Open Stack

On Mon, 9 Sep 2019, Adam Spiers wrote:

> Chris Dent <cdent+os at anticdent.org> wrote:
>> On Fri, 6 Sep 2019, Jeremy Stanley wrote:
>>
>>> I'm disappointed that you don't think the software you're making is
>>> open source. I think the software I'm making is open source, and if
>>> I didn't I wouldn't be here.
>
> I wouldn't either.  I'd be very worried to live in a world where there
> was no serious open source rival to AWS, Azure, GCE etc.

(One possible tl;dr of the below is: For at least some people working
on OpenStack the more direct and immediate cause and result of their
work (whatever their intent) is the enablement of corporate profit
(through sales and support) not individual humans using the
software.)

>From some standpoints I would guess that OpenStack looks and behaves
like open source: people work on it collaboratively and the code is
available for anyone to change.

And I would agree that from that standpoint it is open source, and
the four opens are good both in letter and in spirit.

I would also agree that the academic and other non-profit use of a
OpenStack that Jeremy is very compelling and motivating.

But the context of much of this thread has been about the experience
of the developers making OpenStack. How they come to be in this
situation, how they manage their work, who they work with, what
drives decisions, etc.

(What follows is a ramble. A apologize for not being able to write
less. But you did ask so here it goes.)

In the context of daily developer experience, things are less clear.

They would be more clear and would feel more like open source if I
was more frequently collaborating in the creation of code with people
who were using OpenStack. But I don't. Most frequently I'm
collaborating with people who instead of using OpenStack are helping
to make something for other people (with whom they have infrequent
collaborative contact) to use OpenStack.

For some people this is not the case. For example, many of the people
who have been deeply involved with OpenStack infra use OpenStack all
the time and also work hard to improve the code of OpenStack.

But on a daily basis that isn't my experience. Nor does it feel like
the experience of most of the people I tend to collaborate with.
Yes, sometimes I will collaborate with someone from CERN to create a
feature, but this is rare. Usually I collaborate with people from
Intel, VMware, Red Hat, and a variety of Telco vendors. Doing a
thing to help an existing customer or hoped for notional customer,
both of whom are abstractions at a distance, not humans.

This isn't a bad thing. Organizations collaborating in any way is
great.

But it doesn't _feel_ like "open source" to me. And that feeling is
an important factor (I think) in analyzing the motivations people
experience when working on OpenStack and the choices they make with
regard to how they act in the environment.

As someone who has done what could be called open source since long
before the term was invented, the common failure of corporate
patrons to give maintainability and quality (of product and
(critically) the experience of creating it) sufficient attention is
a source of a great deal of resentment and internal conflict. I am
far too conscious of the necessity to compensate for that failure if
I want to feel a sense of well being with what I'm helping to create
(both in terms of product and the environment it is being created
in). That is: I care enough to try to do what I think is right.

In this thread, and the one that started it, we've put forward the
"maybe we should just chill" as a bit of an antidote to burnout and
overcommitment. While I rationally think that's the right idea,
emotionally it is very hard to do and the source of that difficulty
is this:

OpenStack has constituted itself over the years as the domain of
contributing corporations. Many paid contributors for whom working
on OpenStack is their job. At the same time we have also been very
vocal about being not just open source, but a source of good wisdom
(the four opens) on how to do open source well.

The latter creates a community I want to believe in. A source of
pride. The former creates a conflict of interest, a frequent
inability to do the actually right thing for the long term health of
the community. A source of shame.

Continued pleas to get the corporates to do "open source" well -- that
is with correct attention to:

* developer experience
* maintainability
* architectural integrity
* deeper/closer ties to user engagement and their satisfaction

and thus some akin to "actually open source" -- have fallen on what,
if actions speak louder than words, are deaf ears.

This creates a conundrum. I've tried a variety of ways out of it.
One I'm experimenting with now is realizing that OpenStack really
isn't, now, proper open source.

And if it is not, then

I don't have to care because

they don't.

> Again I'd be very interested to learn more about your take on what we
> can do better.

There are two directions to go:

Maintain the mode of corporate-contribution-driven development. If
this is to be healthy then the corps doing that contribution need to
invest far more heavily in general, but especially in the items I've
listed above at "correct attention". This would grant the community
sufficient resources to evolve out of its aging models for
development and governance. You have to have some free space to have
the head space to get to new spaces.

Start breaking down the corporate-contribution-driven development.
Encourage professional openstack devs (like me) to age out of the
system and discourage new ones coming in. Encourage feature
development from and via users. Feature velocity might drop
drastically but they might be features individuals actually use
within a few weeks of their release rather than a few years.

Some of this latter is already happening. Especially in what some
people call the non-core projects; things associated with deployment
for example. But in projects like nova we're heavily driven by
trying to create a feature base which is predicted to drive sales,
either directly or indirectly.

And, though opinions and experiences differ, my opinion and
experience is that driving sales as a direct factor is anathema to
"open source". Indirect? Sure, whatever, if that floats your boat.

The proper direct factor is humans.

There's a lot more to this than I've stated here, but I hope that
gives at least something in answer to the question.

-- 
Chris Dent                       ٩◔̯◔۶           https://anticdent.org/
freenode: cdent