In article <171 at zebra.UUCP> vern at zebra.UUCP (Vernon C. Hoxie) writes:
>In article <674 at whizz.uucp>, I wrote:
>> Over the past several months, I've become increasingly aware of the huge
>> percentage of articles which cross-post between comp.sys.att and unix-pc.*.
>> When I get into comp.sys.att (second in order in my .newsrc), I find
>> myself virtually wearing out the 'n' key. (An intuitive guess is that fully
>> 70% of the comp.sys.att traffic that I see arriving here is cross-posted
>> stuff from unix-pc.*).
I was being most conservative in stating 70%.
> The reason you have to pour over so many unix-pc postings is
>that these groups are poorly distributed and in order to get to many of
>us, the posters are kind enough to cross post. In fact, I can't post
>except through comp.sys.att.
That was the main point - alot of people CANNOT GET the unix-pc groups in their
local area. Having to set up a special feed in order to get it is plain damn
nonsense when it constitutes the majority volume of another group (att) that
is routinely carried as a standard feed. Likewise, some people CANNOT POST to
the unix-pc groups, having only comp.sys.att provided locally.
For example, I have two feeds for unix-pc groups: one of them gets his by a
direct connection to *California*, the other by a connection to a machine in
*Texas* (if we haven't cross-fed to him already).
We are, please observe, in *******Oregon*********!! (For those of you from the
East Coast, you may want to drag out your Hammond's in order to understand just
what kind of distances are involved here: If you happen to live in the
D.C. area, for example, this would be about comparable to calling Chicago or
Texas, respectively, from there).
> I also would like to know why distribution of the unix-pc group
>is so difficult to come by. I have been trying for the last year to get
>it here. Assuredly, someone in the Denver area does receive it
>since 'boulder' is a backbone site.
I wouldn't count on it. Logic won't get you very far, based one what I've
observed.
Hey, no problem - I'll be glad to give you a feed. According to the idea we
seem to live by, long distance from Portland (Oregon) to Denver is no big
deal (about the same as routinely calling Miami to get *your* unix-pc stuff,
if you live in New York City :-).
> The result is that people like you, who have no interest in the
>unix-pc, but are interested in other discussions on comp.sys.att have to
>grope through our activities.
Umm, well, actually, my machine is a 3B1. But the point I was making is the
same, either way: the redundancy is almost painfully absurd, generating
consequences that are wholey unnecessary.
> I don't know why the distribution system falls down when the
>administrators need only add 'unix-pc,' to the '/usr/lib/news/sys' file.
>Surely this is not a heavy load when they do forward so many "junk" groups.
Yes - I believe that traffic level should dictate whether a group is to be
carried on the standard feed. If, for example, the unix-pc stuff were *not*
carried in the comp.sys.att group, unix-pc would be carried as a standard
feed and 'att' would be an 'alt' group, since (absent the unix-pc stuff) it
would be carrying so little traffic of it's own.
And I agree - if the choice is between talk.valspeak.bizarre and our group,
that's a no-brainer. Unless, of course, t.v.b has alot of traffic :-).
> I have been wondering about suggesting that the group names be
>changed so that it starts with 'comp.xxxx'. Since this is already an
Precisely. But probably too easy a solution.
>entry in the '/usr/lib/news/sys' file, we could all benefit. Perhaps even
>some of our European friends could then join in on the discussions.
>That is, if the full 'comp.' distribution is carried over there.
When I hear people defend the current practice of treating the upc group as
an elective appendage to the net distribution, I am reminded of the old line:
"We won't get wet - I've got a raincoat!". If it isn't a problem for them, then
it simply doesn't warrant change.
Someone recently pointed out that computer-industry 'standards' are defined
not by standards-committees, but by market-share. Why shouldn't this logic
hold for defining usenet groups? Can someone summon any reasonable argument
why a group that is official should get preferential distribution even when
it generates almost no output? And unofficial groups that are weighing in
consistently with solid traffic should be ignored?
For example, my current .newsrc file gives:
unix-pc.bugs: 1-17
unix-pc.general: 1-1106
unix-pc.sources: 1-108
unix-pc.test: 1-10
unix-pc.uucp: 1-42
unix-pc.misc: (this one was lost for awhile :-)
-----------------------------------------------
Total unix-pc: 1-1283
comp.sys.att: 1-1393
This yields 92% potential cross-posting (not scientific, but you get the idea).
In any case, unix-pc, order of magnitude, is *as large* as comp.sys.att, even
*including* the cross-postings! That is to say that without unix-pc, little
else may remain in att, AT ALL!
I'll bet that our tiny band of trusty devotees to the Unix PC are a hell of a
lot more-active than some other groups that are mindlessly passed on across
the net every day, and won't ever create enough total volume to wad a shotgun.
^^^^
I propose that there be two surviving groups, with another a casualty:
comp.sys.upc - all unix-pc, sans other att stuff.
comp.sys.att - give this empty artifact a quiet burial unless AT&T is willing
to put on staffers to write stuff to post in order to keep up
the ol' corporate image. Fact is, little else happens here.
comp.sys.misc - move any tailings of 'att' to the misc group, which itself
isn't exactly a barn-burner. Together, they might be able to
form a combined entity of sufficient size to propogate :-).
Wouldn't there be a certain ripe justice if the Unix PC were sticking it up
AT&T's all-too-capacious orifice, rather than vice-versa? The worm turns :-)
Bud Hovell
USENET: ...!{sun!nosun|tektronix!percival}!whizz!{bbh|postmaster|sysadmin}
USPO: McCormick & Hovell, Inc., PO Box 1812, Lake Oswego, OR USA 97035
MOTD: "Vote NO!"