Re: [gentoo-dev] Proctors - improve the concept or discard it?

[Chris Gianelloni]

On Wed, 2007-06-06 at 18:10 +0200, Wulf C. Krueger wrote:
> On Wednesday, June 6, 2007 05:29:47 PM Grant Goodyear wrote:
> [Proctor system]
> > a way to fix the current system, or should it be chucked entirely, as
> > has been suggested?
>
> Personally, I think we simply don't need the proctors.

As much as I was a part of the creation of the proctors, I agree.

> I'm sure they have the best intentions but I've never seen any clear
> guidelines for them. They use their best judgement what to handle and
> what not to but due to language barriers, cultural differences etc. it's
> difficult to judge.

Well, they've been asked to write guidelines for Council approval, as
well as changes to the Code of Conduct.  Neither of which have been
done.  As it stands now, there are no publicly available guidelines that
I am aware of for the proctors.

> Furthermore, where do we need them? The Forums are moderated by an, IMHO,
> excellent team. IRC is more or less self-moderated.
> That basically leaves the mailinglists and among those, the only one that
> *might* arguably need supervision could be -dev.

One thing I have started to really wonder about is this.

Why do we need the -dev mailing list?  How much real "development" (or
even discussion about it) happens on the mailing list?

Most of the traffic on this list is political in nature and simply
doesn't belong on this list.  Since we've pretty much shown over the
past couple years that the development list isn't being used properly,
why have it?

> Do we really need moderation on the list? Or could we just literally
> moderate ourselves instead? Could we try and succeed to be just ignore
> some flames instead of adding oil to the fire?

Do we really need the list?  We tried self-moderation and it simply
didn't work.  We know it won't work.  There's no point in trying again.
The situation isn't likely to change.

I mean no disrespect to people's age, but I think part of the problem
why we have such a hard time, collectively, acting like adults is we
aren't adults.  A very good number of our developers are in the high
school/college age range.  This means their life experience isn't as
high as a more seasoned adult.  They have no real experiences dealing
with adults in adult situations.  They're simply used to how things are
done with people their age.  It isn't their fault, it is just simply a
lack of life experience.  We simply cannot reasonably expect everyone to
act like a level-headed thirty year old computer professional.  I have
heard people say that our lack of being paid developers compounds this,
as we have people from all walks of life.  I don't think that I believe
that, but I do know that paid developers tend to be older and more
professional.  After all, if they constantly acted like a tool, they'd
be fired.

> And even if we can't: We still have DevRel we can complain to. Yes, DevRel
> is for inter-developer conflicts but let's look back in the archives a
> bit - do we really need more than that? Most conflicts arise between
> active developers and, well, one active retired-dev.

Developer Relations has gone through a few good spots intermixed with
lots of failures.  They keep improving, but the trust level many
developers have with Developer Relations isn't very good.  With the
recent changes within the group, we might see improvement here, and I
think that we will.  I don't mean this to sound like I am throwing
devrel under the bus or anything.  I am not.  I know that those guys
work hard.  However, good intentions and hard work don't necessarily
make up for failing to attain goals.  Part of the problem has been the
fear that Developer Relations has rightly had in using their powers.  I
have always felt that a properly-running distribution should have the
need for a group whose purpose is to resolve internal conflict.  We will
always need recruiters, but the existence of a group just to make the
300 or so of us play nice together shows that our culture is broken.

> Do we really need an entire team for dealing with one former dev in case
> he goes too far? Or could we just agree to ignore him if he again behaves
> inappropriately (or what some of us *feel* might be inappropriate)?

Developer Relations does a bunch more than just deal with problems with
Gentoo developers and Ciaran.  If it really were just Ciaran that was
the cause (or catalyst) of all of our problems, it could be solved very
simply.  It isn't.

Developers simply aren't going to agree all the time.  No matter what,
there will end up being some group responsible for trying to resolve
interpersonal issues.  In companies, that would be the Human Resources
department.  When you think of devrel more as an HR department, you
realize there's more to it than dealing with problems.  After all,
Developer Relations does all the recruiting work.

> When I first read the CoC I had just read about the entire Ciaran-incident
> on the respective bugs, Forums, mailinglists, blogs and many other
> sources. CoC, while not bad in itself, seemed (and still seems) to me
> like a "Lex Ciaran" - a document with that what I had just read clearly
> in mind and targetted at preventing it.

