We can halve volume by not allowing nondevelopers to post
Several people have suggested solutions like creating debian-discuss
for nontechnical discussions, or moving some threads to -policy, or
the like.
These things are not the answer to the problem. I stated the problem
as `volume on -devel', but really the problem is that there are too
many messages in the fora that we'd like most Debian developers to
subscribe to.
At the moment many developers don't read -devel, or read it only very
quickly, because it's too large. Splitting it won't help: instead,
there will be just two lists in the situation where -devel is now:
too much traffic to be worth reading. Splitting is only helpful if
significantly different groups of people would want to subscribe to
the two lists.
It may be that creating a new (pair of) list(s) for discussion of
problems in packages in unstable or frozen would be useful; the
traffic there is something that many people wouldn't need or want to
read.
Trying to move things out of -devel to -policy (other than things that
-policy is currently trying to do) will just make it hard to use
-policy for what it's currently good at.
It's also not clear to me that there is a very high level of off-topic
posting. -devel has always been traditionally the "developers' forum"
- not the one "for development", and has always contained political,
organisational and other material.
However, there is an awful lot of posting, and posting begets more
postings. Particularly, emotive issues like the one about
`censorship' induce people to post whenever they have the smallest
thing to say and each article begets more responses.
Now, on to my Subject line:
Thanks to whoever it was who counted postings.
Reducing the traffic in -devel by a quarter to a third by excluding
nondeveloper postings would IMO be a clearly good thing.
However, I think the effect would be even more dramatic, because of
the effect that one posting generates more postings. If nondevelopers
didn't post, we wouldn't have developers following up to them because
they felt impelled to disagree.
Suppose we assume that every developer followup to a nondeveloper's
posting would not have been posted if the precursor hadn't been
posted, what kind of numbers do we come up with for the reducition in
volume ? It would be interesting to count the number of postings that
were from nondevelopers _or followups directly to articles by
nondevelopers_.
For a rough calculation, if we neglect original articles, and assume
there is no correlation in developerness between follower-up and
followee-up, we find that (1-438/1596)^2 = 52.6% of postings remain.
Thus, neglecting third-level effect, where nondeveloper postings
cause a chain of followups two or more deep, we could HALVE the volume
in -devel merely by closing it to nondeveloper postings !
This, together with the unstable/frozen lists, might put -devel back
where nearly everyone could read it.
Ian.
Reply to: