Subject: the GCC steering committee
From: Tom Lord <lord@regexps.com>
Date: Fri, 15 Mar 2002 18:30:40 -0800 (PST)


Back before the GCC steering committee, there was a formally neutral
arbiter, legally bound to operate in the public interest, who decided
questions about the evolution of the public GCC sources.  In theory,
that's a great idea.  In practice, the arbiter was under-resourced,
essentially run by the controversial judgment of a very small number
of people, and put under intractable political pressure by more-monied
commercial interests.

Now we have a steering committee and one could argue that GCC progress
has improved as a result.  Never one to be easily satisfied, I have
some complaints -- leading to business opportunities and other
positive suggestions.

The steering committee is not legally bound to operate in the public
interest.  Nearly 1/3 of the committee is from just one company.  A
larger number are (presumably) shareholders in that company.  More
than 2/3 of the committee list corporate affiliations, raising the
specter of quid-pro-quo glossing over of conflicts of interest.  Two
of the companies represented have multiple active partnerships
concerning open source software, and some of those touch on areas
related to GCC.  The proceedings and decision making policies of the
committee are not available for public review or, in any meaningful
way, open to public participation.  The membership of the committee
was initially determined in the heat of battle as a political
compromise and there is no public process for changing that
membership.

Where are the minutes?  The votes?  The agenda?  How can I get
something on the agenda?  When do I get to vote for a committee
member?  What's the protest procedure?  Where is the formal RFC/RFI
track spelled out and how can I see the status of the currently active
proposals?  The steering committee isn't a mature political process --
it's almost a "star chamber"; politically, it's structure resembles
the WTO.

I believe the SC members are good guys.  I believe you if you tell me
that they have a history of making tough calls in the right direction.
However, conflicts of interest are insidious and being a good guy
isn't enough to avoid their having impact.  Moreover, by virtue of its
legal structure and public interface, the SC is tooth-less in matters
like giving corporate GCC contributors and their customers incentive
to pay a little bit more to cover general improvements in the general
interest.  The SC is presumably not evil, pure and simple, from the
10th dimension -- but we can do even better (see below).

Let's suppose that most of the corporations who work on or with GCC
agree that they have an interest in seeing things such as cleanups,
overhaul, and experimental R&D get done.  At the same time, none of
them has an individual business justification for spending the
necessary money above and beyond their current (admittedly non-0)
levels.

A more powerful, more clearly public interest, more scrutinized, and
more accountable committee whose neutral authority was endorsed and
supported by the interested corporations and the general public would
be in a better position to establish funding for such activities by,
for example, defining the conditions of membership in a class of "good
citizen corporate contributors" and formally establishing the
privileges of such membership.

As just a crude approximation, the SC might say: "Here we have a list
of 500 GCC tasks that are important over the next N years [of course,
they'd need independent funding to develop and maintain such a list].
You can obtain the privileges of being good corporate citizen of the
GCC world either by throwing money into the pool for these tasks, or
by picking them off yourself at some rate related to your other code
contributions and bug reports."  If you're one of those companies, you
want those privileges.  If you are a customer of one of those
companies, looking to buy some GCC custom development, you want your
service provider to have those privileges even if it means paying a
slight premium over the cost of your development.  It doesn't have to
be an onerous burden -- just enough to assure steady and sufficiently
rapid forward progress.  It helps to implement a "walled garden"
around GCC in a legitimate way.  It helps to shift the competition
over GCC customization from mere price competition, to a larger, more
appropriate agenda: price balanced against stewardship.

What might such privileges include?  One example is membership on a
private mailing list on which good citizens pledge to provide mutual
aid for whatever questions or issues come up.  (For sure, I'm no
lawyer, and I believe such a list would have to be scrutinized against
whatever the legal standards are for avoiding illegally
anti-competitive corporate collusion.)

Of course,

   	(apply (lambda (gcc) this idea) other-projects-as-well)

-t