Why Torvalds is sitting out the GPLv3 process (Linux.com)

What makes someone a Linux Hacker? Right, if he contributes to linux.
What makes someone a GNU Hacker? Right, if he contributes to GNU software.
If the GNU or Linux hacker is employed by any IT company or if he does it in his spare time or if i believes in the words of Linus Torvalds or in the words of the FSF doesn't make any difference. If he contributes to Linux, he is a Linux Hacker and the result will always be linux and if he contribute to GNU he is a GNU Hacker and the result will always be GNU.

So yes, GNU is much larger than Linux.

The kernel has a GPLv2-only license

This is a misinterpretation of Linus' comment at the beginning of the COPYING file in the kernel. Linus can't change the GPL, since the GPL itself is copyrighted by the FSF. He also can't change the license conditions of the original authors of all the files that went into Linux until he made that commend (in the 2.4.0-test series). So what he can do is to clarify under which conditions he's redistributing the assembly of all those GPL and BSD licensed files. And that's "under GPLv2".

As he did single-handedly choose this condition, he can also single-handedly choose to rethink. It's his right to choose the conditions unter which he's distributing the kernel, because all GPLv2 code is GPLv2 or later unless explicitely stated otherwise - by the author (as recipient, you get a direct license from the author, so the author has to say something)! So far, no author of Linux other than Linus himself has made such a statement - if you grep the kernel for copyright statements in file headers, you find that roughly one fourth of the kernel is explicitely GPLv2 or later, and the remaining three fourth don't say anything.

BTW: The FSF requires each file of a FSF project to contain an explicit copyright notice, and this also solves the hijacking problem: You simply take the e.g. ssl.c out of libssl, and state that it is under the Apache license, and that's it. It's compatible to GPLv3, so can be included in a GPLv3 distribution, and it by itself also stays under Apache license. The hijacking is only a problem when you do it like Linus and the other kernel hacker do it: Change the copyright notice on the file. From a tit-for-tat point of view, this is not fair.

This module is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version
2 of the License, or (at your option) any later version.

> So whether the GPLv3 fulfills what the GPLv2 was "supposed to do in the
> FSF's view" is pretty much besides the point - the FSF only wrote the ~600
> lines GPL license, the kernel developers wrote the other 7 million lines of
> code. Thus what matters mostly is the Linux kernel contributors' view on
> this issue.

You are of course right about this - if the kernel developers want to use the GPLv2 license, it is of course their right to do so, and they do not have to justify themselves for that choice.

I also realise that some of the kernel developers seem to have strong feelings against the FSF, and might be disinclined to consider the GPLv3 for that reason. Which is fine too - you can just say you dont want to look at any new FSF license if that's how you feel.

However, the kernel developers *have* been making arguments against GPLv3 and I find these to be quite confusing. So since there has been some arguing, it's only natural that people will try to understand what you mean. In the message you were replying to, khim was arguing (quite convincingly, I think) that the GPLv3 does not do anything beyond what GPLv2 was supposed to do. You say the FSF's opinion of what GPLv2 is supposed to do is irrelevant - fine, but then, what is YOUR opinion of what GPLv2 is supposed to do ?

So far I've heard of the "GPLv3 would prevent tivo from using linux" issue. Linus has been arguing (actually, you did too: http://lwn.net/Articles/200628/) that it's a USE restriction and that GPLv2 did not have those - but that's incorrect, the restriction is only about DISTRIBUTION and GPLv2 already had some of this, i.e. microsoft can not "use" the linux TCP stack in their windows OS either. Or technically they'd be free to USE it internally (GPLv2 does not have any usage restrictions), but they could not DISTRIBUTE the result. If you really wanted a license without distribution restrictions, I think you'd just use BSD. So, in YOUR opinion, why are you happy with the distribution restrictions in GPLv2 but not those in GPLv3 ?

