On Wed, May 05, 1999 at 02:27:07PM -0500, Manoj Srivastava wrote:
> We have lost the opportunity to get a foothold in yet another
> area. I think it would be arrogant to assume that we have won the
> battle, and that we should start coercing people just like MS does.
While I have withdrawn my proposal, I cannot in good conscience let
distorted characterisations like this proceed unchallenged.
Our classification of some software as being in contrib versus main is not
coercive in any way.
All software in contrib is DFSG-free, and this dead proposal would not have
changed that. We do not have legal authority to change (or even, in almost
all cases -- since Debian developers or SPI don't hold the copyright) to
enforce the licenses of the software we package.
It is only within the power of the license to perform "coercion" in any
legal sense. To "coerce" people outside the bounds of the software license
would require illegal activity on Debian's part (e.g., extortion -- "if you
put tik on this CD-ROM which also contains Debian's main, we'll shoot you
dead.")
Please tell me how we would in any way restrict the activities of people
placing software of any kind, let alone Debian packages, on FTP servers,
web pages, CD-ROM's et cetera. We may withhold the privilege of
identifying a product as "official Debian", true, but this is our right as
trademark holder. If you do not feel Debian should exercise this kind of
control over usage of its name, that is a completely separate issue from
how we decide what goes into main and what doesn't.
Debian has *the right* to include in our distribution whatever we choose,
within the constraints of the law (we can't ship Windows 98). We are a
single vendor, and not even the largest, in a fairly large and highly
competitive market. Even from an anti-trust analysis, there is simply no
way that our selection of packages in the main ("official") distribution
plausibly impacts the ability of the individual to obtain a piece of
software, even in .deb format. How you deduce coercion from liberty
escapes me, and I would appreciate a careful explanation.
We might not ship contrib or non-free on our Official Debian CD images.
So what? There are plenty of vendors out there who can make unofficial,
derivative CD images and place on those CD's whatever they want, within the
license terms of the individual packages. As it stands, they are free to
redistribute the contents of main and contrib in any way they choose,
because the DFSG ensures that the software so contained grants them that
freedom. As long as main and contrib continue to be defined as
containing only DFSG-free sofware (quite apart from any other conditions we
use to distinguish them from each other), this is not going to change.
People have carte blanche to archive, mirror, burn CD's and such however
they please.
DFSG-free software leaves in the hands of the individual, or any autonomous
organization, the RIGHT to redistribute and rearrange as they please. They
may have to rename something that they have modified, a la the Artistic
license or "official Debian" versus "unofficial Debian", but we have long
considered this restriction to not be onerous or contrary to the principles
of free software. Are we to start now?
We do not have a gun to anyone's head. From a licensing standpoint,
software in contrib is above reproach, and as long as we continue to define
contrib as containing DFSG-free software, this will be the case (unless we
abandon the DFSG).
Why do we have a contrib at all? The cosmic order of things knows no
fundamental difference between static linking, dynamic linking, wire
protocols, client/server interactions, file formats, or any such minutae.
These issues are simply not significant in and of themselves.
Unless I misunderstand, contrib exists because we recognize that not all
DFSG-free software is of equal utility in building a system based on free
software. From a point of view that evangelizes free software (as RMS
does), or simply one that wishes to work in an environment where non-free
software is excluded (as many of us do), some DFSG-free software can be
seen to better promote the further use, spread, and development of free
software than other DFSG-free software.
The existing distinction between main and contrib is in practice confined
to programs which, at run-time, use non-free software (such as non-free ROM
image or library).
This distinction is not fundamental. It is arbitrary.
Is it easy to evaluate? Usually. Would other distinctions be less easy to
evaluate? Almost certainly.
But what is our guiding principle in making this decision? This may
perhaps be just yet another manifestation of the ESR/RMS conflict.
ESR: Free software is good because it does its job better than non-free
software. (Though he doesn't use the term "free software" anymore.)
RMS: Free software is good because it is free.
ESR is primarily concerned with practical issues, and this concern has led
him along a predictable path in that light. If you were to come up with a
development model that quantifiably produced better software, in terms of
usability or whatever other practical criterion you wanted to adopt, and
still kept the license of the software proprietary, this would be a
fundamental blow to ESR's advocacy (at least, so I have been led to believe
from his own statements).
