the specific, and articles and replies would be
cross-posted to *all* appropriate groups. At least on UNIX systems
this is only marginally more resource-intensive, and it would be much
more convenient for user.
>The *real* problem is not creation of specious groups, but the inability to
>delete groups once they exist. I think it's time we seriously consider some
>form of sunset law -- if a group dies, make it go away. I think it's better
>for the net that we let groups be born and try to find an audience than make
>them stillborn and not see to their potential.
I concur. I also mentioned a while back that we could use a challenge
procedure to declare groups inactive. If a group's been quiet for a
couple months someone could post a challenge on it, providing a 21-day
period during which participants who wanted the group to remain would
reply to the challenger. If at least 100 people respond favorably,
then the group remains. Otherwise it's removed, probably by some net
weanie such as Chuq or Greg (pending a 5-day period for validating the
"vote", of course).
>My personal preference (which I've mentioned a few times over the years) is
>to set a cap to the number of groups on USENET.
Is there some magic number of groups USENET should have? What happens
when the net is so large that there are no dead groups left to remove?
>Instead of arguing about the merits
>of rec.holography, we can argue whether rec.holography is better for the
>network than comp.std.internet (to choose just one piece of deadwood).
Why's it better to argue the latter than the former? If a group's
dead, we should get rid of it. If a new group's needed, we should
create it. They're totally independent.
>I would also suggest that we consider looking at Brians Bottom 40 in
>popularity to see if there are some groups that just don't deserve to
>continue to exist -- especially (and I kno this won't be popular) in the
>very low readership/high volume newsgroups.
The danger in that is that there will always be a Bottom 40. Yes,
that would be a good place to look for possible dead groups, but it
shouldn't be considered evidence of death.
--
Dave Sill (ds...@relay.nswc.navy.mil)
Brad Templeton
Good idea. There's not nearly enough debate and flamage right now
about group creation. We need to add more.
:-(
--
Brad Templeton, Looking Glass Software Ltd. -- Waterloo, Ontario 519/884-7473
Brian Matthews
|>> unaware of sufficient, current traffic on it to justify a group. (If
|>> I am wrong about this, I would probably change my opinion.)
|This is a holdover from ancient days. While I was at one time a strong
|supporter of the rule (primarily because it's difficult/impossible to
|*delete* newsgroups) I now think it's a specious argument. The net is simply
|too large these days for us to expect that all of the people who are
|interested in Lithuanian Sheepdogs to easily find each other in some random
|newsgroup to 'prove' they deserve a group.
But that's basically what has to happen for the vote to be successful -
a bunch of people have to find out somehow that there might be a Lithuanian
Sheepdog group, and if they're interested, they need to send a message to
someone. Instead of a call for votes, why not a call for articles? (Which,
if I understood correctly, is basically what Dave Sill suggests and calls
trial newsgroups.)
|It is also a rule that is inconsistently enforced.
I think there are exception because we want groups to discuss new
technologies. comp.unix.aux and comp.sys.next are two examples.
It's silly to expect there to be traffic on a subject when the
only people who have any information are required not to talk about
the subject! However, this is the exception. Lithuanian Sheepdogs and
holography have both been around quite a while, yet there's no discussion
of either.
|It's primary
|purpose was to try to hold back or limit expansion of the network
|[...]
|It's done a wonderful job at that, based on the statistics we see
|(that's a joke, son).
But there are two possibilities here:
1. Requiring volume doesn't work (which is what you Chuq suggests), or
2. We (the voting public) haven't really required volume, so we don't know
if works or not.
I think 2 is the case. In some instances the volume requirement comes up,
and some people vote based on volume or lack thereof, but not in most cases.
|The *real* problem is not creation of specious groups, but the inability to
|delete groups once they exist. I think it's time we seriously consider some
|form of sunset law -- if a group dies, make it go away. I think it's better
|for the net that we let groups be born and try to find an audience than make
|them stillborn and not see to their potential.
Agreed. If groups went away easily and cleanly, I'd almost say let anyone
create any group they wanted. (Only almost, because we still don't want
groups that duplicate existing groups or groups of questionable legality -
comp.unix.queries or rec.kiddie.porn for example (you decide which is
which :-)))
|My personal preference (which I've mentioned a few times over the years) is
|to set a cap to the number of groups on USENET.
I've never liked this. Who's to say what the cap is, and if there are n
newsgroups (n being the cap), all of which have reasonable volume, how
do we decide which group to delete (and which set of people to annoy).
If a group has reasonable traffic, it should stay, and someone wants to
create a group (assuming non-duplication and legality), it should be
created. Neither should be constrained by an arbitrary limit. Also,
if there are a bunch of groups with zero volume, why wait for new groups
to be created. Just nuke 'em.
|Inspired flamewar folks could, for instance, argue that we should create
|sci.fusion.lukewarm by deleting talk.religion.misc. So we can fix up some of
|the process while not removing the 'fun' from the newsgroup wars!
You've been around Usenet long enough Chuq to know that people don't need
an excuse for a flame war :-)
I'd say any group with a low readers/article ratio or low article count
should be notified of possible deletion. If things don't change over
the next few months in the group (there will certainly be a flurry of
activity when the notice of impending doom is posted - the question will be,
does the increased volume or readership remain increased or drop back to what
it was), the group is nuked, and can't be recreated for a certain period of
time.
Newsgroups can be created as long as the group doesn't duplicate another
group and is basically legal. And those two requirements alone should
provide plenty of flamage opportunities :-)
--
Brian L. Matthews b...@6sigma.UUCP
Six Sigma CASE, Inc. +1 206 854 6578
PO Box 40316, Bellevue, WA 98004
Dave Sill
>private, detailed critique) is that it requires the renaming or deletion of
>groups.
