RE: Linux... OK but definitely second best

[Matthew Benjamin <MBenjamin@comshare.com>]

(Thanks for letting me post once again, Jaime.)  

> Brett, much as I admire your position on Microsoft, your claims about 
> 
> 1. the motives and even the power of the FSF
> 2. the "consequences" of the BSD License
> 3. the "consequences" of the GPL 
> 
> are simply counterfactual.  (Read, as Jamie put it, "plain wrong.")  (No
> offense meant.)
> 
> 1.  The FSF has almost no power or, I suspect, even influence on Linus
> Torvalds.  In fact, although things seem
> to have warmed up in the last couple of years, Richard Stallman (RMS) has
> been an outspoken critic of the Linux movement,
> which he claimed unfairly leveraged the GNU toolchain (GCC, etc), without
> proper credit.  Linux, he said in a lengthy essay, should
> be called by eveyone "GNU/Linux."  
> 
> 2/3. The argument that the BSD license makes software categorically "more
> free" is unfortunately specious, and the vehemence with which you are
> making it is 
> confusing to non-experts, and annoying to the rest of us.  
> 
> Those familiar with the GNU world know that there are in fact two GNU
> licenses, one of which, the LGPL (Library GPL), is expressly designed to 
> allow non-free software to link dynamically or statically with
> pre-compiled, GPL'd software.  Further, there is another group of
> softwares, eg, GNU Bison, which since it generates C program code, comes
> with an explicit license provision stipulating that generated code can be
> encorporated into any software, free or non-free, without changing its
> license in any way.  Thus GNU Bison-generated parsers can be legally used
> in commercial program code, and often are.
> 
> The most questionable claims you make are about the "long-term
> consequences" of pursuing these different licenses.  Such claims are
> dubious on their face, 
> but are further weakened by the obvious counterexamples in the history of
> free software since its beginnings in the early 1980s.  
> 
> First the GPL.  Whereas the GPL is supposed to "encumber" code and make it
> into the property of the FSF, the raw quantity of code available under the
> GPL has risen steadily since 1983, and better than geometrically since
> 1993.  Even a partial list of GNU projects explodes this claim (GCC, GNAT,
> Guile, F77, Small-Eiffel (programming languages); Linux, HURD, GNU Mach
> (Operating System); Emacs, VIM, TeX (Text Processing); XFree86, Fvwm, KDE,
> GNOME (X-Windows graphical environments) -- and so many others, and so
> many coming into being every week that the actual number is almost
> uncountable.  If these projects could be counted exhaustively, they would
> total tens of millions of lines of useful, and widely utilized, program
> code.
> 
> Now the BSD license.  It doesn't take long to see that there are many
> fewer projects using the BSD license than use the GPL.  Although that
> is not necessarily a measure of its value, it bears consideration.  
> 
> On the other hand, the BSD license is noteworthy primarily as a
> "transitional" license for code developed at university, but enhanced and
> "commercialized" by 
> private firms.  It is difficult to see this as negative, yet it is also
> difficult to argue that that this process promoted the evolution of the
> "free" software.  Rather, it heavily  favored the commercialized source
> base.  I know of no case where commercialized BSD source code has been
> returned to the public. The BSD license has proved to be an effective
> vehicle for transferring development of a variety of projects from the
> university community to private enterprise.  Examples are:  UNIX, INGRES.
> The INGRES databases for example, all commercialized as very costly
> software indeed.  And, of course, none of the brilliant enhancements
> produced by commercial firms working with INGRES have made it back into
> the source base.  Similarly, although since 1991 or so the 386/BSD UNIX
> family has evolved as a development community not unlike Linux, this has
> not always been the case.  Several BSD UNIXes have been fully
> commercialized--eg, ULTRIX, BSD on NeXT and OSF/1, AIX--and users
> experiences with them have not resembled those of open-source UNIX users
> in the least.  Rather, they were identical to those who purchased
> commercial System V UNIX.  Ongoing development of UNIX proved expensive,
> so while these commercial vendors enjoyed collecting license fees, they
> did not continue to evolve UNIX, and its commercialized licenses prevented
> the many interested parties from doing so.  One result of this was the
> rise of a series of Microsoft OS monopolies, where each should have failed
> on technical merits.  
