Subject: Re: Do We Need a New Evangelist
Date: 30 Mar 1999 20:20:06 -0000

>So I still see a difference.

FWIW, I, too, see a difference, but share your hesitance to put it
down to important differences in licensing, versus just politics,
in terms of how "the masses" (starting with those roughly in the
overall "freeware community", and expanding outward) view it.

But, I don't think I'm personally concerned about the politics
(I don't particularly care about personalities either), yet I still
do see important differences in the various licensing approaches,
at least as I've seen them presented in the past.

E.g. I personally take the stance that, generally, if anyone wants
me to write code to which they have, or somebody else has, some special
rights, they'll have to pay me substantially (e.g. my going contract
rate) to do it.

Whereas, I'll consider writing code for them, code to which they have
no special rights, as a volunteer (which is essentially what I do
when I fix bugs in, or add features to, egcs, g77, etc. upon request).

And, to me, that means I'm willing to voluntarily contribute to GPL'ed
code, but not, e.g., NPL'ed code.  (Again, generally speaking; there's
always the possibility of exceptions.)

For many, many years, during which I and others defended the viability
of the GPL, we raised this point: to the extent any private entity wishes
to exercise "special" rights over code they don't themselves author (and
thus own the copyrights to), they might indeed be able to attract
additional revenue streams to fund development, but they'll be losing
the willingness of some people, like myself, to voluntarily improve
that code.

I still don't know how important that difference is.  Linux vs. BSD
is an example, but I've been told (fairly reliably) that the legal
threats by AT&T against BSD (I'm shaky with the precise terms
and acronyms here) were mostly, if not entirely, responsible for Linux'

Still, the fact remains that Microsoft can sell proprietary (no-source-
available, pay-per-licensed-copy) versions of any of the "free" BSD
versions of UNIX at *vastly* less up-front and subsequent cost than
it can sell similarly restricted versions of Linux.  (Let's face it,
MS could sell a proprietary version of Linux and face basically no
substantial *legal* threat in response, but it'd take a lot more of
a media-relations beating than if it used FreeBSD instead.)

And, I think it's quite likely that a number of people working on
both Linux and FreeBSD imagined that their work could be interesting,
someday, to Microsoft (for example).  (I wonder to what extent MS is
now considering this, as a means to fight, if not kill off, the
encroachment of Linux on their Windows NT server turf.)

The difference is, the people working on Linux generally would figure
MS would have to either hire them to write a new OS entirely from scratch
for MS, or MS would have to offer its improvements to the Linux code base
to the public as source code.  Whereas, the people working on FreeBSD
generally would figure MS would just use their code, without having to
get any specific permission, in a proprietary version, and *maybe* would
hire them if MS felt it needed more such code -- but that the
economic attractiveness of this proprietary route would make this
scenario, instead of the Linux/GPL scenario, more likely.

It might well be that there ended up being more Linux developers of
this sort than FreeBSD developers.  Because, given the pace of MS
hiring, if you're going to write code that MS might use, why not get
paid by MS (and, someday, vested) to do it?

That's a subtle distinction for the GPL vs. BSD licenses, amplified
in this instance because of MS's monopoly.  (That is, MS isn't the
only entity that can ship proprietary versions of FreeBSD, so, in
that sense, it's still "free software".  But, as a monopoly, MS
would be much more likely to succeed in convincing the industry that
*its* FreeBSD code base -- under whatever name it gave it -- was
*the* new standard for (BSD) UNIX, making the utility of the BSD
license's equanimity vastly lower, and the theoretical freedom of
less practical utility.  It's much harder, though perhaps possible,
for MS to do this to Linux and the GPL as well.)

So, I can certainly believe that the greater differences between
the GPL/BSD form of license and the NPL form of license would be seen
as even more pronounced.  The GPL/BSD form grants basically equal rights
to authors of code; equal rights to users of code; and have
differences mostly in the rights of authors vs. users.  Whereas,
as I understand it, the NPL form grants greater rights to certain
entities than they do to authors and/or users of code.