I meant to reply to Chuq's critique, but I was too busy and didn't get
around to it. Apparently it's not clear from my original proposal
that the trial newsgroup is *not* a real new group. I *thought* I
made it clear in my posting on this thread that that wasn't the case.
Let me make it perfectly clear: trial newsgroup discussion takes place
in an existing *.misc group, *not* a new group or a new subgroup of a
*.misc group. The whole idea of my proposal was to avoid creating new
groups until we (USENET) were sure they were warranted. What makes
the trial newsgroup workable is:
-the Trial Newsgroup Announcement, so people know it exists
-the [trial.group.name] appended to subject lines that enables
ready distinction of trial newsgroup traffic from the normal
*.misc traffic.
>Functionally and administratively, deletion of a newsgroup is a difficult
>task (trending towards impossible) on USENET for various arcane reasons.
Well please enlighten us. Until we know the problem we can't find a
solution.
>To successfully delete a group requires the physical intervention of the
>administrator on every site, plus the addition of aliases and other toys.
How is the rmgroup mechanism different from the newgroup mechanism?
>The great renaming scheme of a few years ago was
>originally brought up by me, to massive, emotionally painful flamewars
>("net.unix-wizards was good enough for my granny, it's good enough for me,
>you communist!"). I finally dropped it. A couple of years and lots of
>newsgroups/volume later, Spafford and Rick Adams regenerated it, fixed it up
>and survived a second, lesser flamewar to get it implemented.
This issue is tangential to newsgroup creation/deletion article, but
why are the sciences legitimized by their own top-level category,
while the arts are all stashed away under the recreational/hobby
category"? It sure looks like those who organized it were biased
toward the sciences.
>>It's silly to expect there to be traffic on a subject when the
>>only people who have any information are required not to talk about
>>the subject! However, this is the exception. Lithuanian Sheepdogs and
>>holography have both been around quite a while, yet there's no discussion
>>of either.
>
>I think that's a false assumption. The entire existence of the alt.all
>hierarchy proves it wrong.
I'm sorry, Chuq, but from what I've seen in my 3 months on the net the
alt hierarchy proves little, if anything. Too many sites choose not to
carry alt for various reasons, many of them quite valid. (Like not
wanting to carry alt.drugs, alt.sex{.bondage}, or any of the other
questionably legal/moral/legitimate/useful/work-related groups.)
>The reality is that trying to require existing volume effectively
>does nothing useful except give people something to argue about.
The reality is that arguing about existing volume is stupid. Either
we should require verifiable existing volume ala a successful alt.*
group or Trial Group (tm), or we should shut up about it.
>>Agreed. If groups went away easily and cleanly, I'd almost say let anyone
>>create any group they wanted.
>
>I've been arguing that position on and off since the really early days of
>USENET.
I don't believe it! You guys are saying that USENET as one giant alt
hierarchy would be fine so long as we could easily delete the dead
groups. What a mess that would be; the sum of the idiosyncrasies of
anyone that felt like creating a new group. I, for one, prefer that
some kind of consensus be reached before new groups are formed.
Granted, the current organization has its flaws, but it's certainly
better than what would have been formed out of the chaos of newgroup
stew.
>I agree. One of the things I would do would be to put a cap on the number of
>groups with the provision that if people decide that there are no excess
>groups, then you waive the cap and raise it.
Kind of a "flexible freeze"? Q: When is a cap not a cap? A: When you
can raise it whenever you want. So why bother?
>The idea is not to kill off
>good groups, but to make people aware of the fact that (1) USENET is not an
>infinite resource, and if you want your piece, you'll have to take it from
>someone who doesn't need it, and (2) that there are groups out there that
>aren't being used. [note: this is an old argument. I'm not sure the former
>point is really valid any more. Every time I think USENET has reached it's
>limitations, someone finds a new way to invent new infinite resources....]
Whether USENET, by whatever definition, is a finite resource or not is
irrelevant. It *is* a boundless resource. The resources of any given
node or link of USENET are, of course, finite. But USENET is
constantly growing in sites, links, users, average site resources,
number of viable groups, etc.
(2) Is definitely a fact. But how much of a burden are unused groups?
A couple dozen bytes and an i-node on each site? Yes, you get big
numbers if you multiply those numbers by the estimated number of
USENET sites, but such numbers are irrelevant. Aesthetically, it
would be nice to get rid of them, but I just don't see how they have
any significant negative effect on the net.
--
Dave Sill (ds...@relay.nswc.navy.mil)
Chuq Von Rospach
>a bunch of people have to find out somehow that there might be a Lithuanian
>Sheepdog group, and if they're interested, they need to send a message to
>someone. Instead of a call for votes, why not a call for articles? (Which,
>if I understood correctly, is basically what Dave Sill suggests and calls
>trial newsgroups.)
We've actually tried things like *call for articles* in the past. One time
was when we really tried to enforce a "start a mailing list, and when you
can show a readership and a message stream, talk to us..." rule. That was
how comp.text.desktop started, as a matter of fact, and it eventually
switched to moderated (by me) and eventually to unmoderated.
It is, also, a *lot* of work for one person, and it essentially
disenfranchised most of the readership. the mailing list of c.t.d was maybe
350 wehn I converted it to a newsgroup. Within 45 days it was 5,000. It's
now around 8,000.
It is also hard to police, hard to track and doesn't really solve the "is
this *real* traffic or is this *fake* traffic arguments" at all.
Another problem is that the Lithuanian Sheepdog fanciers can all get
together and 'take over' an existing group -- even something like rec.pets --
and fine that everyone else is pissed at them for doing so. So you don't get
traffic, you get flamewars. It's worse when you get the "if you don't create
talk.bizarre, we'll just go take over rec.wobegon and use *it*" type
blackmail/threat.