> 
> On the other hand, two of the most popular Internet protocol servers, BIND
> and Sendmail, are Berkeley licensed.  And, the PERL environment is
> licensed in a manner similar to, but not identical with, the Berkeley
> license.  Obviously, it is possible to build a free software community
> around such code.  Would such efforts have come into being without the
> influence of such FSF projects as GNU GCC and EMACS?  It seems unlikely,
> but no one can say for sure.  Possibly the best example you could leverage
> is that of X-Windows, which for more than a decade remained free and
> open-source, under a Berkeley-like license.  Unfortunately, this example
> is significantly weakened by the recent decision of the X-Consortium to
> encumber the latest X-Windows release, X11R6.4, with a paid commercial
> license.  (According to the X/Open www site, a commercial source license
> for X11R6.4 costs $30,000.)  For this reason, the XFree86 project has
> obviously chosen to continue to develop from the X11R6.3 base.  The new
> license resembles that of DCE/DFS, the next generation computing
> environment that, it now seems, will not have any significant role in
> ongoing distributed computing efforts--except, in modified form, as the
> transport layer for Microsoft's proprietary DCOM protocol.  (Microsoft
> calls this, "Microsoft RPC.")
> 
> My Own Opinion:  The value of free software inheres in the _idea_ of
> making, and keeping, software free--not in the specific license that
> protects it.  Nevertheless,
> this idea that software should be free was first articulated not by
> Berkeley researchers--many of whom are now wealthy members of corporations
> like Sun Microsystems and Informix---but by Richard Stallman.  And the
> rise of Open Source software today has coincided with the growing success
> of a symbiosis
> of free software projects, the most important are:  GCC, GNU C Libraries,
> Linux, *BSD, XFree86.  The rise of Open Source has also ushered in a true
> rebirth 
> of UNIX--made possible by free access to the source.    
> 
> As you see, I have presented BSD and Linux development as a symbios, not a
> conflict.  That is not wishful thinking, those are the facts on the
> ground.  In 
> spite of the differences in license specifics, Linux and *BSD are
> influencing each other in productive ways.  Further, the developers of
> these systems are
> sick and tired of arguing about which OS to develop for.  Jordan Hubbard,
> the FreeBSD project leader, has said as much in unequivocal terms.
> FreeBSD, Linux, who cares?  They are both freely extendable, and free for
> commercial use.  Also,  *BSD and Linux are BOTH unequivacally freely
> usable as a springboard for a virtually any kind of commercial development
> or portation of commercial software.  This doesn't pollute their freedom,
> and it doesn't pollute corporate code.  It is 
> a win/win.
> 
> We could argue about the Linux and BSD development models all day long.  I
> used to agree with you on this--specifically, BSD has clearly been more
> stable.  But if FreeBSD has more stability, it lacks a list of features as
> long as my arm--all of which exist in Linux.  But again, this is not
> relevant.  As free UNIX continues to evolve, FreeBSD and Linux have many
> opportunities to challenge and enhance one another.  In response to your
> complaint that Oracle has not ported to *BSD, let me point out that Oracle
> has ported to *BSD in the past.  They might do so again, when *BSD has the
> base POSIX features (asynchronous I/O, SMP, kernel threads) that make
> Linux a viable target for their software's current source code base.
> 
> Therefore, this argument is silly.  Make free software, under whatever
> license you favor.  Tell your family, friends, and your boss about it.
> That is creative use of 
> your freedom--and that is what I think this movement is about.
