🚀 go-pugleaf

Aydin Edguer says:
> Paul Barton-Davis writes:
> > This is something ideally suited to general purpose software.
> > I'm not sure the same model can be extended to special purpose
> > systems created for 1 or 2 clients.
> Why not?
> I think your division between specialized and general purpose is entirely
> artificial.  But if you insist on it, then I will state, "I think specialized
> code is a better candidate for the GPL than general purpose code."

('Scuse me while I get up on this soap box...)

The difference between an operating system and a specialized piece of
software is that an operating system is well-trodden territory.  It
may have some innovations in it, but mostly the work involved in creating
the operating system lies in the actual design and implementation of the
software itself, and in the quality of the author's craftsmanship.  For
many specialized programs on the other hand, the cleverness of the
implementation, and trade secrets embodied in the implementation, more
than the work involved in making that implementation, are what make the
software author's livelihood.

For example:  Most commercial Computer Aided Design software is shipped
without sources.  Companies invest a great deal of time, effort, and
money into finding the best possible algorithm for, say, routing wires
on a chip or simulating the effects of faults in a circuit.  These
programs are not so specialized as to have only one or two customers,
but they're still fundamentally different from programs like gcc or
the GNU kernel.  The companies are selling expertise in the form of
software; giving away the software would be equivalent to giving away
this expertise, and most companies simply can't afford to do that.

Think of it this way:  Company X knows all there is to know about, say,
laying out circuit boards.  They write software that uses this knowledge
to do a good job of layout.  If you have a design you want made into a
board, you could mail them the design, let their engineers run it through
their software, and have them mail back the result.  Instead, company X
graciously allows you to execute the software on your own machines, thus
avoiding all the overhead of communicating your design to their people.
But for company X to give you their source code under the GPL would be
comparable to offering to have their engineers teach your people all their
best tricks, and to encourage you to tell these tricks to all your friends.
They might get a plaque for distinguished public service, but they would
also go out of business because nobody would come to them for their
expertise anymore.

You can say that the public still benefits more if the company gives away
its secrets, but, practically speaking, the company could never have
afforded to develop and refine those secrets, at the cost of hundreds of
thousands or millions of dollars, unless it could expect to make money
from them.  It can't afford to continue improving this expertise unless
it continues to make money from that expertise.

This is what patent law is for, by the way.  In a perfect world, company X
could patent their layout algorithms, then release them without a worry.
But wait---algorithm patents are evil, too!  We can't stamp out every way for
a person with expertise to make a living.  I personally think that hoarding
*truly* proprietary software is a lesser evil than allowing people to patent
software left and right.  If we could trust people to patent only truly
specialized and innovative things, like their clever PC layout algorithm,
then maybe I'd feel differently, but experience shows that the lawyers will
patent anything they can get their hands on.  So it looks like patents are
a civilized luxury we can't afford in the software biz.

Richard Stallman's approach, of giving away the software and making a living
from support, is a good idea and it works well for his kind of software.
But it's not practical for everything.  I have to agree with Paul here that
I believe some of the FSF's literature is guilty of pretending that all
software is just like gcc and GNU Emacs.  (I won't pick on the GPL, though,
because the GPL accompanies specific programs, and nobody needs to put it
on their program if they think it's not right for that program.)

The GLPL allows companies that can't afford to give away their software
to benefit from free software libraries.  Those companies will owe a debt
of gratitude to the authors of those libraries, and, as the FSF has seen,
some of those companies will decide to repay that debt by donating
generously to the Free Software cause.  In view of this, I think the GLPL
is a healthy move in the right direction for the FSF.  It's not so much
a compromise as a broadening of scope.

								-- Dave