The primary problem with Dave Sill's proposal (of which I sent him a
private, detailed critique) is that it requires the renaming or deletion of
groups. Functionally and administratively, deletion of a newsgroup is a
difficult task (trending towards impossible) on USENET for various arcane
reasons. To successfully delete a group requires the physical intervention
of the administrator on every site, plus the addition of aliases and other
toys. So you can rmgroup all you want, and you'll still see messages flowing
from those 'dead' groups for a good six months, sometimes longer.
Believe it or not, in the time that I've been on USENET these kinds of
proposals have come up before. I've sponsored at least one. Entire
heirarchies of groups (tmp.all) as one variation. There are, frankly,
relatively few original ideas on how to run USENET. Many of them have been
hashed many times and tossed each times because they simply aren't
practical. Which is not to imply they shouldn't be brought up again -- over
time, needs and requirements change and something that isn't practical
now might work fine a year from now. People come up with new angles that make
them more workable. The great renaming scheme of a few years ago was
originally brought up by me, to massive, emotionally painful flamewars
("net.unix-wizards was good enough for my granny, it's good enough for me,
you communist!"). I finally dropped it. A couple of years and lots of
newsgroups/volume later, Spafford and Rick Adams regenerated it, fixed it up
and survived a second, lesser flamewar to get it implemented. Which is not
to say the net was stupid not to listen to me -- or that I was stupid when I
brought it up. There's a third aspect of "good idea, we're not ready yet."
It's still really difficult to delete groups. Because of this, any proposal
that requires deletion of groups and/or temporary groups has serious
administrative and technical flaws in it that have to be overcome before
it's practical. None has. I doubt any *will* until we finally deal with the
issue of how to administratively delete newsgroups in a consensual way like
we have with newgroup voting. USENET has progressed somewhat from the days
of the Nuke Wobegon flamewars (of which I still have scars....) but has
never felt it's important enough of an issue to finally settle. When it
does, it'll be time to look at temporary newsgroups. Until then, we have to
assume that once it's created, with few exceptions, it stays created.
>I think there are exception because we want groups to discuss new
>technologies. comp.unix.aux and comp.sys.next are two examples.
>It's silly to expect there to be traffic on a subject when the
>only people who have any information are required not to talk about
>the subject! However, this is the exception. Lithuanian Sheepdogs and
>holography have both been around quite a while, yet there's no discussion
>of either.
I think that's a false assumption. The entire existence of the alt.all
hierarchy proves it wrong. Almost every group there was created because
someone thought there might be enough interest to warrant the group, but
without a track record of volume to back it up. Some, like alt.sca, have
proven that you *can* create a good newsgroup with a reasonable volume out
of thin air, even when both the people and the topic have been around for a
long time -- without ever meeting on the network in the middle.
I've seen too many exceptions to this well-believed usenet 'theory' to
believe in it any more. Then again, it's also very possible to find a group
of people to vote in a group and then never use it. rec.wobegon and
comp.std.internat for two examples.
>2. We (the voting public) haven't really required volume, so we don't know
> if works or not.
>I think 2 is the case. In some instances the volume requirement comes up,
>and some people vote based on volume or lack thereof, but not in most cases.
Yes, in the past we *have* required traffic volume. I don't think it bought
us anything. Some groups (alt.sca, for one) taht could have been created and
survived happily were forced off the net, while on the other hand, other
groups that probably should have been caught (comp.std.internat) and killed
were created anyway and went on to obscurity. The reality is that trying to
require existing volume effectively does nothing useful except give people
something to argue about.
Now that we have reasonable voting criteria, we should trust them. Have
someone come up with a group proposal and it either lives or dies on the
basis of the proposal -- you can use existing volume (or lack of it) as
arguments for or against the group, but don't use lack of existing volume as
an excuse to keep it from a vote.
The fact that politicians don't have money in the treasury doesn't keep them
from voting appropriations, does it? The same sort of twisted logic applies
here.
>Agreed. If groups went away easily and cleanly, I'd almost say let anyone
>create any group they wanted.
I've been arguing that position on and off since the really early days of
USENET. It dates back to the Nuke Wobegon wars (what was that, six, seven
years ago now?). I still think we're doing it bass-ackward by making it
tough to create a group because it's impossible to delete them. The proper
way is to create a placeholder for a discussion and then delete it when that
discussion dies off. What we're trying to do instead is create a finite
number of placeholders and then wedge each discussion into the most
appropriate slot -- even when such a slot doesn't exist.
>I've never liked this. Who's to say what the cap is, and if there are n
>newsgroups (n being the cap), all of which have reasonable volume, how
>do we decide which group to delete (and which set of people to annoy).
Those are all technical issues that could be worked out -- we worked out the
voting rules over time, after all.
>If a group has reasonable traffic, it should stay,
I agree. One of the things I would do would be to put a cap on the number of
groups with the provision that if people decide that there are no excess
groups, then you waive the cap and raise it. The idea is not to kill off
good groups, but to make people aware of the fact that (1) USENET is not an
infinite resource, and if you want your piece, you'll have to take it from
someone who doesn't need it, and (2) that there are groups out there that
aren't being used. [note: this is an old argument. I'm not sure the former
point is really valid any more. Every time I think USENET has reached it's
limitations, someone finds a new way to invent new infinite resources....]
>Also,
>if there are a bunch of groups with zero volume, why wait for new groups
>to be created. Just nuke 'em.
It's been tried. It's truly amazing how emotional some people can get over a
group they don't use. "We might want to use it in a few months! honest!".
