[news.software.b] Some documentation on the broken references line problem

brad@looking.on.ca (Brad Templeton) (11/18/89)

Some people asked why it mattered if only 6% of sites were putting out
articles with broken references lines.   To find out, I sorted the articles
in 'news.groups' (a busy group with lots of big threads) by the References
line.

What I found out was that there were 106 distinct trees of articles in
news.groups for that period.

BUT, only 75 had distinct subjects.  31 were fragments of original trees
that had been broken off due to broken references lines.

In addition, several of the 'distinct' subjects were only slightly edited
subjects that fooled sort -u.  I

And a whole *bunch* of the threads were really part of the 'aquaria' thread,
deliberately broken off by people who either used postnews instead of
a followup with a different subject, or who deliberately cut the references
line.  Here are the Subjects that were found in more than one message tree,
along with how many trees had that subject:

  10 +-Results of sci.aquaria vote
   4 +-The disservice of pushing for sci.aquaria
   4 +-Some observations on this whole mess.
   4 +-Give it up, folks
   3 +-Images Group
   3 +-Discussion: talk.religion.pagan
   3 +-A proposal for a new voting scheme
   2 +-call for discussion: renaming comp.emacs to comp.editors.emacs
   2 +-Will not allow sci.aquaria to exist at my site.
   2 +-Sci.ad.nauseum.aquaria redux
   2 +-Pierless Punditry
   2 +-Moderation
   2 +-Call for discussion -- comp.binaries.usenet-ps-maps
   2 +-Call for discussion - comp.sys.ncr
   2 +-A Few Observations

Several of the 10 "Results of sci.aquaria" vote trees seemed to come from
Richard Sexton himself.

The result -- a small number of sites breaking reference lines results in
lots of broken trees -- at least 35% extra trees.

Making the References line useful, but unfortunately very sloppy-looking.
If you had a tree-kill mechanism, for example, as I do, you would have
had to kill 'Results of sci.aquaria vote' at *least* 10 times before it
finally went away.   Or resort to the kludge of killing the subject as RN
does.

References went into news years ago.  Isn't it time it worked?
-- 
Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473