Regardless, my sorely-under-researched impression of "Open Source"
as a term is that it has its usefulness, as a trademark, as a
concept, and as an indicator of certain forms of robustness --
which can be useful to insist upon for products used within an

In particular, it'd have done less good to publicize "free software"
as widely, when it came to stressing the importance of having the
source code, etc., because nothing would have prevented MS from
claiming, upon seeing the term gain importance, that Internet Explorer
was "free software" (as my MS-encrusted relatives have), perhaps
even that Windows was "free software" (because you always got it
with new PCs, right?).  Trademark law makes that sort of hijacking
of "Open Source" much harder.

Still, "Open Source" is quite different from "free software" to me,
even though the former is a superset of the latter.  As purely an
end user, the difference might be less important, but if I'm even
remotely considering caring about, finding, and fixing bugs in the
code or the documentation, I want it to be *free software* and,
ideally, GPL'ed software.  Else I have to think through just how
much volunteer work I'm willing to do for someone else's special

        tq vm, (burley)

P.S. I do wonder how FreeBSD developers, especially those fitting the
above description, feel about the success of Linux.  On the one hand,
it increases the likelihood that MS might want to offer, and thus hire
known-good people to write, a proprietary UNIX that competes head-to-head
with Linux.  OTOH, the name recognition of Linux reduces the likelihood
that MS will think it will be able to convince people to see any UNIX
*not* called Linux, and perhaps even not *actual* Linux, as worth
bothering about.

Say MS decided it had to offer "Microsoft Linux" soon.  If it went
against its own grain, by sticking to the spirit and ethics of the
Linux/GPL mindset, MS Linux would be akin to RedHat Linux, perhaps
with some pretty useful MS compatibility modes added (as GPL'ed
software to the kernel, surely proprietary anyplace they could get
away with it).  The end result would be no need to hire FreeBSD
developers specifically, no cachet from having done so, and, in fact,
the possible perception that they were hiring people who were working
on a code base with a head start, more initial PR, a short-term
legal hassle, and yet still failed to compete against Linux.

If MS decided "Microsoft Linux" could be something like FreeBSD
modified to be Linux *compatible*, but still proprietary, that might
make hiring FreeBSD developers worthwhile.  But how seriously would
the industry take "MS Linux" if it couldn't look at the source code
and thus see how Linux-like it really was?  Especially given the
increasing number of people who seem to "get it" that the success
of Linux isn't due to any snapshot of its code base per se, but
its ongoing improvement by a large contingent of volunteers willing
to help out -- the vast majority of whom would *not* help out people
wanting help with an "MS Linux" of this form.

So I wonder if there are FreeBSD developers out there who feel they
invested in their form of free software and are more and more likely
to come up "empty", because their "free" code is not catching on,
and the value of its being available to take proprietary seems to
be rapidly dropping.

It's a question that, directed at me regarding g77, I answer "I'm
happy to have done what I've done, wish I could have done more,
but generally no regrets", because I didn't work on g77 to try in
any way to drum up interest among compiler vendors to hire me to
work on their Fortran products (heck, that's what I was doing, for
plenty 'o' cash, before I quit to do g77).  Yes, I'd be much
happier if g77 was more widely seen as worthwhile, but not for
myself personally as much as for what I see is the value and promise
of GPL'ed free software.  If g77 really goes "gently into that good
night" and no GPL'ed Fortran 90+ compiler is produced, the overall
attempt to show that the GPL model can produce a quality, cutting-edge
Fortran compiler can be said to have failed.  But that doesn't make
me feel bad, personally, except in the sense that maybe I could have
done better -- because my motivation was to help everyone have
source code to their Fortran compiler, and that's still something
many can choose, and choose to help with, even after g77 is declared
"dead", should that ever happen.

(I'm not sure it'll happen, by the way: the egcs project, a bazaar-
style one, is where the g77 action is these days, and I've been
discouraging people from trying to work on the "soon-to-be-rewritten"
portions of g77 for years now.  We're going to try to iron all this
out at a Fortran BOF at LinuxExpo -- see my web page,
<>, for more info.)

Ultimately, though, I hope FreeBSD developers are able to be as happy
to have done some fun work, to have learned a lot, and to have helped
others, as I've been working on g77.  Not being annointed as the
"winner" is really only a problem for those who see the competition
itself, and the work they do engaging in it, as of little value.