Re: [gentoo-dev] Democracy: No silver bullet

[Stuart Herbert]

On 8/24/06, Donnie Berkholz <dberkholz@g.o> wrote:
> A distribution is more than just an entity that packages upstream
> tarballs. I agree with your point, but it misses a large chunk of what
> we do.

We do more than that, sure, but the vast majority of the day to day
work in Gentoo is exactly that - packaging software released by
upstream, and fixing bugs reported back from users.

What do you do that goes beyond this?

> If this is the Gentoo vision, then why are we even doing anything else?

Because folks want to?  Because we've been recruiting people to
shoulder the load, instead of recruiting them into a culture?  Because
we want to see Gentoo run on a wider variety of hardware than
$upstream has access to?  Because we want to make Gentoo more
accessible to folks than it was in the past?

What activities are we doing that don't directly support the Gentoo vision?

> We've already reached our only goal, which is packaging stuff, and all
> we need to do is bump it.
>
> People need to feel that Gentoo is _moving forward_, that it's actually
> going somewhere.

We have no organisation that's going out there making deals with
commercial entities, ISV partners, nor users.  In that respect, we're
a completely different beast to RedHat, SuSE and Ubuntu.

You're not the first, and you won't be the last, to complain that
we're not going anywhere.  My question is simple : where do folks want
to go, and what is stopping them getting there?  Seriously - what
exactly is this enormous brick wall that folks need a boost from
management to climb over?

> Then why wasn't the hierarchy fixed? Instead we somehow ended up in this
> huge metastructure debate and changed everything around.

It was hardly a "huge" debate, unless your only metric of measurement
is number of posts.  Take that debate, and then re-imagine it as an
event in the physical world, with folks having face to face contact.
You'll find that none of these debates are really that big.  They just
seem big, because electronc communications can be so inefficient.

Personally, I'm opposed to a return that that hierarchy.  The idea
that somehow desktop, server, and other such projects should sit at an
exclusive top-table doesn't work for me.

Gentoo would be much more effective with having a core management team
that covered our key operations (infra, devrel, userrel, pr, releng,
and 'tools' - portage and catalyst), and which ensured that they all
worked together to give the outward appearance of an organised
distribution.  Have management focus on what forms the "spine" of the
Gentoo organisation.

The lack of this management structure is, to pick one example, behind
the grief Infra are getting over the long-term problems with bugzilla.
 Folks aren't complaining about bugzilla any more; they're complaining
about the problem continuing.  Effective senior management would have
done three things in particular here which would each have made a
difference:

  a) They would have provided oversight on Infra's handling of the problems.
  b) They would have communicated effectively with the wider
organisation, explaining what was going on, why, and when it would be
resolved.  This communication would be early, it would be frequent.
  c) They would provide Infra with resources they can't get on their
own to solve the problem, including additional budget.

It's been agreed on -dev that it's not the existing Council's job to
do any of these things wrt the ongoing bugzilla problems.  So
everyone's left with a service that's not fit for purpose at the
moment, and only Infra to grumble about.  Everyone loses sight of the
steps Infra is taking to resolve matters, and nobody wins.

Your "top table" of herds does nothing to address what Gentoo really
needs.  It's a step backwards at best.

> "Official" votes, sure. But what about GLEP discussions on -dev? That's
> the only way anything major ever happens, and it might as well be a
> requirement for a unanimous vote among all ~350 developers. The only
> time I can recall even a single dissenter before a GLEP moved on to the
> council was brix on Sunrise.

I call bullshit on this.  Big time.

There are lots of major things happening all the time - you're one of
the people who make this happen - and they don't require GLEPs.  GCC
upgrades, X.Org 7, Portage 2.1, Gentoo Overlays, Java 1.5 - these and
many _many_ more are all major things for the users affected by them.

What major things do you want to see that aren't getting done because
of the perceived need for GLEPs?

It's also worth pointing out that we're hardly snowed under with
GLEPs.  There has been only 51 in the last three years; that's less
than two a month on average, and just under 50% of GLEPs were filed in
the first twelve months of the GLEP process's existance.

Your recollection is faulty; there _is_ no GLEP for sunrise.

> > The basic cause always comes down to weak or non-existent management.
>
> Yes, and that's exactly my point. We need stronger management.

We need _appropriate_ management.  You can over-manage something just
as easily as under-managing it.  Strong management is just as
misguided.  It leads to bullying, and certainly over here in the UK
there is serious debate about whether it has gotten so far out of hand
that the law needs changing to address it.

> > I'm not sure how you can justify that statement.  To the best of my
> > knowledge, that system has only been tested in full the once - when
> > Brian was suspended from the project and Ciaran was expelled.
>
> That in itself is proof enough. There were numerous instances where it
> _should_ have been tested but wasn't, because of all the hassle required
> to do anything.

You're accusing devrel of not taking disciplinary action against
Gentoo devs because the process is too much hassle?  That would be a
very serious charge.

Or you're saying that Gentoo devs are not making complaints to devrel
because devrel's process is too much hassle?  In that case, why are
you complaining on -dev about it?  You know our conflict resolution
rules, and they don't include bitching about it on -dev.

> > Can you back this up with three examples in the last twelve months
> > where this has happened?
>
> Any long debate where more than 25% of the posts came from a single person.

I find that a poor criteria.  If you think about how few folks in
Gentoo are involved in any one area, and that most change usually has
one person acting as 'poster boy' for it, it's inevitable that you'll
end up with long debates matching that sort of criteria.

Please, provide specific examples to support your arguments.

> > Hrm.  Where is this lack of respect for devrel being displayed today?
> > What forms does this lack of respect take?  If there's a lack of
> > respect at the moment, it's not for devrel.
>
> How about in Gentoo's complete inability to do anything about the
> constant trolling and people acting like assholes?

How is that a lack of respect for _devrel_?  Wouldn't that be more
accurately described as a lack of respect between the trolls /
assholes and everyone else?

One person's troll isn't always another person's, as we'll see in a moment.

Who are the people you think Gentoo is completely unable to do anything about?

> We say we're about
> courtesy but we don't (can't?) do a damn thing about it, because it
> requires a huge, convoluted investigation and trial and nobody's willing
> to waste that much time.

What is stopping you fixing devrel?  And why are you complaining to
-dev about devrel?  Shouldn't you be complaining to _them_?  And if
your complaint to them has been unsuccessful, have you complained to
the council?  I can't find a record of that in the council logs (my
apologies if I've missed it).

I don't see how bitching on -dev is going to achieve anything - or how
it makes you any different from the unnamed folks you're complaining
about.

> I know this is partially changing, but I'm unsure that any group outside
> of the council will ever be trusted to suspend or kick people out.

The folks who don't accept devrel ... I don't see any reason why they
would accept the council on this matter.  These things don't seem to
be about _who_ is doing the kicking ... it seems to be more about
whether the kicking should be happening at all.

I don't see how bringing in a dictator is going to suddenly change the
trust in these matters, either.

> Some Debian developers commented on my blog about how valuable DebConf
> was for this.

I've been told the same from other groups too.  We'll see with the
trustee elections whether or not there's enough support for it amongst
Gentoo devs for it.

> > I'd also argue that we're _not_ powerless.  It wasn't pleasant, but
> > the old system has shown that we can deal with genuine trouble makers.
>
> Barely, and with enormous effort ...

If I wanted to fire an employee at work, the effort involved is
_substantially_ more than what we went through with that process.
Sure, we can learn from it and improve matters (and devrel are doing
exactly that; they're not exactly sitting around doing sweet FA about
it), but you have to see things in perspective.

> > We don't have a democracy.  Gentoo is largely a workocracy (there must
> > be a better word for it ;), where the vast majority decisions are made
> > by the folks who actually do the work.
>
> Only the small-scale decisions.

I can't agree with that.  I think that's demeaning to all the package
maintainers, arch teams, releng folks, and other staffers who each and
every day make decisions that are very important to the users that
they are focused on.

> > Folks don't vote on stuff.  To pick a recent example, none of the
> > folks who opposed Sunrise actually had any means to vote to prevent it
> > happening.  What they had to do was to lobby the council, who were the
> > only folks with a vote.
>
> Oh, gimme a break. Screaming about it on -dev for hundreds of posts
> isn't just equivalent to a vote, it's better. It makes people think
> there's more than 2 developers opposed to it.

Bullshit.

You're shifting the argument here.  You started by arguing that we
have a democracy - and that it's a bad thing - and when I give you a
real example of how folks didn't have a vote on something that was
important to them, you shift the argument to complain about folks
voicing their opposition.

For the record, there _were_ more than two developers opposed to it.
Both Ramereth and Kloeri have repeatedly voiced concerns about the
project.  It was me who suspended the project in the first place, and
referred the whole matter to the council for guidance.  That makes
five right there.

And even if there were only two, that's not important.  Chris and Brix
brought up many valid points, and their efforts ensured the
fundamental problems with Sunrise were sorted out before they became a
wider problem for Gentoo.  They did us all a favour in the long run.

I can't agree with your implied statement that they were trolling over
Sunrise, or being assholes.  I do agree that there was a lack of
respect with our rules - which _clearly_ state that projects can go
ahead with no announcements, no discussions; all they need is to
create a project page.  Sunrise followed those rules; I personally
made sure of it.

Those _rules_ created that mess, just as much as the Sunrise folks did.

But where was the outcry from folks, telling Chris and Brix to leave
things along, because Sunrise had followed our rules?  There was none.
 Where was the support for updating the rules, and ensuring that
another project couldn't do what Sunrise did the way they did it?
That was nowhere to be seen either.

Frankly, no-one seemed to give a damn about the rules either way.
That seems to be the real problem.

>From the perspective of what I work on, you've done _far_ more damage
with your comment on LWN about the Gentoo Overlays project than Brix
and Chris did with the Sunrise debate.  I think you have a bloody
nerve complaining about trolls and arseholes on here after so casually
dismissing the overlays project as "a hack to allow for
quasi-distributed development without using a distributed
version-control system such as git, mercurial, etc."  We set that
project up to strengthen our relationships with users, to help get
more folks involved in Gentoo, and you go to LWN of all places and so
casually dismiss it.

We're having a _terrible_ time getting folks outside Gentoo to take
any notice of what we do these days.  LWN was just about the only
mainstream place that mentioned it at all (even GWN hasn't covered it
yet, but I'm sure it will).

Don't _ever_ do that again to one of my projects.

> Untrue, voices make a democracy.

No, they just make a noise. It's widely accepted that you can't have a
democracy without freedom of speech, but speech alone does not make a
democracy.

> I'd rather get rid of devrel altogether (at least its conflict
> resolution role) and have the council deal with this.

Where would someone make their appeal to, after the council decides to
kick someone out?

> You say "unelected" like it's evil. In a company, nobody gets elected,
> but a hell of a lot of work happens.

We are _not_ a company.  We are a community; or at least, we're trying
our collective best to be.

There's nothing stopping you moving downstream and forming your own
company if you're passionate about adopting that model.  Just because
Genux was a total disaster, it doesn't mean that the idea's a bad one
in principle.

(As an aside, folks _do_ get elected in publicly-traded companies, and
plenty of companies are terrible at getting work done, which is why
many thousands of them close down each year, and why many more of them
have weak earnings/costs ratios.  It is competition which drives
progress, not whether you are a company or not.  The whole Linux
revolution will stand forever in history as proof of that).

> What vote? I'm not running for anything, and I have no desire to do so.

My mistake.

Don't get me wrong, I have a lot of respect for you, and you do a
cracking job on the X11 stuff for Gentoo.  But your arguments in this
debate have been surprisingly inaccurate, vague, and badly thought
through.  I have such a high opinion of you from your Gentoo package
work that the low quality of your contribution here has really taken
me by surprise.

> I'm just trying to get people interested in fixing Gentoo so it's not
> stuck in the mud.

I respect that, but I don't see how you're going to change anything at
all going about it like this.

There's no detail in what you want to do, only a vague unhappiness
with how things are, a desire to return to the "good old days" that
never were, backed up by arguments that are demonstrably and factually
incorrect or incomplete.

What is your plan?  Where do you want to take Gentoo, where it isn't
already going?

Where is your vision, your strategy, your roadmap, your action plans?
What are your targets for the next quarter, half, year, and five
years?  What are your strategic project milestones, and what are the
enablers and tactical project milestones required to support them?
What are your resource requirements, and your budgets?  Which
organisations do you need to partner with, and which do you need to
engage as suppliers?  Which rivals do you intend to compete with, and
on what terms?  Which values do you need to adjust to gain new
markets, and create blue oceans?  Who are your senior staff who will
share and deliver these plans?  What do you need to do to get them
onboard?  Who are the core customers, and what are their values?  Who
are the opportunistic customers draining your resources, and how can
you get rid of them w/out pissing everyone off?

These are exactly the things that Mark Shuttlework and Canonical have
answers for, which is one of the reasons Ubuntu are successful at what
they want to do.

_If_ you're looking at Ubuntu with envious eyes, my advice is that you
cross the floor and join them.  There's no sense whatsoever in putting
Gentoo head-to-head with any of the other Linux distros, unless they
try to come after what we are good at.

> The goal?

The idea would be to devolve power to developers, so that folks who
are slackers / trolls / arseholes either have nowhere to go (and
therefore drop out by default), or at least are grouped together in
one tiny corner that the rest of us can ignore.  Most disciplinary
matters get delt with locally, and swiftly.  The ones that can't ...
well, they're exactly the ones where a trial by peers are appropriate.

The thing is, businesses have spent centuries looking for a magic
bullet for managing staff.  They've tried everything they could think
of, from outright slavery at one extreme to worker co-operatives at
the other.  If there was one true way to do it, there wouldn't be any
sort of debate about it.  But all forms of government come down to the
same fundamental - it only works if folks buy into it.

Best regards,
Stu
--
-- 
gentoo-dev@g.o mailing list