Some groups *have* been deleted over time. Very, very few. The last major
purge was during the great renaming when it was convenient to do so anyway.
The one previous to that was the Nuke Wobegon wars, of which the less said
the better.
Chuq Von Rospach
>in an existing *.misc group, *not* a new group or a new subgroup of a
>*.misc group.
No, it wasn't at all clear to me. You're proposing that we keep around some
groups just in case we want to test out a group proposal? I'm not sure I
like that -- first, people are likely to just start using them anyway, and
you end up (practically speaking) with having to have one for each major
distribution -- and then, what do you do if I want to create
comp.sys.mac.hardware at the same time someone else wants to create
comp.periphs.floptical?
>How is the rmgroup mechanism different from the newgroup mechanism?
As the software is set up, a newgroup will automatically create a group and
then send mail to the administrator telling him what it's done. A rmgroup,
on the other hand, will send mail to the administrator saying "I got a
rmgroup! Delete this!" -- forcing the administrator to take action before
anything happens. Newgroups are automatic. Rmgroups are advisory.
On many (perhaps *most* USENET systems, administration happens whenever the
administrator feels like it. For a good chunk, that means never. This means
that when you send out a rmgroup, some folks do it, some folks ignore it and
some folks never read their usenet mail anyway and don't even know it
happened. You end up with a group that is still legal on many machines that
dumps out into junk elsewhere.
To delete/rename a group correctly also requires changing the aliases so
that any messages aimed at the now dead group gets rerouted to a more likely
successor. That also requires action by the administrator.
To some degree, these problems can be mitigated by it being done on the
backbone, but you still end up with high levels of user confusion ("I posted
a question to rec.pets.sheepdog.lithuanian, and everyone is answering is in
alt.aquarium? What gives"). It also creates a bunch of discontinuous subnets
on the network where the group is still alive and well, separated by people
who have killed it off.
It takes months and months for a rmgroup to settle out. There isn't much you
can do to fix taht, either, the way USENET is designed.
>This issue is tangential to newsgroup creation/deletion article, but
>why are the sciences legitimized by their own top-level category,
>while the arts are all stashed away under the recreational/hobby
>category"? It sure looks like those who organized it were biased
>toward the sciences.
Not true. sci.all was a proposal of mine, as a matter of fact. So was the
last, unlamented net.theater, which was created and died a quiet death. Same
with the net.mystery proposal (sf-lovers for mystery readers). The reason
it's rec.arts.all instead of arts.all is really simple -- there never was
enough interest in arts.all to justify a top-level domain. It was
considered, but there aren't enough groups and there wasn't any indication
that there was going to be enough interest to make it worthwhile. I think,
looking at what is *in* rec.arts, that this decision was correct, also.
>>I think that's a false assumption. The entire existence of the alt.all
>>hierarchy proves it wrong.
>I'm sorry, Chuq, but from what I've seen in my 3 months on the net the
>alt hierarchy proves little, if anything. Too many sites choose not to
>carry alt for various reasons, many of them quite valid.
You are welcome to disagree. You also missed the classic days of alt.all.
As alt has grown, it has shifted much closer to USENET until it's frankly
hard to tell the difference any more. You missed the time when Alt really
was different.
>The reality is that arguing about existing volume is stupid. Either
>we should require verifiable existing volume ala a successful alt.*
>group or Trial Group (tm), or we should shut up about it.
Agreed. I've been trying to convince people to shut up about it for years.
It's a strawman issue.
>I don't believe it! You guys are saying that USENET as one giant alt
>hierarchy would be fine so long as we could easily delete the dead
>groups.
Yup. I don't think the network is nearly as stupid as some people seem to
believe. It self-polices itself now pretty well, overall. There will be
abuses -- there are *already* abuses. So what?
>I, for one, prefer that
>some kind of consensus be reached before new groups are formed.
I thought that way about USENET for years. I've seen the light....
>Kind of a "flexible freeze"? Q: When is a cap not a cap? A: When you
>can raise it whenever you want. So why bother?
Why Bother? It makes people think. It makes people aware of what's going on
all over USENET, not just in their little corner of the world. It raises
enlightenment. It makes people justify groups in terms of tradeoffs rather
than just "I want." -- it makes people think "This is more important to the
net than this other thing is." -- it brings up the spectre of finite
resources, something USENET is always scared about and something it never
does anything to mitigate.
>Whether USENET, by whatever definition, is a finite resource or not is
>irrelevant.
No, it *isn't* -- if USENET grows to the point where people shut it off and
the net dies, the death of USENET sure isn't irrelevant. So far, we've been
able to always find a technological fix for USENET's ills (well, most of
them). Someday that won't work.
>(2) Is definitely a fact. But how much of a burden are unused groups?
>A couple dozen bytes and an i-node on each site?
Gads, do I need to do this argument again? There is also an interface issue.
The technology is simple. What is important is the namespace and how readers
and posters interact. An empty slot in the namespace causes user confusion
and makes it harder for users to find the appropriate places to post things.
It also, IMHO, increases the likelihood of unnecessary and incorrect
cross-posting.
My interest is in an efficient name-space. That implies that every place has
its slot, that each slot is unique and doesn't overlap some other slot, and
that there aren't excess slots around making it hard to figure out what
slots we're supposed to be used. The reason to delete off dead groups isn't
technological -- it's to help the newer USENET reader who's floundering
around the namespace looking in vain for Lithuanian Sheepdog fanciers. It's
to help *that* person find the groups they want to read in an efficient
manner.
Flabby namespaces create flabby networks with high cross-postings and lots
of ambiguity. There is, for instance, almost no cross posting between
comp.sys.mac and it's children. I will say, preening slightly, that a lot of
that is because the children were very carefully designed and implemented
after great thought and feedback about appropriate names (c.s.m.programmer,
hypercard and hardware were all fostered by me, I might add, hence the
preen). Other groups have not done that kind of thought and feedback (and
self-policing) leading to ambiguous names and lots of cross-posting.
rec.music.all, for instance.
Whenever someone brings up a topic, people argue it from technological
terms. The technology of USENET is easy. The hard part is designing for the
people using it. That's what I look at. I don't *care* about the technology.
It's just a tool for the people. We should be arguing the sociology of
USENET, not whether to program in C or pascal...
Dave Sill
>>in an existing *.misc group, *not* a new group or a new subgroup of a
>>*.misc group.
>
>No, it wasn't at all clear to me. You're proposing that we keep around some
>groups just in case we want to test out a group proposal?
Last I checked, rec.misc, comp.misc, misc.misc, sci.misc and soc.misc
were already extant. Conducting a trial newsgroup in a *.misc
newsgroup makes sense since that's where the discussion on topics with
no dedicated group belongs.
>I'm not sure I
>like that -- first, people are likely to just start using them anyway, and
>you end up (practically speaking) with having to have one for each major
>distribution --
That's exactly what *.misc groups are for, Chuq.
>and then, what do you do if I want to create
>comp.sys.mac.hardware at the same time someone else wants to create
>comp.periphs.floptical?
You both share comp.misc with the rest of the net. The
[trial.news.group]-string-in-the-subject-line convention, in this case
either [comp.sys.mac.hardware] or [comp.periphs.floptical], combined
with kill files would make it easy to block out trial groups one has
no interest in. I understand a kill file could also be set up to
follow *only* the trial newsgroup traffic, too.
>>How is the rmgroup mechanism different from the newgroup mechanism?
>
>As the software is set up, a newgroup will automatically create a group and
>then send mail to the administrator telling him what it's done. A rmgroup,
>on the other hand, will send mail to the administrator saying "I got a
>rmgroup! Delete this!" -- forcing the administrator to take action before
>anything happens. Newgroups are automatic. Rmgroups are advisory.
What we're talking about here are the default settings of the macros
NORMGROUPS and NONEWGROUPS. As shipped, the news software sets the
former to "all", causing rmgroups *not* to be executed and a mail
message to be sent to the news admin. NONEWGROUPS is left undefined,
so newgroups are automatically executed. I believe there are a good
number of sites that set NONEWGROUPS to "all" because they want to
verify that newgroups are legitimate.
I don't know for sure, but I suspect rmgroups cause all spooled
articles in the removed group to be deleted. Since this is not
something we'd want to happen to, say, new.groups, it's probably not
likely that NORMGROUPS will ever be relaxed. If some asshole running
news on his Xenix PC decides to send out a newgroup for
comp.sys.eniac, it's at most annoying. Not so if he sends out random
rmgroups.
>On many (perhaps *most* USENET systems, administration happens whenever the
>administrator feels like it. For a good chunk, that means never. This means
>that when you send out a rmgroup, some folks do it, some folks ignore it and
>some folks never read their usenet mail anyway and don't even know it
>happened. You end up with a group that is still legal on many machines that
>dumps out into junk elsewhere.
Isn't there a script or something posted periodically that makes sure
the active file's got the right groups in it? That would help, but I
suppose there're many admins that don't run it.
>To some degree, these problems can be mitigated by it being done on the
>backbone, but you still end up with high levels of user confusion ("I posted
>a question to rec.pets.sheepdog.lithuanian, and everyone is answering is in
>alt.aquarium? What gives"). It also creates a bunch of discontinuous subnets
>on the network where the group is still alive and well, separated by people
>who have killed it off.
Most people, if there even know where to look for a list of groups,
check Spaf's list. Removing defunct groups from that list combined
with rmgroups on the major/active/backbone sites would, I think,
eliminate most confusion.
>The reason
>it's rec.arts.all instead of arts.all is really simple -- there never was
>enough interest in arts.all to justify a top-level domain.
Do top-level domains cost more than subdomains? So what if there are
only a half dozen arts.all groups? At least the user knows where to
look for arts-related newsgroups. "Hmmm, poetry...that would be under
`rec' wouldn't it..." Not exactly intuitive.
>>I don't believe it! You guys are saying that USENET as one giant alt
>>hierarchy would be fine so long as we could easily delete the dead
>>groups.
>
>Yup. I don't think the network is nearly as stupid as some people seem to
>believe. It self-polices itself now pretty well, overall. There will be
>abuses -- there are *already* abuses. So what?
Two heads are better than one. If ten people decide within a one-week
period to create a new group dedicated to belly-button lint, the odds
are that you'll get 5-10 different lint groups. (sci.belly-buttons.lint,
sci.lint.belly-button, rec.belly-buttons.lint, talk.belly-button-lint, etc)
>>Whether USENET, by whatever definition, is a finite resource or not is
>>irrelevant.
>
>No, it *isn't* -- if USENET grows to the point where people shut it off and
>the net dies, the death of USENET sure isn't irrelevant. So far, we've been
>able to always find a technological fix for USENET's ills (well, most of
>them). Someday that won't work.
The ol' Chicken Little Theory of USENET ("The net's collapsing! The
net's collapsing!). It's nothing more than a theory. My theory is
that USENET is no more likely to go away than the telephone system.
Sure, it could happen. But it won't, at least until there's something
better. Already, USENET proper could cease tomorrow and the
Internet/NNTP sites alone (of which ark1 is a member) would continue,
albeit with a tad less traffic. :-) As long as the hardware's out
there there will be people dedicated to making them talk to each
other. USENET isn't some freak of technology, and it's not
threatening to collapse at a moment's notice. At least that's the way
*I* see it.
>>(2) Is definitely a fact. But how much of a burden are unused groups?
>>A couple dozen bytes and an i-node on each site?
>
>Gads, do I need to do this argument again?
Apparently. I'm surprised you're surprised there are new faces on the
net since the last time you argued the point.
>There is also an interface issue.
>The technology is simple. What is important is the namespace and how readers
>and posters interact. An empty slot in the namespace causes user confusion
>and makes it harder for users to find the appropriate places to post things.
>It also, IMHO, increases the likelihood of unnecessary and incorrect
>cross-posting.
If it was purely an interface issue then wouldn't arts-related groups
belong under their own domain? Why would we care at all if a group
was active or inactive? Inactive groups would simply be placeholders
waiting for discussions to happen.
>Flabby namespaces create flabby networks with high cross-postings and lots
>of ambiguity.
"Crossposting is Evil" is right up there in my book with the Chicken
Little Theory, but I'll save that for another thread.
>[stuff about mac newsgroups deleted.]
>Other groups have not done that kind of thought and feedback (and
>self-policing) leading to ambiguous names and lots of cross-posting.
>rec.music.all, for instance.
And that's with the current voting method. If we had the anarchy you
champion how long do you suppose it would take for things to settle
down?
>Whenever someone brings up a topic, people argue it from technological
>terms. The technology of USENET is easy. The hard part is designing for the
>people using it. That's what I look at. I don't *care* about the technology.
>It's just a tool for the people. We should be arguing the sociology of
>USENET, not whether to program in C or pascal...
I agree that it's the interface that's important. Ideally, we'd
simply specify a list of topics that we wanted to follow and the
software would handle all the details. All articles would be
classified by their content only. No need for artificial
"newsgroups". Threads would weave in and out of various topics,
splitting off, perhaps joining again later. But no discussion would
ever be out of place. There'd be no need to look for existing
discussion on a topic. Rather than looking in vain for an appropriate
newsgroup to discuss something, one would simply ask the net "Is
anyone interested in Lithuanian sheepdogs?" People tracking the
keywords "Lithuania", "dog", "sheepdog", or "Lithuanian sheepdog"
would automatically see the article.
Ahhh, someday...
--
Dave Sill (ds...@relay.nswc.navy.mil)
Brian Matthews
|>Functionally and administratively, deletion of a newsgroup is a difficult
|>task (trending towards impossible) on USENET for various arcane reasons.
|Well please enlighten us. Until we know the problem we can't find a
|solution.
A couple of reasons:
1. Any time anyone suggests deleting a newsgroup, there will be people
who will whine and complain. Even if there's been no traffic in the group
in the last five years, they'll claim the group is still needed.
2. Most administrators have their news configured so rmgroup control
messages are sent to the administrator and not acted upon. This is to
prevent Joe Random Bozo from sending out an rmgroup (either accidentally
or maliciously) for a perfectly valid group, and having all of the articles
for that group removed (and with the speed of news transmission today,
on most machines in the world.) This means a group can't be incorrectly
removed without an administrators intervention, but it also means that
a group can't be correctly removed without an administrators intervention.
Historically, both of these occur and are a problem. Also, there's no
established procedure for deleting a group as there is for creating a group
(however perfect or imperfect it may be.) Hopefully one will come of
the discussions taking place now.
|How is the rmgroup mechanism different from the newgroup mechanism?
See reason 2 above. Also, newgroups usually happen automatically because
adding a group doesn't loose articles, and it's easy enough for the
admin to rmgroup the bad group locally. If an existing group is
rmgrouped, the articles in the group generally can't be recovered, so
you don't want accidental rmgroups happening too often.
|This issue is tangential to newsgroup creation/deletion article, but
|why are the sciences legitimized by their own top-level category,
|while the arts are all stashed away under the recreational/hobby
|category"? It sure looks like those who organized it were biased
|toward the sciences.
The current hierarchy and set of top level groups is based upon the
old net. hierarchy, from before the Great Renaming. The net. hierarchy
evolved, based on what groups people wanted to see, so there isn't
an arts hierarchy today because there weren't all that many arts groups
in the net. hierarchy, and this was because they weren't needed. I
don't think there is any intentional bias, it's just that the way
most of the people using the net make a living is via the sciences,
so that's where most of the heavy technical discussion occurs. Even
today I don't think there's a lot of highly technical discussion of the
various arts, just a lot of recreational discussion.
|>The reality is that trying to require existing volume effectively
|>does nothing useful except give people something to argue about.
|The reality is that arguing about existing volume is stupid. Either
|we should require verifiable existing volume ala a successful alt.*
|group or Trial Group (tm), or we should shut up about it.
Agreed. Either require proof of volume or don't even mention volume.
|>>Agreed. If groups went away easily and cleanly, I'd almost say let anyone
|>>create any group they wanted.
|>I've been arguing that position on and off since the really early days of
|>USENET.
|I don't believe it! You guys are saying that USENET as one giant alt
|hierarchy would be fine so long as we could easily delete the dead
|groups. What a mess that would be; the sum of the idiosyncrasies of
|anyone that felt like creating a new group. I, for one, prefer that
|some kind of consensus be reached before new groups are formed.
Me too, and I stated this, but a lot of context was left out. There
still needs to be consensus on things like:
1. Does the group duplicate existing group(s)?
2. Is the discussion that will take place in the group basically legal?
3. Is the name of the group proper (thanks to Greg Woods for pointing
out that I didn't include this in my earlier article. I meant to, but
just forgot.)
