Subject: Re: Can open source cost money?
From: Craig Burley <burley@gnu.org>
Date: Fri, 28 Aug 1998 11:18:34 -0400 (EDT)

>> How soon we forget.  The 800-pound example is SunOS, and the tens
>> (hundreds?) of other proprietary BSD Unix derivatives.
>
>Yes...but what conclusion do you draw from that example.  Here's
>what I draw:
>
>* Some people took the BSD work commercial, and created a lot of
>customer value from it.
>
>* Some people continued to develop the BSD work in the open source
>world, and created a lot of customer value from it.
>
>Where's the problem?

Well, SunOS has not been supported by Sun for (AFAICT) quite some time
now, and plenty of bug-fixes and improvements that have made it into
FreeBSD cannot be pulled into SunOS even by customers who *paid* for
SunOS -- because they didn't pay enough to get the source code and the
rights to improve it.

Of course, there's always Solaris, the "upgrade path", which Sun
has continued to improve to (again, AFAICT) the point where it's
generally better than SunOS.  But that presumably still leaves some
older customers with older machinery in the lurch -- they can't
get improvements to SunOS anymore, and they can't upgrade to
Solaris without paying $$$ for new hardware to run it as well as
their old hardware runs SunOS.

OTOH, FreeBSD (presumably) and Linux run on machines that, by today's
standards, are very ancient and small -- I think Linux can be made
to run on 386's with 4MB of memory and 40MB hard disk, or some
such thing.  (I gather it's not easy to squeeze it down to a small
size.  It's probably much easier when one has the source, though.)

>What was libre didn't become non-libre.  What happened was that someone
>added value, and didn't make the added-value libre.
>
>The only party that has the right to complain is the original author, if they
>wanted their contribution always to remain libre, and used the wrong license.
>That's what the GPL is for.  But if the author doesn't mind, why should
>anyone else?

The *real* debate here is, why should *customers* mind when they pay
big $$ for a system that they cannot be sure they'll be able to improve,
maintain, and so on using the resources *they*, not the proprietors of
their software, have on hand?

>I guess what I'm arguing is the moral right of the author of the software
>to put it under whatever license they want.

I think everyone understands that.  And it is the moral, legal, and
economic right of the *end user* to require software to be licensed
to them under whatever license *they* want.

>Personally, I find the world that allows both Solaris and FreeBSD to
>be created on the base of the original Berkeley UNIX work to be superior
>to one in which only FreeBSD could exist.

(You mean SunOS.  Or is it known that lots of BSD code remains in
Solaris?)

>Often, the choice is not between non-libre and libre, but between non-libre
>and nothing.  Not everyone has the luxury to create libre software.  I
>applaud anyone who creates value for others, under whatever license
>they choose to use.

Absolutely.  And I applaud anyone who *insists* on value for themselves,
when they acquire software, in particular anyone who recognizes the
substantial long-term benefits to *themselves*, not to mention the
better overall global economics, of GPL'ed software.

After all, it's pretty reasonable to conclude that the licensing scheme
that permitted both SunOS and FreeBSD to exist *didn't* allow FreeBSD
to share the improvements made to SunOS in that original BSD core, so
everyone could enjoy them.

And the Big Question for the free-software community these days is,
is it precisely *because* Linux is GPL'ed and thus there is really
only "one" Linux (with many variants, apparently), that led to Linux
basically overtaking FreeBSD in the customer-acceptance arena?  Is
the fact that worthwhile improvements to Linux almost *always* find
their way back into free-Linux (usually without requiring an expensive
re-engineering effort), while the same kinds of improvements to the
BSD OS source base often *don't* find their way back into free-FreeBSD,
one of the Big Reasons for why Forbes (among others) talk about Linux
and not FreeBSD, and Linux gains market share (or is it seats, I forget)
while no other OS but a MS one does so?

I'm leaning towards "yes", obviously.  Though I respect the way Linus
has done his job, and believe that, too, contributed to the current
situation vis-a-vis FreeBSD, I find it hard to believe it amounts to
a substantial-enough difference to make the use of the GPL irrelevant
or especially a hindrance.

Other people (esp. GPL opponents) *claim* that the GPL has nothing
to do with it (some say it has *hurt* Linux vis-a-vis BSD), but the
only reasonable alternative is that Linus did an unbelievably fantastic
job.

IMO, if Linus himself is substantially responsible for this success,
and selecting the GPL was, at best, coincidental (ignoring his
claims to the contrary), then why haven't any of the people who
believe this hired Linus to do *just* project management?  Given his
success rate and the supposed irrelevance of the GPL, he could,
with these superpowers, plus VC backing, create a serious competitor
to anything Microsoft produces within about 3 years, and that'd make
him worth at least $500K/yr plus stock options.

Does anybody know roughly how much Transmeta pays him?  I.e. do *they*
believe it was he, much more than the GPL, that has been responsible
for the success for Linux, and show it with their compensation package?
(They'd pretty much have to do the latter if they believe the former,
since Linus' work is so highly visible, unless Linus doesn't place a
typical value on typical forms of compensation and Transmeta has something
special to offer that he *does* value.)

But my point is, and has been for several years now: the *real* advantages
of the GPL don't show themselves until *end users* decide they're going
to value having the complete source for *programs* they use to a much
higher degree than they do today.  And, with the significant technical
advantages such source availability gives (which I believe we're barely
scratching the surface of, in terms of offering, as a community or
industry), I think such an increase in perceived value is quite likely.

Until then, the *other* market factor contributing to "GPL superiority"
is the fact that some *programmers* (and other kinds of contributors)
place a high value on being sure the source code they write will be
available to them at any site that is using it.  Ignoring the
altruistic aspects of this (which can be significant), the market
factor is simple: a programmer who knows he'll always have his own
source code to modify, no matter who hires him to improve programs
containing it, knows he'll have a wider, longer-term market that'll
value his particular skills.  (Yes, this is not the only value; other
concerns include knowing that there'll be widespread *use* of
his source code, which gets back to the original question.  But, right
now, I bet core Linux programmers are feeling pretty good about their
long-term ability to find work on Linux as compared to core programmers
who've spent several years working on FreeBSD...or, for that matter,
any version of MS Windows.)

        tq vm, (burley)