Subject: Re: Successful FSBs
From: "Stephen J. Turnbull" <stephen@xemacs.org>
Date: Tue, 24 Sep 2002 13:46:05 +0900

Executive summary:  weakening the definition of FSB because few
companies qualify is a bad idea, IMO.  Maybe we should weaken it, but
if so, what we should do is weaken the definition of FS from RMS's
definition or the OSD to "achieves the goals of education, sharing
among programmers, and modifiability/repairability by users to a very
high degree."

Interesting back door for O'Reilly, for example, because even though
the books are (mostly) not free, they are clearly enabling for all
those goals, not least through their (well-deserved) popularity.

>>>>> "David" == David Kaufman <david@gigawatt.com> writes:

    David> so my question then is, if we want an FSB to be something
    David> in the middle,

I don't see any point in defining an FSB as "something in the middle".
We can admit that all will fall short, but still aim for perfection.

    David> between idealisticly aligned development businesses and
    David> greedy users using the software to produce proprietary
    David> source-secret software, then what exactly do we want an FSB
    David> to be?  must it assume one or more of the roles
    David> specifically laid out in the GPL license?

No.

    David> to me, Open System Consultants appears to be a Free
    David> Software User.  They *are* an important part of the Free
    David> Software Industry, as members of it's *consumer* market.

Bull.  As consumers, they're tiny.  "Part," yes, but negligible, just
as I am.  Let's be realistic: the only potentially important consumers
of open source software as individual entities are governments.

BTW, I think it's Bernard Lang and Jean Camp we owe thanks to for
keeping those possibilities "on the radar" as they develop in the EC
and Peru, etc.

    David> the Altruistic Goals of the open source community,

The open source community's goals are in general not altruistic, or at
least not necessarily so.  We have a classic multiperson prisoners'
dilemma here (it is a dominant strategy, or nearly so, for everyone to
free ride) with substantial beneficial externalities to choosing the
dominated strategy.  These beneficial externalities are especially
strong because of the network effects.

Thus, compared to any "classic game theoretic equilibrium" it is in
everyone's interest (including Microsoft's, I suspect, although I
haven't computed it) to have an open source outcome where all
contribute their software to the public domain, and all return
appropriate financial remuneration to developers whose software they
use.

This may be a pipe dream, but OTOH there may be institutions we can
construct which will enable us to come far closer to the ideal than
the current situation.  Let's not give up the ideal.  That unfairly
cheapens the achievements of those who "made it" as well as those who
came damn close.

    David> if we as a Community invalidate experimentation by
    David> businesses that think they can innovate and be profitable,
    David> we'd be stifling any further innovation by limiting what is
    David> "allowed" beyond those 4 (or 8) business practices
    David> explicitly sanctioned by the text of the GPL.

I think you're projecting here.  Nobody has said "Only if RMS would
like it is it an FSB."

While I don't want to restrict FSB to the roles enumerated by RMS in
his various writings, I do insist that if your profit comes from
selling non-free software you are not an FSB.  Not even if your
operating profit is used to subsidize substantial amounts of
(loss-making) FS development---in that case you are a proprietary
software business philanthropist, specializing in FS philanthropy.  We
know how long that philanthropy will last if you fall into the hands
of a conventional VC, right?

I'm more than happy to give honorable mention or even honorary FSB
status to companies like TrollTech, where their primary product is
both free on Unix and probably freely available on Windows via Cygwin,
although their Windows license is proprietary, and Aladdin, which if I
understand the story right would have accepted the GNU GPL as the
primary public distribution license if RMS had permitted exclusion of
the hardware companies which, although they may have distributed
source, were distributing immutable binaries and thus (IMO) do not
qualify for GPL privileges in spirit.  But unless the free software is
what pays for the early retirement of the proprietors, those are
technically not FSBs IMO.

I'm also willing to give a _lot_ of credit to firms like those and
like SleepyCat (I'm mentioning these firms because I've studied their
licenses and practices, not because I think they deserve more credit
than anyone else) because they have gone and (1) given away their
software to very large numbers of potential retail customers under
nearly free licenses (by which I mean they allow everyone to see the
source, and most of the customers by head count to freely modify and
redistribute the software) and (2) become their own toughest
competitors by releasing their previous versions of their commercial
products as free software.  This (to my mind) clearly distinguishes
them from say RSA where a reasonably good free library (RSAREF) was
publically available, but the "good stuff" never was.[1]  Thus RSAREF did
not have the educational value (sharing) that GNU Ghostscript, GPLed
Qt, or Berkeley db v. (current-1) have.

If you want to weaken the definition of free software business, IMO
that's the way to go: measure the freeness by the degree to which the
software product achieves the goals of sharing among programmers,
education of those interested, and modifiability/repairability to suit
the needs of users.  Not by "factoring in [business success]":

    David> reserving our subjective judgements about what is right and
    David> wrong, good or bad, until the successes and failures of
    David> meeting *business* goals can be factored in.

We will always be able to find businesses that work to satisfy
business goals, sacrificing "lifestyle" and "ethical" goals.  If
satisfying portfolio value growth goals is incompatible with being a
"women's business mutual fund", a "black-owned business mutual fund,"
or a "green mutual fund", I think that's a pity, but no excuse for
changing the definition of those ethically constrained portfolios.

Similarly, the success or failure of FSBs as businesses is no reason
to change the definition.

    David> if they *are* in the black (even as a lowly lifestyle

Lifestyle businesses are a symptom of _highly_ evolved economies, not
the reverse.  Lifestyle businesses are being excluded because
typically proprietors of successful one are pretty stellar, either as
businessmen or developers (or whatever they do).  They can afford it.

What we want to know here is whether "FSB" is a viable business niche
for people who are not yet assured that they can afford to send any of
their kids to Stanford.  That wealth can be generated either as 100s
of proprietors or (as LMA keeps saying, I think rightly) as employees
on a large (100s/firm) scale.  But a few examples of successful
lifestyle FSBs and a bunch of profitable "not even close" wannabes is
not what we want to see.

    David> business), would you not want to at least mention
    David> businesses like theirs in your article, if only as an
    David> example of one extreme on the spectrum of possible
    David> experiments now taking place to try to make money in the
    David> Free Software world?

If it were my article, as I understand their business, the answer is
"No chance."  They're also "in the black" in the sense that they're
not part of the rainbow at all.


Footnotes: 
[1]  This may not be a good example, several people have testified on
this list that they consider implementing RSA a trivial (just kidding)
and straightforward task.

-- 
Institute of Policy and Planning Sciences     http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
 My nostalgia for Icon makes me forget about any of the bad things.  I don't
have much nostalgia for Perl, so its faults I remember.  Scott Gilbert c.l.py