[TYPES] meta thread on types-announces filtering of announces

[Gabriel Scherer]

Dear types list,

A new thread on academic announces moderation on coq-club
( https://sympa.inria.fr/sympa/arc/coq-club/2018-06/msg00104.html )
reminded me that some may be interested in discussing the
types-announce moderation policy. (I apologize for the people in
types-discuss who do not subscribe to types-announce for the
noise. I don't think there is much to discuss about types-list
moderation, as essentially every message is interesting and is
accepted.)

(Reminder: I am the main moderator of both lists since November 2017
(see the full list of moderators at
https://lists.seas.upenn.edu/mailman/listinfo/types-list ), so I was
the one filtering almost all announces sent through the list.)

The current types-announce policy is roughly:
- off-topic announces are not transmitted, everything that seems
  relevant to type theory, programming language or formal verification
  is kept (one of those themes has to be of significant importance
  in the announce, not one checkbox among a bazillion hot topics)
- calls for submissions are more important than calls for
  participations (except in no-submissions events like summer school);
  only one call-for-participation for conference, and none for
  workshops, if they have sent a call for submission on the list
- position announces are transmitted when they are in-topic (or when
  they are generic but the sender comes from an in-topic community,
  implying that the topic will be considered), no judgment on the
  importance of the position or the announce is attempted.

Some numbers. There were 442 postings in 2018, about 2.5 per day. 2200
people receive each types-announce message, and 300 get daily digests.

I understand that people prefer to receive less emails. Here are some
options I have considered, but currently there is none that I strongly
prefer to the statu quo, so the plan is to roughly keep the current
policy.

1. I could use a more subjective policy, letting through only
   announces that pass the filter above *and* that I find
   interesting. However, based on sampling the archives, this would
   not remove more than 5-10% of the announces, for a decrease in
   predictability and fairness.

2. We could stop propagating announcements for positions/jobs, which
   tend to have a smaller potentially-interested audience than
   international events. On the other hand, it does seem useful to
   have a place to publicize positions.
   This would filter out 10-15% of the announces.

3. We could only send one announce per event, instead of allowing
   re-sends. I have mixed feelings about this idea. In particular,
   I think that it is not a good idea to reject announces for deadline
   extensions: if someone learned of an event from our list, they would
   probably expect to learn of the extension from our list as well, and
   they may miss it if we don't propagate it.
   A lot of the events multi-announce (with at least a month delay in between),
   so this could decrease the traffic a bit more, maybe 30%.

4. We could restrict announcements of workshops that are affiliated
   with conferences, systematically asking for joint workshop calls
   instead of individual calls. We had a lot of traffic for the FLoC
   2018 workshops, for example -- for the Calls for Participation,
   I decided to only let the joint call through.
   Workshops are 20% of our traffic, but not all of them are co-hosted.

5. I have had discussions with other moderators about splitting the
   list in smaller, more focused lists, for example one list
   "types -- and software verification" another "types -- and logic".
   Personally I think it's pretty clear on which sides most announces
   would fall, so that someone subscribed to a single of those list
   could expect a good 40% traffic reduction. But it's not clear
   how to organize such a transition, and how many people would not
   still subscribe to both lists.

I'm open to other suggestions for filters -- they have to be easily
implementable by a human, and justifiable to people whose announces
get rejected. It is useful, to evaluate a proposition, to use the list
archives ( all 2018 annunces are at
http://lists.seas.upenn.edu/pipermail/types-announce/2018/thread.html )
to get a quantitative estimation for how much of the traffic you would
filter out.

Any other comment on the moderation is also welcome is welcome --
please consider whether you should send it to the whole list or just me.

Cheers