RMS is first and foremost concerned that our computing environment is
unfettered by traditional, closed-door (or expensive-access) approaches to
intellectual property in general; it is not just an issue of software
licenses. To observe this fact, read the articles in the philosophy
section of the GNU website.
A person who sees the issues from a position more sympathetic to ESR's
position would, logically, be opposed to any distinction between main and
contrib at all (assuming the DFSG is consonant with his understanding of
free software, which for the most part we can consider it to be). All
DFSG-free software should be in main because it can all benefit from the
development model of free software. And the more software we have in main
the better, because it is thus more likely that the user will be able to
find a free tool to do what he wants (I will not argue that Debian packages
outside of main may be harder to stumble across, but even so they are
hardly kept secret).
On the other hand, RMS and those in sympathy with his ideals have a loftier
goal. It is very good to license your software freely (even better to do
so under terms which assure that it and its derivatives will stay free,
hence the GPL), but this is not the fundamental issue. Nor is the manner
in which the software is developed (cathedral, bazaar, random code
generator, etc.). Even more important than good and effective software
licensed freely is the goal of having a completely free computing
environment. That means completely open hardware specifications, BIOSes,
firmware, flash-reprogrammable devices, et cetera. From this perspective,
it can be seen that free implementations of propritary interfaces
(especially when it is forbidden by patent law or the like to
reverse-engineer or compete with these implementations), are less desirable
than a free implementation of a free interface.
Perhaps you personally are more closely aligned what I have identified
(hopefully not unfairly) as the "ESR philosophy" than the "RMS philosophy",
and therefore, you desire the arbitrary distinction between main and
contrib to permit more, rather than less, software into main. But there
may be others that disagree with you, and I feel it is irresponsible of
you to cloud the issue of an arbitrary distinction with charges that a
change in that distinction will somehow be a manifestation of coercive
action.
Debian does not have a monopoly, or anything close to it, on the
communication channels between developers and users. We do not control the
Internet. We are not even all that influential among the distributions,
a market which is increasingly more dominated by corporate interests.
Therefore, since we do not have general control over production of
software, and since we do not have general control over its means of
distribution, we cannot, even accidentally, exert a coercive influence on
this industry. At least, not without resorting to actions of the most
extremist and criminal sort, such as putting prices on the heads of people
we don't like and advocating their assassination. You and I both, I am
sure, agree that such activities are unbeliveably far removed from the
principles of Debian developers as a whole, and likely to each individual
developer, to a man.
So let us cease this unwarranted use of the term "coercion". What we CAN
do is withhold our endorsement from certain kinds of free software, and
this we *already* do (last I checked, there were more than zero packages in
contrib).
The change in the contrib policy that was proposed, as far-reaching as some
people may think it is, simply was not fundamental. Important enough that
it lies outside the purview of policy, yes; and therefore a topic more
appropriately broached on debian-devel-announce and debian-vote if it were
to be put to a consensus, and not handled solely within the policy group.
Again, it's not fundamental from the perspective you are approaching it.
Even if we *randomly* selected 1000 packages from main and tossed them into
contrib instead, this would not be a coercive action.
Would it be detrimental to the utility of our distribution? Most probably.
Would it be a PR blunder? Almost certainly.
Would the selection criterion be open to fair criticism? Absolutely.
The above questions are completely relevant to proposal of redrawing the
line between main and contrib as well. But suggestions of coercion are
utterly unfounded.
> But not quite enough. When Linux has 25-30% of the desktop, we
> may be able to coerce people like microsoft deos, "Do it my way or
> else". I don't think we can do that yet. I also don't think we should
> do it ever.
Putting something in contrib is simply not telling people "do it our way or
else." Microsoft doesn't put things on their web site that aren't "their
way". Microsoft doesn't let vendors add or subtract at will from the
distribution of their operating system or application software.
There is no "or else". Even if we removed contrib entirely from our
official FTP sites, mirrors, web pages, and what have you (let alone our
official CD's), people who wanted the software would simply say "f--- you"
and go get it from the original author, another distribution, some archive
on the net, repackage it themselves, share it with friends, et cetera. The
licenses of the software in contrib assure them this freedom, just as they
do in main.
> I really would not like to think we would ever think we are
> strong enough to start working agsainst giving the users
> functionality no, and to piss off writers of DFSG free software.
We already work against giving the users functionality, assuming the
existing contents of contrib can be adjuged to be functional. We have
pissed off writers of DFSG software in the past. We will probably do so in
the future. Unless we want to craft the distinction between main and
contrib based foremost on the desires of the authors of DFSG software, we
cannot let popularity considerations override the criteria we have
selected.
> Marcus> The question is if we can continue to get stronger
> Marcus> indefinitely by embracing the proprietary world, or if it is
> Marcus> time at one point to actively go against proprietary
> Marcus> protocols and software patents. The other question is when
> Marcus> the time has come if it is time to fight.
>
> Right now I think we have a hrad enough time getting to the
> desk top to start the fight.
That is your personal assessment; it implies to me that you think the day
may come when taking up such a "fight" is appropriate, but that seems to be
contrary to what you have stated above. Could you elaborate on this?
In the course of this message I have enumerated at least three criteria
that I regard as important and relevant factors in the weighing of such a
decision, and if the proposal succeeded or failed in overwhelming
proportion due to sound judgements on those factors alone, I would not be
upset.
Of course, I cannot be the sole arbiter of what are considered to be
germane critera for assessment of the issue, but I can appeal to standards
of logical and rhetorical analysis that are literally millenia old. I
submit that we evaluate the proposal based on existing, well-known
principles of logical discourse. If someone wishes to argue with me about
selection criteria for arguments for or against the proposal, they are
entitled to do so; but in the course of holding that argument I will insist
on abiding by the rules of logic, so I must plead down to my axiom of
argumentation :).
What distresses me is the widespread and sometimes thoroughgoing (on an
individual basis) misinterpretation of the proposal and its ramifications,
sometimes to a degree that seems to me grounded in hysterics. This is not
how intelligent people resolve disputes. Intelligent people resolve
disputes by rationally analyzing them, and/or by compromise.
Aside from the objections you have raised (which I quoted), the other
argument against the proposal I have seen which I regard as irrational and
ungrounded is the protest that this will be a slippery slope and we will be
forced to throw LILO (for instance) out of main because it depends on
non-free BIOSes. My rebuttal is that the distinction between main and
contrib is FUNDAMENTALLY an arbirtary one. It is not a slippery slope, it
is a line in the sand. We can draw that line where we choose; in time the
tide may come in and wash it away, and we must draw it anew, perhaps in a
different place.
There was one other irrelevant objection, and that was personal offense
taken to the specifics of James's action. I hope I don't really need to
belabor the point of how utterly non-germane that is to the issue. If we
take what we proclaim to be a principled action, all the other people in
the world who share that principle are not logically impelled to support
the action. A person may go on a rampage of serial infanticide, and
proclaim that he did so because Thomas Malthus was right and we've got to
get the population under control. People can agree with the principle of
controlled population growth without lauding the actions of a serial
murderer. So please, let us stick to the point. One can agree or disagree
with what James did, or hate his guts, but all that really should have
absolutely no bearing on a person's assessment of the merits of the
proposal in question.
[DISCLAIMER: I apologize in advance if I have misrepresented the
philosophical stances of anyone I have mentioned in this message. I
believe I understand those of ESR and RMS to the extent that I have used
them as examples; I do not pretend to have a great deal of understanding of
James's or Manoj's personal philosophies, and have tried not to put words
in their mouths. Otherwise, my characterizations of nonspecific
individuals are derived from what I have observed in discussions of this
matter on the debian-policy mailing list and in the Debian developers' IRC
channel.]
--
G. Branden Robinson | Damnit, we're all going to die; let's
Debian GNU/Linux | die doing something *useful*!
branden@ecn.purdue.edu | -- Hal Clement, on comments that space
cartoon.ecn.purdue.edu/~branden/ | exploration is dangerous
Attachment:
pgpEWOKtYKK_l.pgp
Description: PGP signature