Subject: Re: Dealing with the Open Source community
Date: Sat, 18 Nov 2000 15:28:50 -0800
Sat, 18 Nov 2000 15:28:50 -0800
on Sat, Nov 18, 2000 at 01:21:30PM +0000, Simon Cozens ( wrote:
> On Sat, Nov 18, 2000 at 04:54:40AM -0800, wrote:
> > Dumping dead product on the free software community doesn't bring it
> > back to life, and broken code developed in a proprietary environment may
> > well be beyond resucitation.
> Yes, this was another thing I was going to talk about. Can you say
> "OpenOffice"?

<g>  Don't get me started.  

Cliff's Notes summary:  I consider OpenOffice a failure until otherwise
demonstrated, for many of the reasons touched on elsewhere.  The most
useful immediate products are the file format translation libraries
(I've heard of these, not sure they are actually released), and some
press.  While the product works, for broad definitions of the word, and
I've been known to use it for presentation tools, I far prefer other
alternatives -- mutt and vi for most work, AbiWord, Gnumeric, and any of
the various browsers-that-suck (w3m, lynx, Skipstone, Galeon) for their
relative apps.

I've long felt that a high degree of modularity is essential for
successful free software projects.  Documentation and comprehensibility
are also core requirements.  OpenOffice has a steep curve to climb.

> > Mild disagreement.  Status translates.  Attitude doesn't.  Or rather,
> > bad attitude gets old fast.
> OK - putting it another way: you can't assert status.



> > > open source community do it for one simple reason - fun. We don't get
> > > paid for it, we don't *have* to do it. 
> > 
> > I dispute the last sentence, particularly the first clause, strongly.
> > Free software wont scale if it can't be a livelihood, and hasn't scaled
> > without becoming so. 
> ...
> > But free software programmers, like any other kind, have bills to pay,
> > families to feed, and dreams not (necessarily) existing online to
> > fullfill.  And that takes dollars, euros, yen, and rupees.
> I addressed this is my last article, which I posted here. :) That as may
> be, I would still maintain that those not paid for working on free
> software are the vast majority.
> Yes, everyone needs bills to pay, and most free software hackers have a
> day job as well. It's still primarily a volunteer effort.

There's a paper writen by Gerald P. Dwyer, Jr. of the Federal Reserve
Bank of Atlanta, which tries to explain the same PoV:  that the majority
of FS development is part of a gift culture.  I'm uncomfortable with

There is alternative evidence, some concerning the Linux kernel
development effort, another Mockus, Fielding, and Herbsleb's "A Case
Study of Open Source Software Development:  The Apache Server".  I
believe there's similar work oriented toward Perl.  In most cases,
while there is often a large development community -- Apache lists some
3,000 developers who've contributed code and/or bug reports -- the core
development effort is often quite small.  In Apache's case, the core
consists of some 15 developers total, at any time six or so of whom
might be active.

In the case of Apache:

    The top 15 developers contributed more than 85% of the MRs
    [modification requests] and [code] deltas, 88% of added lines and
    91% of deleted lines.  Very little code and, presumably,
    correspondingly small effort is spent by non-core developers.

    (Mockus, Fielding, and Herbsleb.)

I'll disagree slightly with the conclusion that little additional effort
is spent by the non-core developers, particularly in aggregate.  The
debugging and support rôles, in particular, are key.  However, they are
mungeable, and can occur as part of an encompassing rôle.  Particularly
in the case of project such as Perl, a widely used, extensible
development environment itself, used strongly by technical users, I
suspect there's much "fringe" development of Perl modules by developers
who use Perl in their daily work.  While not paid directly to develop
Perl, there is Perl development occuring as part of their work.

My own employer could be a case in point.  Zelerate's AllCommerce
product could, it's been suggested, be rewritten as a set of Perl
modules.  There's a fair bit of argument that this would make the
project rather more manageable.  Which would effectively turn both our
in-house and external developers into Perl developers.

While I agree that there's much free software development which occurs
out of work hours, by students, etc., I'm not sure this is any different
from the hobby tinkering that occurs in other development circles.  It's
effort which can be collectively shared, however.  I've felt for years,
though, that ultimately it's going to be paid work, likely from
consulting giants (I'd fingered IBM as a likely very good fit in 1998),
as well as other sources.  

More recently, I'm favoring the "free agency" developer model which I'm
starting to see expressed more frequently.  I know of several developers
who's basic attitude is "my primary allegiance is to $PROJECT, and
$EMPLOYER knows it".  This reflects my own professional evolution:  from
employee at a company, to association with a project, to association
with a development tool (in my case, SAS), to a computing environment
(Unix), translated to GNU/Linux, to free software in general.  The
shifts have been toward associations which increase direct benefit and
flexibility to me.  I see it as a trend.

Karsten M. Self <>
 Evangelist, Zelerate, Inc.            
  What part of "Gestalt" don't you understand?      There is no K5 cabal

["application/pgp-signature" not shown]