> 
> 
> Matt Benjamin
> 
> 
> 
> 
> 
> 
> 
> 
> -----Original Message-----
> From:	Brett Glass [SMTP:brett@lariat.org]
> Sent:	Monday, July 27, 1998 11:18 PM
> To:	love@cptech.org; Multiple recipients of list AM-INFO
> Subject:	Re: Linux... OK but definitely second best
> 
> At 10:46 PM 7/27/98 -0400, James Love wrote:
>  
> >	The people who do the GPL software are every bit as hard working and
> >smart as the people who make software with a profit motive in mind. 
> >They are entitled to attach conditions to their software.
> 
> You're absolutely correct. But the problem with the GPL is that it
> attaches
> conditions that, in the long run, will have negative effects. Therefore,
> it is
> important to ask them not to publish their software under that license. I
> believe that if they truly at the big picture -- understanding both the
> true
> philosophy that spawned the GPL and the negative results it will produce
> --
> they will consider using a license that allows commercial re-use of their
> code.
> 
> >Why is it
> >"good" for a proprietary copyright that restricts reuse, because we love
> >profit making enterprises so much, but "bad" for developers to dedicate
> >the fuits of their labor to the general community of programers and
> >users?  
> 
> Unfortunately, the GPL doesn't dedicate their labor to the "general
> community"
> of programmers and users. It leaves out a specific group: those who would
> like
> to use it as the basis of commercial software. 
> 
> The most benefit would be realized if the base of open source code is a
> truly
> universal resource -- a "public library" of classic code that can be used
> by
> anyone for any purpose. This is akin to saying that individuals can use
> the 
> material they public library, but businesses can't -- or that individuals
> can quote Shakespeare, but someone making a speech in a business setting
> cannot.
> As one of the guiding forces behind a movement to protect the public
> domain,
> you can understand, I'm sure, why such things are important. We mustn't
> FORCE
> code out into the open, but should encourage its completely free use once
> it's 
> there.
> 
> >> Again, I must disagree. The BSD license is better in all cases, since
> the
> >> GPL's sole purpose is to reserve power for one specific faction (more
> below).
> >
> >    Brett, I don't see how you can be more one sided than this.  "better
> >in *all* cases" ???    Is this the same as better because it is defined
> >to be better?
> 
> No.... I've given a great deal of serious thought to this issue, and can
> envision 
> no situation in which the GPL would produce a better long-term result than
> a 
> license that allowed commercial re-use. 
> 
> >This "one specific faction" stuff seems a bit bitter. 
> 
> Nope, it's reality. The FSF, on its Web pages and in public speeches,
> derides
> all other forms of open source software. It thus positions itself as a
> faction
> that is opposed to all others.
> 
> >Aren't these people using their own private property in the way that
> >they are entitled to?  
> 
> Microsoft makes that same argument for its actions. ;-) Seriously, though:
> of 
> COURSE these people are entitled to do what they want with their
> intellectual
> property. However, we should ENCOURAGE them to do what produces the
> greatest
> long-term benefit for all concerned -- including them.
> 
> >Even if some software firms are unhappy about the
> >restrictions, it doesn't mean the GPL restrictions are not rationale or
> >appropriate. 
> 
> The reason that the GPL restrictions are inappropriate is not that "some
> software firms" are unhappy with it, but rather because -- if you look,
> again,
> at the big picture -- they are counterproductive. They stifle
> productivity,
> innovation, and entrepreneurship.... Which are not things that I believe
> any of 
> us want.
> 
> >	Not everyone shares your opinion on the dangers of embrace and
> extend
> >strategies, and after watching what MS is trying to do to Web protocols,
> >who can dissmiss the danagers?
> 
> Open source is not more or less subject to "embrace and extend" strategies
> than any other type of software. If Microsoft, for example, really wants
> to create
> proprietary extensions to a protocol, it can do that with or without open
> source. 
> 
> Historically, however, it has been shown that by providing the open source
> to 
> commercial software vendors for re-use, one can establish the standard so
> solidly 
> that it's tough for ANYONE to create incompatible versions. This advantage
> does not 
> occur unless the commercial vendors can use the source. If they must
> reimplement, 
> they're tempted to tinker and often add incompatibilities -- if for no
> other reason
> than that they've interpreted the standard in different ways.