I agree completely that teaching everybody how to create a group and
letting them have there merry way would be a disaster. (Oh no! IDoUP :-)
|>I agree. One of the things I would do would be to put a cap on the number of
|>groups with the provision that if people decide that there are no excess
|>groups, then you waive the cap and raise it.
|Kind of a "flexible freeze"? Q: When is a cap not a cap? A: When you
|can raise it whenever you want. So why bother?
I've got to agree with Dave. I don't think a cap makes sense. If a group
has no traffic, nuke it. If someone wants to create a group (and it's
not a duplicate, etc.), they shouldn't have to cajole/browbeat/threaten
people into deleting another group. Let them create it. If no one uses
it, it gets deleted.
|(2) Is definitely a fact. But how much of a burden are unused groups?
|A couple dozen bytes and an i-node on each site? Yes, you get big
|numbers if you multiply those numbers by the estimated number of
|USENET sites, but such numbers are irrelevant.
The problem with extraneous newsgroups isn't so much a physical resources
problem (although I suspect there aren't a lot of sites that have too
much disk space or cpu time), but an organizational problem. The newsgroup
names give pointers to where to find and post articles (often the only
pointers), and if these pointers are misleading and confusing, it will make
finding and posting articles that much harder. There's a vast amount of
information available via Usenet, and the easier it is to find the
information you need, the more useful Usenet will be to everybody.
Greg Woods
>in an existing *.misc group, *not* a new group or a new subgroup of a
>*.misc group. The whole idea of my proposal was to avoid creating new
>groups until we (USENET) were sure they were warranted.
I see nothing wrong with encouraging this practice. While I don't advocate
necessarily making it a REQUIREMENT for creation of a new group, there is
no doubt that an active demonstration that there is enough interest in the
topic to sustain discussion on it is a better barometer than the surveys
we now use. Such a discussion is also likely to help generate YES votes,
both from people who want the new group, and people who want to lower the
volume in the *.misc group. Also helps eliminate NO votes from people who
don't think there is any interest (since they would already have been
proven wrong). Better still, if the discussion FAILS to generate ongoing
interest, we can avoid a vote that would fail, or worse, a successful vote
that creates an unused newsgroup.
Best of all, this can be done with NO changes in official policy and without
any control messages needing to be issued and with no action on the part of
site administrators. All it takes is an attitude change on the part of new
group proponents. Instead of posting an article to news.announce.newgroups
asking for a discussion on a new group for Lithuanian sheepdogs, the article
could instead say that a discussion of Lithuanian sheepdogs is being started
in rec.misc which MIGHT lead to the formation of rec.pets.lithuanian-sheepdogs.
Then actually post an article on the subject and SEE if anyone is really
interested in discussing it! What a novel concept! :-)
--Greg
Peter da Silva
> The net is simply
> too large these days for us to expect that all of the people who are
> interested in Lithuanian Sheepdogs to easily find each other in some random
> newsgroup to 'prove' they deserve a group.
This is only marginally true, and is true only insofar as the hierarchy has
been screwed up by bad group names. And the biggest source of bad group
proposals come from the BBS folks who want to create a group for their
favorite <whatever>. The very folks you're encouraging.
Lithuanian sheepdogs have a place on the net, in rec.pets.
> It is also a rule that is inconsistently enforced. If 'we' (the generic we)
> don't think 'you' (the generic you) deserve the newsgroup, we pull up this
> strawman as a way of stopping the creation process.
What do you mean "we", white man? I may not vote "no" against all such
inappropriate votes, but I always bring it up (at least in a letter to the
person proposing the group) even if I agree with the group.
> It's time to put this stupid strawman to rest once and for all. It's primary
> purpose was to try to hold back or limit expansion of the network -- by
> limiting groups to being better place-holders for things we were already
> doing. It's done a wonderful job at that, based on the statistics we see
> (that's a joke, son).
If it's been consistently ignored, by your own admission, how can you put
the blame for the current mess on its head?
> It's a subjective rule that is observed subjectively. Let's get rid of it.
No, let's start paying attention to it before things get worse.
> If someone comes up with a good idea for a group and can find enough people
> to vote it into service, let the group live or die on its own merits. I
> don't see *any* rational basis for a "we have no volume, so it's not worthy"
> argument. Time to let it die.
What do you mean "worthy"? A newsgroup is not a place, it's a category. A
system of classifying traffic. "A place-holder for things we were already
discussing", as you put it. It's not supposed to be created because a topic
is "worth discussing".
> The *real* problem is not creation of specious groups, but the inability to
> delete groups once they exist. I think it's time we seriously consider some
> form of sunset law -- if a group dies, make it go away.
What you're saying here is simply that a group *is* such a place-holder.
--
Peter da Silva, *NIX support guy @ Ferranti International Controls Corporation.
Biz: pe...@ficc.uu.net, +1 713 274 5180. Fun: pe...@sugar.hackercorp.com. `-_-'
"That is not the Usenet tradition, but it's a solidly-entrenched U
delusion now." -- br...@ucsd.Edu (Brian Kantor)
Peter da Silva
> Now that we have reasonable voting criteria, we should trust them. Have
> someone come up with a group proposal and it either lives or dies on the
> basis of the proposal -- you can use existing volume (or lack of it) as
> arguments for or against the group, but don't use lack of existing volume as
> an excuse to keep it from a vote.
Chuq, all that was happening was the people were using existing volume as
arguments for or against a group. Why start flaming against making volume
a requirement when nobody's suggesting that?
There's nothing in the guidelines against making any reasonable arguments,
or even electioneering (though I consider excessive electioneering slimey
and have voted against groups where it went on). And whether you agree ort
not there are still lots of folks who think existiong volume is the most
reasonable argument of all.
