Στις 03-09-2010, ημέρα Παρ, και ώρα 05:34 -0700, ο/η Conrad Irwin
έγραψε:
On 3 September 2010 00:51, Danese Cooper
<dcooper(a)wikimedia.org> wrote:
1. Eliminate single points of failure / bottlenecks
2. Reconfigure into teams designed to encourage faster (shorter
duration) and more accurate projects / deployments
3. Develop programs to encourage / grow volunteers into paid
staff...recognizing that as a non-profit WMF needs to plan for more
frequent turnover in the Tech department to ensure that we can grow
expertise across the dev community rather than concentrating it in a few
core people.
4. Restore the balance between community-led and foundation-led efforts
so WMF feels again like a partnership rather than concentric circles of
permission / blame
I really like item 2. When I was working on MediaWiki related stuff,
the main problem I had was working out who to talk to. The answer, if
you ask, seems always to be "Tim" — which is not very scalable
(brilliant though Tim is). I'd hope that along with an organisation
into more formal teams would come a structure where individuals are
responsible (either permanently, or on rotation like the chromium) for
subsets of MediaWiki/Wikimedia.
I think of the "who to talk to" issue (which in part reduces to a "who
can officially sign off on your code so it may be deployed" issue) as
#1, bottlenecks. Over the past few years that I have been involved in
various ways in the Wikimedia projects, I have seen people give up after
waiting a couple of years for their code to make to the deployable
stage. This in term impacts both #3 (getting/attracting volunteers who
can later turn into paid staff) and #4 (dynamics between WMF and the
community, to the extent that WMF staff are not considered community
members.
It sort of puts a new spin on the old "If you want it done, write it
yourself because we are too busy" line. When writing it yourself (and
being willing to revise it until it's considered good) isn't available
as an option any more, an open source project is rather less open.
I know that is by no means the only bottleneck we have; it's just the
one that looms large in my mind at the moment.
This would imply that the structure used by the
Usability Initiative
is something we want to emulate — a few tight-knit teams that can
focus on specific concerns, containing people with the power to say
"yes" or "no" to particular features/ideas. Having a person
responsible for saying "no" is essential; as Danese says, you can't
accept every random patch or idea that gets thrown your way, and
leaving things languishing forever is less kind than just saying no
(ideally with a reason). Hiding behind team decisions is impersonal to
the point of rudeness — if I'm committing into your area of interest,
I am part of your team.
Having teams responsible (and with authority) for different areas, able
to approve features and code, sounds fabulous. I know we are moving
towards broader team structure already. Ideally there might be more than
one person on each team who could give the thumbs up/down.
The other advantage of having teams is that it makes
it more explicit
which areas of development are being focussed on, and by whom.
Wikimedia obviously concentrates a reasonable amount on fundraising,
which is essential as a means, but it doesn't achieve anything
directly. I'd hope that having some kind of explicit structure would
expose any less obvious blind-spots — my personal gripe is that no
time gets spent on Wiktionary (cf. bug 20246).
Clearly there are downsides, cliques are likely to form. I'd certainly
regard it as a failure of the system if it stopped someone from doing
something merely because they were currently on the wrong "team" —
there's not much point in keeping lists of team members, perhaps all
that is needed is a team leader (responsible for accepting or refusing
changes/ideas), and the team is simple those who are talking to him.
In case it's not obvious, I would not make team leaders exclusively
employees, but use the meritocracy previously described that would
only make it likely that most of them would be.
The other sentiment I very much agree with is that more communication
should be public — for the simple reason that if I don't know what
you're doing, I can't help. Keeping track (however loosely) of what's
being worked on is much more efficient than trying to second guess. If
the channels we have are too noisy, it's easy to split them by topic.
I like the idea of making #mediawiki the support channel, and having
#mediawiki-dev for development — this is a model used by lots of
projects on freenode. We could even have #mediawiki-offtopic too, if
people want to do a lot of misc chatting. Being able to talk to other
developers is very useful for getting things done, whereas having some
developers who can't be contacted at all by others is very troublesome
indeed. I had assumed that it was a requirement for a wikimedia staff
member to be somewhere public on freenode — clearly I was mistaken.
I'm one of those who posts almost never to this list (in fact almost
never to any list, I have an allergy :-P). However I live in IRC in
public channels where anyone can and does find me. Between those two
sets of tools, I would hope that everyone could be similarly available
and that their development processes would be accessible for public
participation.
A brief comment about closed mailing lists, since this has come up a few
times:
We have a mailing list that is typically used for operations matters.
I'm not sure how much of that *has* to be private but I am sure some of
it does: things like Sun contract numbers and invoices don't belong on
a public mailing list. If we were to close it, we would simply have to
find another way (less convenient I suppose) to communicate that same
information.
AFAIK we don't have a s3krit mailing list for development. (Now someone
will prove me wrong by revealing the existence of wmfdevcabal-l. :-P)
I hope that items 1, 3 and 4 will become much easier
if we solve 2
correctly, and communications will always be improvable. Good luck
with your plans, and please keep us updated — I'd much rather have to
add you to my spam filter than to not know what's happening.
We hope to produce enough quality spam to keep you and everyone else
happy :-)
Ariel
Conrad
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l