[Gnu-arch-users] while waiting for the coward, Cats and Bazs

[Thomas Lord]

While we wait to hear from Mark, I note that today's /. brings
us yet another round of self-promotion from ESR.   In a refreshing
bit of honesty he reports a fact relevant to understanding what's
going on with baz, Canonical, Red Hat, and other corporate efforts
in the free software and open source space:

  ESR:
  > You’ve maybe heard about this “open source” thing? You get
  > one guess who wrote most of the theory and propaganda for it 
  > [....]

He ventures into fantasy with the rest of that sentence:

  > and talked IBM and Wall Street and the Fortune 500 into buying in.

but a propagandist can be forgiven for overrating the significance
of his own work as surely as I can mine.

In a recent message I pushed aside a reference made to "The Cathedral
and the Bazaar", dismissing it merely as a garbage essay.   A private
correspondent prompted to elaborate further.   Since it's relevant
to the strategies and tactics I'm railing so self-righteously against,
in light of today /. post, I'll reproduce that off-list message here.


Private correspondent:

> I'm curious about your criticism of the essay.

I have both technical and political criticisms.

Politically, it has been propagated, waved-around, and put to
use by business interests who have a need for uncompensated
labor and a need to spin a yarn around the false claims they make
about the quality of their products.   To the "kids" in the
industry, it becomes a bogus received wisdom about their careers
and about software engineering practices.

Technically:

>  I've been thinking of 
> writing up a criticism for some time [....]
> Could you  clue me in a bit on your criticism?


The essay rambles on an on in an unfocused and self-congratulatory
way. If you want to write a serious critique of it, a good place to
start might be trying to rescue a thesis and argument from it from
amongst the detritus.  The argument, such as it is, is predicated on
the idea that certain programs (e.g., the Linux kernel) have a high
quality which  needs to be explained -- measuring the quality of
software is a  controversial topic but if you are willing to take a
stand on it you might critique by undermining that predicate.  (Of
course, any claim that today's GNU/Linux systems are less than
stellar is not going to win you many friends in the FOSS developer
community.)

I think that what the essay most *shows* (anecdotally) is that when lots
of people are excited about the opportunity to hack with others over the
network, and when there is a paucity of projects to join, through a bit
of glad-handing and name recognition you can attract lots of helpers,
get an *interesting* code-coverage and portability work-out during
"beta" phase, and often collect a lot of patches.   In that regard, its
popularity among a certain species of business leaders is unsurprising
-- I have the impression that ESR is "writing between the lines" to
that group with exactly that intent.

A corollary of what is shown would be that when there is an abundance
of projects to choose from, competition for these helpers heats up.
In that situation, one would predict certain businesses to invest
more in glad-handing.   In historical fact, we've seen not only that
but even the meta-level: e.g., Collabnet selling training to other
companies about how best to "win friends and influence people" in 
the developer community.

The essay is often *taken* to be about software engineering practices
and I suspect it is in that light that you ponder the Cathedral/Bazaar
dichotomy.   It is not, for the most part, about software engineering.
It is more about management of and investment in programmers.

Where it tangentially touches on engineering issues it does so
shallowly and in ways that undermine any conclusion that there
is such a thing as "Bazaar-style Engineering" which leads to good
result.  E.g., working alone, ESR improved his inherited program
by making architectural changes;  ESR needed to rework the program
in isolation until he felt he understood it from top to bottom; etc.
He wanted to launch his experiment with a starting code base he
considered to be controlled and extensible and to achieve those aims
in the most cost effective way he employed "Cathedral" techniques.

Absent from the essay are any metrics or objective arguments that the
result of going Bazaar did much more than find a couple of bugs he
didn't expect and win him a modest-sized mailing list.  There's not
much there to poke at which, apart from the political digressions,
is why I commented on-list that not much could be said about the
garbage essay.

Also absent from the essay are any metrics that show how the 
Bazaar is a sound investment strategy for managers of software
projects.   The upside (free debugging help, higher quality
bug reports) is shown -- but no downside or evaluation of the 
net.

It's hard to say much about such hype other than "It's garbage."

-t