[Fs-phil] Defining software freedom, and a sidetrack on embedded systems

[Stephen J. Turnbull]

Per Glenn's suggestion, moving to an alternative venue.
Grovel the List-* headers for how to subscribe, etc.

Please trim!

Jambunathan K writes:

 > > This is a serious philosophical problem for the movement.  The
 > > movement's propaganda equates "software freedom" with "freedom of
 > > speech", but in fact they belong to different classes.
 > 
 > Somehow I think that the last paragraph is incomplete.
 > 
 > Could you please clarify to what class "software freedom" rightfully
 > belongs in the way you see it?

Sure.  Software freedom as espoused by the Free Software Foundation
("morally speaking, all software should be free") really is for
hackers.  Nobody else cares, and it really doesn't affect their lives,
except economically, which is irrelevant to the issue of software
freedom as such (in other words, bitching about monopoly is an open
source issue).  As I argued in the previous post, it *can't*,
especially now that most applications come in free versions, or are
embedded and unhackable by non-specialists anyway.  That is very
different from free speech, where the right can be exercised by any
halfwit blowhard, and frequently is.

And free speech *does* matter to your daily life in your relation with
the government, especially if you want to translate the names of
people killed in an earthquake into roman letters for the benefit of
people abroad whose PCs don't speak Japanese (did that, a long time
ago, and got a cease-and-desist lawyer letter from the TV company
whose broadcasts were the native source of the names for my pains), or
if you live downwind of a meltdown (I do, and the fact that people
living near me freely published Gieger counter readings made staying
in Japan thinkable -- unfortunately for those *near* the Fukushima
power station, there weren't any national universities or research
institutes downwind, and the power company and the bureaucracy
suppressed the readings they had), or if you make enough money to
interest the tax authorities (I don't, but if I did you can be sure
I'd lobby the government to give me a break).

This is not to say that eliminating software patent and software
copyright might not be an excellent idea (I personally am in favor of
the former and on the fence about the latter), just that I really
don't see in so-called "intellectual property" the kind of government
infringement of individual rights that are dealt with by the U.S. Bill
of Rights and similar legal provisions.

 > GNU/Linux is becoming particularly popular to build sophisticated
 > embedded systems

Yes.

 > They also provide drivers in the kernel and distribute it alongside the
 > vanilla kernel. The distribution of driver and their apparent upgrades
 > is a mere posturing by these companies to "express solidarity with the
 > GNU/Linux" system.

I doubt they're expressing "solidarity".  AFAICS they're just doing
the minimum they think is needed to avoid getting trashed the way
nVidia does in the community.

 > In truth, these drivers do not exploit the full funcitonality of
 > the underlying hardware and are often very poor substitutes for
 > their "commercial" linux offerings (Think HP and IBM here).

So what?  You want commercial grade drivers, buy the OS.  I bet you
get source then.  Or write them yourself.

 > Does GPL lays emphasis on the right to distribute or a (mandatory
 > contractual) obligation to redistribute? Note the subtle difference
 > - the right to distribute would mean the right to not distribute as
 > well.

It's not a question of "emphasis".  The GPL imposes the obligation to
distribute a reasonable form of source if you distribute any form of a
GPLed program.  If you have used any part of a GPLed program in your
program, your *whole* program is a derivative, and you must do one of
three things:

1) refrain from distributing your program at all
2) get a license permitting you to do what you want to do from the
   copyright holder of the GPL program, or
3) distribute the whole source of the program to those you have
   distributed any form of the program too.

There is some legal question about definitions of "derivative" and of
"distribution", but mostly they probably mean what you think they do.

However, in the case of kernel drivers, Linus has explicitly stated
that kernel drivers that access only "public APIs" of the kernel are
"really" extensions of the firmware, not the kernel, and do not incur
GPL obligations.  Viz nVidia.  This does not apply to private APIs or
internal variables of the kernel; if you use those, the GPL bites.

 > Law could be lax on individual users should it also be lax on such
 > enterprises, which in my view, playi the devil and actually "hoard
 > and piggback" on community produce and accumulate humongous private
 > fortune.