Dave Mack
>private, detailed critique) is that it requires the renaming or deletion of
>groups. Functionally and administratively, deletion of a newsgroup is a
>difficult task (trending towards impossible) on USENET for various arcane
>reasons. To successfully delete a group requires the physical intervention
>of the administrator on every site, plus the addition of aliases and other
>toys. So you can rmgroup all you want, and you'll still see messages flowing
>from those 'dead' groups for a good six months, sometimes longer.
Until they hit a node where the group in question has been rmgroup'd,
then they go into junk. End of the line. What's the problem? Sure,
there will be pockets of the net where a newsgroup continues to exist
for a while, but so what? Does the fact that I still have the entire
net.all hierarchy in my active file cause problems for the rest of the
net? I doubt it. Anyone who's interested in knowing which groups are
blessed and which aren't only has to take a look Spaf's monthly posting
in news.lists (or whatever.) If they can't be bothered with checking,
why should that constrain the rest of the net?
I think this obsession some of you people have with *controlling* the net
is really unhealthy. What my active file looks like is really none of your
damned business, and vice versa.
And no, I don't really have the net.all hierarchy in my active file,
but you'd be amazed at some of the alt groups I still have.
--
Dave Mack McDonnell Douglas Electronic Systems
uunet!inco!mack, inco%ma...@uunet.uu.net (703)883-3911
All opinions expressed are my own and do not reflect those of MDESC. Ever.
Chris Lewis
>Until they hit a node where the group in question has been rmgroup'd,
>then they go into junk. End of the line.
Nope. They don't stop. They keep going as long as the sys file's distribution
permits it. net.* would probably not get very far, but there's a lot
of newsgroups that do get far beyond systems that don't have specific
newsgroups... Inet groups f'r instance are always leaking out.
--
He's a consultant: | Chris Lewis, Elegant Communications Inc.
Lend him your watch | UUCP {uunet!attcan,utzoo}!lsuc!eci386!clewis
and he'll tell you the time. | Moderator of the Ferret mailing list.
Don Munroe, Cosmic Commander|
David Wright
ARTICLES' to an existing group - I think this is an excellent idea (but
note "may" not "must"). The group to use would not always be a misc group;
for the sheepdogs in question the call would presumably be for discussion
in rec.pets.
Greg Woods has supported this, saying "and it needs no change in the rules".
True, but a small change in the procedures might be in order: I think Greg
should post such calls in news.announce.newgroups if they are submitted to
him.
Re deleting old groups:
In article <56...@inco.UUCP> ma...@inco.UUCP (Dave Mack) writes:
#In article <34...@apple.Apple.COM> ch...@Apple.COM (Chuq Von Rospach) writes:
#>.... So you can rmgroup all you want, and you'll still see messages flowing
#>from those 'dead' groups for a good six months, sometimes longer.
#
#Until they hit a node where the group in question has been rmgroup'd,
#then they go into junk. End of the line. What's the problem?
The problem is that some (lots of?) news admins like me would rather have
junk newgroups than junk articles, so when someone posts something to
old-but-thought-to-be-dead group comp.lang.asm370 (to take a recent case)
which ends up in our 'junk', I run mvjunk to re-create the group for the
message. Then in a month or six someone else in some far corner of a
foreign field will not know the group had died and will post something
else, and then ....
Of course there are checkgroups messages, but here I keep so many US groups
to hold cross posting to groups that don't officially make it to Europe
that it's hard to pick out the few that really are dead from the rest.
I think the only way to kill old groups is for someone to conduct a
campaign on the issue (as is starting on news.admin), get some sort of
concensus on the dead group list, and then send out rmgroup's for that list
at regular (2 or 4 week?) intervals for the next 6-12 months. Or maybe a
monthly list of "old but now deleted" groups as a new Spaf posting?
Either way, something that would remind us admins what to remove.
Last time someone campaigned on this it finished up with a list of 3, two of
which by then had recent messages (not about the deletion), so I let them be
here. The third group was comp.lang.asm370, I think?
--
Regards, David Wright STL, London Road, Harlow, Essex CM17 9NA, UK
d...@stl.stc.co.uk <or> ...uunet!mcvax!ukc!stl!dww <or> PSI%234237100122::DWW
Living in a country without a written constitution means having to trust in
the Good Will of the Government and the Generosity of Civil Servants.
Dave Mack
>In article <56...@inco.UUCP> ma...@inco.UUCP (Dave Mack) writes:
>
>>Until they hit a node where the group in question has been rmgroup'd,
>>then they go into junk. End of the line.
>
>Nope. They don't stop. They keep going as long as the sys file's distribution
>permits it. net.* would probably not get very far, but there's a lot
>of newsgroups that do get far beyond systems that don't have specific
>newsgroups... Inet groups f'r instance are always leaking out.
Wrong, at least in news2.11 and C news. If an article is received with
no valid news groups, it goes into junk and isn't broadcast to neighboring
sites, regardless of the Distribution field. NNTP may behave differently.
Dave Mack
David C Lawrence
DM> Wrong, at least in news2.11 and C news. If an article is received with
DM> no valid news groups, it goes into junk and isn't broadcast to neighboring
DM> sites, regardless of the Distribution field. NNTP may behave differently.
NNTP doesn't behave differently because it doesn't behave on that
level at all. NNTP is just a transport mechanism; it interacts with
the existing "News Overlord Software" which makes the real decisions
about where to put articles and to whom to forward them.
Just trying to clear up a misunderstanding before it spreads too far.
Follow-ups to news.software.b which has become a general news.software
group. Perhaps it is time to really change the name.
Dave
--
(setq mail '("ta...@pawl.rpi.edu" "ta...@itsgw.rpi.edu" "ta...@rpitsmts.bitnet"))