Re: [tlug] GPL vs. paid version and ethics

["Stephen J. Turnbull" <stephen@example.com>]

Attila Kinali writes:

 > Why? Because most paid programmer groups i've seen so far work
 > in a group of lone-gunman kind of way.

You mean the folks who gave us Unity?  I suppose so, but then how
about GNOME 3?  GNU Emacs, anyone?  ;-)

 > While on the other hand, hobbyists talk with each other constantly,
 > review each others work and thus improve each others programming
 > skills and dublicated work is greatly limited. New people entering
 > the group usually start from an "easy" point they have choosen
 > themselves and diskussion and feedback comes to them, as soon as
 > they publish their work on the mailinglist. This greatly simplifies
 > and speeds up the initiation of new members.

Yeah, yeah, yeah, CatB! HtNS! etc, but (I hate to break it to you) as
a general proposition that theory was bullshit when a group of us
FSBers published it in IEEE Software in 1999 or so as a response to a
Steve McConnell editorial, and it's still bullshit today.[1]

That's not to say that there aren't significant, high-value areas of
software where hobbyists can do the job, namely those areas where the
hobbyists are the users who matter.[2]  Prominent examples include the
Linux kernel itself and the Apache webserver.  But if you think
ordinary hobbyists contribute a large fraction to those projects any
more (if they ever actually did aside from Linus Torvalds and Marc
Andreessen themselves, and IIRC Andreessen was actually a paid
employee of UIUC when he was working for NCSA), AFAIK you're wrong.  I
only know a very few people who have managed to get code into either
of those two projects (aside from the "big names" I could claim
acquaintance with based on a single personal email ;-), although the
list of wannabes includes folks like Steve Baur.

The thing is, there never would have been a Mozilla without Netscape,
and no OpenOffice without Sun Microsystems.  Admittedly, Mozilla today
is a more open project and still is a big improvement over IE, but
OpenOffice is just a wannabe.  Octave is not Matlab, R is not S, and
Maxima is not Mathematica.  MySQL and PostgreSQL are wonderful, but
they're not Oracle or DB2.  GNOME and KDE are not the Mac OS X GUI.
In any major system where the devil is in the details, you need to pay
somebody to deal with the devil, and you need to pay bosses to watch
the somebodies and accountants to keep track of the bosses and the
somebodies.  Even stuff like Ghostscript and GCC are at best somewhere
in the middle.

And I know a lot of small shops where people are paid to do the work
they love, and they do communicate, review, and train each other while
keeping an eye on the relevance of all those activities as well as
direct development work to the bottom line.


Footnotes: 
[1]  ESR's essays are a *lot* more nuanced, and read carefully they're
a lot deeper than the "commercial incentives lead to cutting corners"
theory.
[2]  Users of embedded systems and people who have never built a
kernel don't count as "Linux users" here because it's rare that they
can recognize a kernel issue when they see it, so there wouldn't be
much point in having a toll-free number with paid operators standing
by to handle their bug reports ("I have a 16-core system, so if it's
slow there must be something wrong with Linux" but they don't tell you
it has a full complement of 640KB RAM ;-).  OTOH, if OOo won't read
your boss's Word file, or if you can make the background any color but
yellow, the OOo developers can probably figure out what went wrong
from that -- but it's useful to have paid operators (ie, developers
with an unusual talent for diagnosis based on user reports) standing
by to filter and edit a bit.  It would *not* be useful to have a
volunteer on the mailing list who just responds to every n00b with
"good idea, patch welcome", though: you really need people to do work
that any of us would insist on a handsome salary before taking the job.