It's not the law, it's Linus who is lax.  He considers it an issue of
"consenting adults" if users want to use hardware accessed via
proprietary drivers.  He has publicly stated that he does not require
adults to be sane or competent; he thinks that consenting to such
drivers is stupid.  The FSF's lawyers, however, say that kernel
drivers are derivatives of the kernel and therefore covered by the
GPL, absent this special exception.

Richard Stallman is also lax on the issue of firmware etc.  That is
the issue that drove Ghostscript out of the free software fold.  Peter
wanted an interpretation of "software" that excluded firmware not
upgradable by the user so that he could use the GPL for Ghostscript as
software, while charging for use in embedded systems like printers and
fax.  Richard told him to take a hike, this is a desirable use of free
software.  So Peter hired a lawyer and switched to the Aladdin license
from the GPL for current release, then switched each release back to
GPL after one year.

(Richard will tell you that there are legal complications in trying to
make this distinction, I imagine, and there probably are.  Cf. how
complicated the discussions around the Affero clause and the anti-
technical-means clauses in GPL v3 were.  Nevertheless, he also told
Peter that what he wanted to do violated the definition of software
freedom.)

 > > And of course this assumes that we concede to RMS the right to define
 > > "software freedom".  
 > 
 > Some one has to define it.

Sure, the common use definition needs to be done by someone, and
Richard's is a good one.  That's not the point.  The question is "when
I call myself a free software advocate, what do I mean by that?"

 > > Many of my friends do not, and use less stringent definitions (ie, not
 > > requiring redistributability).  It is fair for you to complain that
 > > this is not "true" software freedom, but that misses the point.  These
 > > folks are "almost there"!
 > 
 > I have highlighted aspects of "requiring redistributability" in the
 > above para and contrasted that against "right to redistributability".

Above you talked about the GPL.  The GPL is only the restrictive
extreme of free software licenses, but there are a wide variety of
more permissive licenses.  The definition of free software does not
require redistributability of downstream, but it permits it.  Some
kinds of free software allow fully proprietary redistribution, with
no obligations at all on the licensee, except those that indemnify the
licensor and protect her reputation, and "author's rights" (eg, to be
known as the author).

The "less stringent definitions" referred to above are nonfree because
they allow licenses *prohibiting* redistribution, as long as the user
gets source, and permissions to make modifications and local copies
for one's personal (including corporate personal) use.  Everybody
acknowledges that such licenses are non-free.  The folks I'm referring
to simply try to have it both ways: those licenses do not cause them
to get their shorts in a knot (although they avoid products carrying
such licenses to avoid lock-in, etc), but they want to consider
themselves advocates of software freedom, too.

 > > It seems likely to me that they are fairly easy to persuade to accept
 > > the full definition.  
 > 
 > I don't think the embedded companies

Please READ WHAT PEOPLE WRITE, and respond to that.  In the sentence
above, "they" refers, first, to the people I drink beer with, and
second to others in that general neighborhood of the spectrum of the
open source community.  NOT to "the embedded companies".

If you don't want to reply to what I wrote, I have no problem with
that.  Write what you want to write.  But don't quote my words if
you're just going to ignore them.

 > I believe Free Software makes for a stronger case in conjunction
 > with Free/Libre Hardware Designs. I think a "Category Killer" in
 > the Open Hardware is all that is needed to tilt most of the "Open
 > Source" camp in favor of "Free" camp.

Unlikely IMO.  I suspect it would rather have the opposite effect,
attracting a lot of people who would otherwise say "it can only be
done with proprietary systems" to the "hey, open source can be all of
cheap, good, soon!" point of view.  Some might move toward the
position I currently hold ("there's probably a theoretical
modification of the patent system that would create a better
environment for software users and developers than eliminating
software patents altogether, but it isn't ever going to get enacted,
so let's get rid of software patents").  But all of this is about the
tangible benefits of free software, not about software freedom, as
David would say.

So I don't see how it would attract them to the "all laws creating
intellectual property are morally wrong" position.  The "I created
program X, I should have a say in who uses it and how" feeling seems
to be instinctive, and most people seem to both feel it and sympathize
with others who do.