You've also been arguing that tivo should be free to lock down their software, so that you get the source but you can't recompile it and run the result on your tivo. You seem to think it's OK since if one wants to be able to run their modified code, they can just buy a general-purpose computer. Would you still feel that way if it somehow became impossible to buy a general-purpose computer with TV inputs and outputs ?

I'm just trying to understand your opinion...

Thanks,

the big difference is: if you are unhappy with Linus you can fork the Linux codebase and try your luck. That keeps Linus honest and sharp every day. If he messes up, people will leave his project.

On the other hand the FSF licensing process cannot be "forked". What RMS decides goes into the GPLv3. No matter what the "Committees" do, or what the "comment submitters" do. There's no competitive pressure whatsoever to keep him honest or sharp.

Look at the RMS quotes i linked to: "Most of our community does not believe in free software". And the GPLv3 process very much reflects this belief of him: RMS does not trust the community. So he (and some of his more vocal supporters) should really not be surprised if the community does not trust him.

For what it's worth, I personally think signed software installable only
by a service operator makes sense if a machine is provided as part of a
service, eg. cable TV or a telephone, just like you used to have to get
permission from the BOFH. Clearly you don't have the four freedoms here;
maybe that's okay but maybe *not* if the software is copylefted. On the
other hand if you've bought the box, you own it and your right to tinker
with it should extend to the ability to reflash it if that is technically
possible. And if it's running free software, that means the ability to
modify the source, recompile and reflash that. But that's just *my*
opinion, and has nothing whatever to do with copyright law and licences,
nor with the FSF position on the Four Freedoms.

The "symmetry" that manufacturers should provide the ability to change
the software if they have it themselves is not "essential" to the four
freedoms, but Tivoisation and ROM *both* violate freedoms 1 and 3 in
exactly the same way as not having the source does. You can't change and
improve the software you have.

The difference is that in one case it might as well be a law of nature,
and in the other it's something deliberately imposed by someone who is
supposed to be granting their users the four freedoms they had from
upstream.

I think this point and "symmetry" are logically equivalent, or nearly so.
Tivoisation unquestionably violates the *spirit* of the GPL, if not the
letter of version 2.

I guess that the FSF have considered this similarity between keyed but
reinstallable software and ROM, and decided that a hardware limitation
like ROM is an acceptable exception to the four freedoms (in the same way
as, for instance, distributing free software binaries for computers which
aren't capable of running the compiler is acceptable), but actively using
a technical measure like TPM to restrict the ability of the user to do
what would otherwise be possible is not acceptable.

I see what you're saying, and it would be justifiable on utterly pedantic
grounds to add equivalent restrictions forbidding free software on ROM or
on inadequate computers. But the FSF's goal, in the end, is not to be
utterly pedantic but to expand freedoms. The Tivo and similar devices do
the exact opposite; they actively work against the goal, and so a new
licence which stops them from doing that is called for.

I will be very, very surprised if any of the Tivoisers switch from GPLv2
software to GPLv3 software in ROM; they are vastly more likely simply to
stick with the GPLv2 version. Remember that anything ever released under
GPLv2 will always be usable under those terms. Only when upstream
increments to "version 3, or at your option, any later version" will the
GPLv3 "innovation stream" be closed to the poor deprived Tivoisers.

It is only in the long term that this might impose *any* cost on
Tivoisers, and it's only an opportunity cost. The real pressure is peer
pressure; far more effective. So why you even mention ROM as an
alternative except to question the consistency of Eben Moglen (ha ha!) is
beyond me. You're sooooo pedantic ;-)

I can't see the logic behind there argument, which is strange because they are usually quite clear and direct. That makes me think it's more than just the license, and possibly a clash of personality/goals/values. It would clear things up quite a bit if they would make known how they would like the GPL changed instead of how they don't want it changed.

I don't buy for a minute the argument that GPLv2 doesn't need changing. They accept/tolerate binary only modules which the license doesn't.