Re: Policy for expiring lists WAS: Idea for a secondary list server

On 03/02/2015 07:37 PM, Josh Berkus wrote:
> On 02/28/2015 08:52 AM, Stefan Kaltenbrunner wrote:
>> On 02/27/2015 01:36 AM, Josh Berkus wrote:
>>> On 02/24/2015 01:32 PM, Alvaro Herrera wrote:
>>>> Josh Berkus wrote:
>
>> all nice and clear - but who is going to check whether lists fall under
>> that termination rule or not on a regular base? Are you volunteering?
>
> I don't have the permissions. And this is a task better automated, no?

well your proposal specifically included the requirement that an admin
or the list owner sends a "termination notice" - not sure we want to
fully automate that?

>
>>
>> Also to put some numbers to this - we currently have 35 PUG lists, out
>> of those only 5(6 if the cutoff would be 10) would make the cut per the
>> above rules and only 13(!) had more than a single mail in all of 2014.
>> There have been only 268 mails in total over all of those 35 lists.
>
> Good, we can immediately clean some stuff out then. That's what we
> want, correct?
>
> Also: does that include the non-English lists? If so, I'm surprised.

well those numbers are for all lists in the PUG section:

http://www.postgresql.org/list/group/6/ (just click into some of the
archives to see the sad state of things for yourself if you dont believe me)

My argument was mostly that with SFPUG removed (which is part of the "5"
lists that are make at least some traffic) we would be down to 4.
Based on those numbers I find it very hard to advocate for making list
addition easier and even paying money for some secondary service because
frankly - nobody actually seems to use them :(
Right now I think the mailinglists are actually doing us a disservice
because the only impression a new visitor can get from looking at the
archives of most of them is "they are dead".

>
>>
>>>
>>> One question is: for terminated lists, what is our policy/practice on
>>> archives?
>>
>> dont think we have one
>
> Right, I posted here so that we would *create* one. I'm not in a
> position to *propose* one because I don't know enough about the
> infrastructure of our archives.

well the actual archive can just keep all the mails - so it is more of a
policy decision of whether we want to keep displaying them (or make them
searchable or whatever)

>
>>> For example, I'd like to terminate the SFPUG list. Given that we have
>>> Meetup, RSS *and* Twitter, we really don't need it anymore. However,
>>> I'm reluctant to delete the archives.
>>
>> uh isnt that actually _the_ prime example why we dont actually need more
>> lists? If a PUG as large and successful as SFPUG does not need one
>> because there are better ways to coordinate a PUG and make it successful
>> why are we not promoting those?
>
> No, it's not. We need the mailing lists for PUGs which are not
> successful *yet*. The reason that SFPUG has outgrown its mailing list
> is that we have other resources:
>
> * Meetup.com, paid for by David for $140/year.
> * Website/RSS feed, paid for by me, $75/year.
> * Twitter acct., registered by me.
>
> Prior to having those things and getting big, our mailing list was
> *essential* for the success of SFPUG; we used nothing else for the first
> 6 years we existed. There's also the fact that most of our local folks
> are very active on the main lists (-hackers, general, etc.) which is the
> reason why we don't see a lot of discussion and peer-to-peer help on
> SFPUG anymore.

interesting - so why is it that only 1/8th(or 15%) of the lists we
created so far seem to be able to sustain even a minimum of traffic?

>
> This is not a solution for new PUGs because:
>
> * We're not proposing to pay for Meetup accounts for every new PUG.
> * Meetup is only used in certain cities in the US/Canada, and is unknown
> elsewhere. It's also not multilingual.
> * Some people hate Meetup because it's proprietary and external.
> * We're not proposing to offer website space to all new PUGs.
> * Twitter is not a substitute for other ways of contact.
>
> Now, if you said to me "Josh, I think we should be offering those other
> things *instead of* mailing lists to new PUGs," then I think that would
> be a possibility worth discussing. I didn't think the infra team was
> eager to offer hosted websites for PUGs again, though.

no we are not eager to host websites for PUGs but I also dont think that
it is good to run mailinglists just for the sake of it, with the
sysadmin team hat on I think we have much more important tasks than
running a listserver nobody actually uses (and running mail crap is not
_cheap_ from a manpower pov) and with a community hat on I would find it
weird for us to spend money on a service nobody will use.

>
> Also, I'm only suggesting that we could terminate the SFPUG list *if* I
> can re-create the list later if something happens to Meetup. If I can't
> get the list back once it's gone (which has been the de-facto policy in
> the past), then I'm going to fight to hold onto it.

hmm do you have any reference where reinstating an formerly existing
list was refused?

>
> Further, we're not just talking about PUGs; people will also request new
> mailing lists for other purposes, such as specific development projects.
> For example, I could see that a pgsql-fpga@ or mailing list could be
> useful at this point, except that the involved developers haven't asked
> for it. Certainly if such a list were requested, we wouldn't tell them
> to use Meetup.

we already had a few such topic/development related lists (like
pgsql-cluster-hackers which basically boils down to a meeting
coordination list for the meeting at pgcon) so no sure those are going
to succeed either.

>
> And ... to be blunt, I really feel like a bunch of folks on this thread
> are discussing this in bad faith; that is, they're looking for any
> excuse at all to say no to new mailing lists, regardless of other
> considerations. If that's the case, then please say so now, we can see
> if this is a majority of the infra team, and if it is, I'll stop wasting
> my time.

to be blunt as well I find this an offence - some of us are spending a
_significant_ amount of time on infrastructure stuff, and this is not at
all saying "no to mailinglists", but this is saying that there is no
point in creating additional systems that will have a very very low
level of usage (and the numbers we have on the current lists are imho
proof for that).

>
>> Per the above numbers it is obvious that we dont need more mailing lists
>> but what we need is a solid set of recommendations on how to run a
>> successful pug and what tools to use for that (wiki?).
>>
>>>
>>> A second question: what about reactivating lists?
>>>
>>> Example: SLCPUG stops meeting and their list goes dead. We terminate
>>> the list but keep the archives. Two years later, a new community member
>>> wants to re-organize SLCPUG. Do we have a way to give them a list which
>>> will archive to the same place?
>>
>> dont think there is any technical issue with doing it that way but again
>> I have some doubts that it is actually needed at all.
>
> This is *exactly* what I'm talking about. The reason the infra team
> runs up against a wall on terminating lists is that list members know
> that once a list has been killed, the infra team will throw every
> possible sandbag in the way of reviving it. If you guys were more
> flexible about reviving lists, then the owners would be more flexible
> about terminating them. You're digging your own hole, here.

again not sure why you are picking on the sysadmin team here - what we
should be focusing on is that:

* the overall mailinglist traffic on all our lists is declining
* 85% of the current PUG lists are dead by your own definition
* the most recently requested and granted mailinglists basically were
never used:
* pgmke (created half a year ago - only a testmail no real traffic)
* seasiapug (create Nov 2013 - not a single mail yet)
* triangle-nc-pug (created january 2014 only a single "created" mail
from alvaro)

If we want help our PUgs in growing we certainly need to provide
something, but from all the numbers I see it does not seem that
mailinglists is what helps them. This feels a bit like picking on people
spelling out the fact that times have changed and what made us
successful as a community 20y ago might not be the thing we "really"
need today.

Stefan