The Code of Conduct was written with the hopes that its existence would
help to curb the flamewars and other general nastiness between people
within the community.  The proctors were created to enforce the Code of
Conduct.  Their mandate was to be very fast moving and to try to keep
flames from spreading.  For some time, I was working with the proctors.
I ended up disliking the bureaucratic direction they were taking and
chose to have myself removed from the group.  Since that time, I have
pretty much felt that the proctors *have* taken it upon themselves to
single out and target particular individuals.  Whether this was
intentional or not is really beside the point.  The perception is all
that really matters, as it is all that gets propagated to the world.  I
think this is something that people seem to forget.  It doesn't matter
what the real truth is for anything.  All that matters "to the world" is
what they perceive.  If the perception is that Gentoo is nothing but a
bunch of guys waiting to flame people, it doesn't matter that there
might be 98% of the developer pool that has never engaged in a flamewar.
(Numbers completely made up...)

> While preventing it is a good goal in itself, writing a CoC based on an
> actual case which has only recently occurred, usually leads to this
> result and damages the whatever good intentions were involved because
> other people will see the similarities as well.

The Code of Conduct wasn't written in response to a particular case.
The timing suggests that it was written against Ciaran.  It wasn't.  I
know this will sound a bit harsh, but if we really were trying to just
get rid of Ciaran, we would have just banned him and been done with it.
There wouldn't have been a point in creating yet another project and
staffing it.  The goal *was* and still *is* to reduce the flames, no
matter what parties are involved.

> More than that, it puts a strain on those who are entrusted with enforcing
> the CoC because they will try, with the best motives, to prevent anything
> like that happening again. And they will do it, as the proctors stated
> themselves, pro-actively.

No, re-actively.  If it were proactive, it would be done before the
flames started.  The proctors *have* tried to react as quickly as
possible.  The problem is that there are no published guidelines, and
decisions from the proctors are completely arbitrary to any outside
observer.  I think they've failed.  Again, I don't think that the guys
didn't have the best intentions, and I know that some people took my
voicing of their failure as a direct personal assault.  It wasn't meant
that way, but I'm not going to apologize for my observations.  I see no
point in apologizing for what *I* perceived, even if it does hurt a few
feelings.  I just think people are being overly-sensitive.  It's
Gentoo's curse.

> The problem is, though: In an asynchronous communications medium, you
> simply cannot pro-actively do anything without bordering on what some
> like to call censorship. You can only *re*act in such a situation.
>
> Even *trying* to act pro-actively will lead to unrest as we've only very
> recently seen it. If we accept my hypothesis of asynchronous
> communication and the implications I described, we come to the conclusion
> that reaction is the most likely way not to open Pandora's Box.

Attempts to become more proactive were dismissed.  One such attempt was
to enforce bans on all mediums.  For example, if someone is banned for
24 hours for their actions on IRC, they should be banned from all of our
media.  Why?  Because there's nothing keeping the person from just
moving "next door" and starting more problems.  We've even seen it
happen in at least one occasion that I am aware of with this list and
the forums.

> That leads back to DevRel. We have them to deal reactively with conflicts
> after a complaint by either party involved. I stated, that on the
> mailinglists, we mainly see inter-developer conflicts and those can be
> handled by DevRel.

I think we mostly see developer<->user conflict.  That was one of the
reasons we created a new group to monitor our media.  We felt that
Developer Relations was *not* the group, since it dealt only with
inter-developer communications and we needed something broader.

> A small improvement to DevRel might be achieved, at least from what I've
> seen by reading lots and lots of DevRel bugs, by taking action on
> unfounded complaints, too. I'm speaking of trivial complaints, of course.

If Developer Relations were able to act fast, it would help immensely.
Again, we have tied ourselves in so much red tape and procedure, that
getting things done is now secondary to following protocol.  I am not
pointing fingers at devrel on this.  I think it is a failure across
most, if not all, of Gentoo.

> If, after both sides were investigated properly, the complaining party is
> found to be exaggerating or too easily offended, disciplinary action
> should be taken against it. Of course, this should be done light-handedly
> but it should give the complaining party some time to learn from their
> mistake. Maybe this is what's already intended - it's just that I haven't
> found any examples. :)

It is actually what was intended.  The problem is that even the most
light-handed actions have been met with resignations, flames, people
being general assholes, and all kinds of other fun things that compound
the problems rather than resolve them.  I know that one of kloeri's
biggest fears as creating more problems than he solved, every time
devrel had to do just about anything.

We are an open source project that is completely community-based.  We
simply don't all think alike and can't expect that to ever change.

> I apologise for the long mail but I wanted to state clearly and without
> too much emotions why I think we don't need the proctors and why we
> should thank them for attempting to bring some order to the chaos and
> give up on the concept as a whole.

We don't really have any sort of replacement for the proctors.  The
original User Relations was supposed to do that job, but that was before
Christel came around and reinvented userrel as a great
community<->developer gateway.  The problem was that we *needed* to have
the "old" userrel to compliment devrel.  The proctors were supposed to
fill in that gap.

I know I am planning on bringing up discussion on this at the next
Council meeting and we'll simply go from there.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation