when companies join over free software projects

[Mark Galassi <rosalia galassi org>]

Here are some observations I have made about what happens when
companies are involved in free software projects.

The overall effect tends to be good, because they do pay people to
work on those projects, and to do some of the ugly grunt work.  The
Helix Code distribution, for example, is awesome.  GCC as a
cross-compiler is awesome.  Some Linux distributions are awesome.

On the other hand, if things are not done well, they can hijack a
project and take it down a bad path.

I think that the correct approach for the company to take is to have
some very good programmers join the projects without being fettered by
management in any way at all, and to do things that are unrelated to
the company's marketing plan.  Sort of "hackers without portfolio".

These hackers-without-portfolio would be a most welcome contribution
to the free software project.

The company would then have other programmers working on the project
doing customization toward its specific goals.  These programmers
would not do all that much for the community, except maybe indirectly
(as they would make the company feel that they are making a profit
from this free software collaboration).  They would, though, be doing
good work for the company, and not hurt the overall effort, especially
if they pass their changes back to the maintainers.

Examples (does anyone want to add more):
----------------------------------------

Cygnus's contributions to GCC have been one of the best things that
happened to free software.  The main hackers that Cygnus assigned to
gcc worked with the community, and they did not have to answer to any
marketing pressure.  Their purpose was to make gcc advance.  Cygnus
then also had other people scheduled to meet its contractual
requirements.

GDB, instead, is frequently considered a failure of the "company takes
maintainership of a program".  Cygnus was unable to redesign gdb, and
it is frequently thought that this was because Cygnus was the sole
maintainer of GDB, and decisions were made in-house, based on what
contracts Cygnus had to deliver on.  Cygnus would frequently try to
re-design GDB, but the design efforts always fell apart because they
were subject to various pressures inside the company.  If there had
been a net-wide design effort, and Cygnus had assigned a few
programmers to work with that effort *not* representing Cygnus, then
we would probably have had a new GDB design.

Red Hat has Alan Cox (anyone else?) working on the Linux kernel, and
he works with the kernel team instead of pushing a Red Hat agenda.
Red Hat also has many people who patch the kernel and make packages
and do things to prepare it for a Red Hat release.  Clearly Alan Cox's
contributions are the most useful ones to the free software community,
and I'm sure even Red Hat's competitors appreciate them.  Conversely,
I have worked with people at Red Hat who package my tools, and it has
not been a good collaboration: they are too busy with Red Hat
deadlines to coordinate their .spec files with me.

LinuxCare Italia, formerly Prosa, has programmers working on Embedded
Linux, GNOME, and other free software projects.  These guys' job
description is indeed "hacker without portfolio": they are chartered
with doing the best thing for free software.

In the case of eCos, the free software "Embedded Cygnus Operating
System", the sources were put on the net.  Originally one person was
appointed as the only person who should answer net questions in a
rather official capacity.  When they opened it up so several guys on
the team could participate, the collaborative nature of the project
increased drammatically.


What's happening with GNOME:
----------------------------

There have been times in which people at Helix Code did things that
irked the rest of the community.  They were not major, and they were
resolved, but they made some people start to think that GNOME was
being hijacked.  It would have been better if that had not happened.

Helix Code and Eazel also played some weird games with the VFS and
file manager which would not have happened if all the players had
still been volunteers.

The current public relations fiasco (NY Times article, and various
other articles) is another sad result of the corporate interest in
GNOME.

I'm sure there are other examples.

So some things have irked GNOME developers.  This should have been
avoided.  Maybe it could be avoided in the future with a different
model for corporate participation in GNOME.


How should corporate involvement work?
--------------------------------------

I think Miguel has been a wonderful project leader for GNOME.  I think
he can continue to do so, but he has to make sure that when he makes
GNOME decisions, he is not tied to any company or corporate
consortium.

I think that all the companies involved in GNOME should make it clear
that some of the programmers are just working with all the other
programmers, as if they were independelty wealthy (or sysadmins with a
light load) and volunteering.

The companies should then have other people working on their
deliverables.

If this were to happen, I think the volunteers feel good about the
companies and they would be quite happy to rally behind these
companies.  For example, they help coordinate with their deadlines.

For example, I remember that I had no weird feelings when Red Hat labs
first collaborated on GNOME: they seemed to be good citizens.  When I
found out they had a new OS release coming out, and GNOME was *very*
young and had no intro docs, I took a bunch of time to write the first
introduction to GNOME so that they could ship GNOME with an intro.  I
would also coordinate releases of the DocBook tools with them so that
they could use them.  I think other people felt the same way.

With today's twists I feel that we are being a little hijacked by the
companies, and I think many feel as I do that they are less motivated
to work with these companies to make them succeed.

If Helix Code, Eazel, Sun and others make it clear that they guys on
the gnome-hackers list are functioning "without portfolio" and with no
vested interest, I think we would all feel differently about their
contribution.

------------

What do you all think of thise point of view?  Does it seem
reasonable?  Does it look like something we want to suggest to these
companies?  Should be a recommendation in the GNOME Foundation
Charter?

------------

PS: this is rather unpolished, and I did not put as much effort as I
should in making sure my examples are checked for accuracy and that
everyone else perceives them the same way I do.  I think it's OK, and
I want to get people thinking about this before tomorrow's
announcement.

PPS: I wonder if the two dudes from Sun who appeard on gnome-hackers
get this list; if not, could someone forward this note to them?

PPPS: disclaimer.  I used to work for Cygnus, but I based my Cygnus
perceptions on what I remember other people telling me about what they
saw coming out of the company, since that allows those perceptions to
be calibrated to my perception of other companies where I have not
been on the inside.