> 
> But if the code *is* the standard, this doesn't happen; everyone just uses
> it, and
> interoperability is greatly enhanced. This is what happened with BIND, the
> Berkeley
> Internet Name Daemon -- it literally holds the Internet together.
> 
> 
> >    This is plainly wrong.  THe FSF doesn't even control the
> >distribution of its own software, and it clearly doesn't control Linus
> >Torvald.  Even Linus doesn't control access to Linux anymore.  He shares
> >the copryight with 50 or so developers who have made various
> >contributions, and everyone one of them would have to agree before the
> >license was changed in a significant way.
> 
> This is part of the danger. The GPL "locks itself in" in a way that makes
> it
> difficult to undo. It's tough to veer away from the FSF's agenda later,
> once
> it becomes apparent that it would have been better to allow commercial
> vendors
> to use the code.
> 
> >>What's more, it seeks to tilt the playing
> >> field away from ALL commercial software vendors.... Not a good plan.
> Commercial
> >> and free software can, and should, always coexist.
> 
> >     Not true in reality, as the rash of commercial announcements for
> >Linux products demonstrates.  Heck, some of the Linux Windows manager
> >programs are not subject to GPL, and much of the software that is used
> >in the Linux distributions comes from a variety of licensing schemes.
> 
> These are not counterexamples, because they do not incorporate the GPLed 
> source code in the commercial code. They can't.
> 
> However, there are good examples of positive effects from the ability to
> include OTHER open source code in one's own products. BSDI is one example.
> And
> there are the many embedded systems that use code from Apache, FreeBSD,
> BIND, and
> other programs that are published under more flexible licenses -- for
> example, 
> Internet routers for small offices. Or print servers small enough to clip
> onto
> the back of the printer and can be controlled via a Web browser.
> 
> >	Everyone was very happy when Corel and Oracle made their linux
> >announcements.  
> 
> Not quite "everyone." I'm sure Microsoft wasn't happy! And as a FreeBSD
> user, I 
> would of course rather not run these companies' products under Linux
> emulation.
> I think users should have a choice of OSes in the open source world as
> well as in
> the world of commercial software! In fact, I think it is just as
> dangerous, in 
> the long run, to have these vendors embrace only one open source OS as to
> have them
> embrace only one commercial OS. As you and Ralph have said many times,
> users should 
> have a real choice.
> 
> >> In an earlier message, Jamie expressed concern that open source
> software could
> >> be "monopolized" too. Well, it can -- and the GPL is the mechanism via
> which this
> >> particular faction seeks to do it. As the song so brilliantly puts it:
> "Meet the
> >> new boss.... Same as the old boss." The "revolutionaries" of the FSF
> are attempting
> >> to fool us again.
> 
> >       Except, they aren't bosses.  Brett, for all your bitterness
> >toward Richard Stallman, 
> 
> I'm not "bitter" toward Richard Stallman; in fact, I take great pains to
> correctly represent his views and aims. It is true that those views and
> aims
> are extreme, but (as you know) even Richard admits that.
> 
> >how many shots do you think he calls in Linux
> >development?  The people calling the shots are the people who write the
> >code.  For once.
> 
> It seems that some developers are allowing the FSF to call the shots
> vis-a-vis licensing, and this is dangerous. I believe that they should
> reconsider 
> their actions and release their code under other licensees.
> 
> >	You don't have to be a communist to appreciate a little
> non-commerical
> >space in your life.
> 
> Agreed.
> 
> >I think the Internet, which is based upon open
> >public domain standards and no "ownership" of the net itself, shows how
> >our lives can be enriched by public infrastructure.
> 
> I agree with this as well. Let's start building that public infrastructure
> in a
> way that allows everyone to benefit and excludes no one. Instead of
> closing the 
> "public library" of code to commercial traffic, why not open it up to all
> comers?
> 
> --Brett Glass
>