The GPLv3 violates the 4 GNU freedoms ...

Posted Oct 4, 2006 2:23 UTC (Wed) by sanjoy (subscriber, #5026) [Link]

The GPLv3 tries to control hardware that was not copied, modified or from any GPL codebase, and prevents the redistribution of that copy of GPL-ed code, and thus GPLv3 violates freedom 2.

and (from an earlier message)

trying to control what hardware does is immoral

It does not control the hardware, even if people claim many times that it does. It says only that if you distribute the code in such hardware, you need to provide another piece of information (a small piece of software, you could say): the authorization key. Tivo (if they used GPLv3 software) would be free to make whatever broken DRM hardware they like, and you would be free to fix it. How is that control over hardware?

A similar clause in the GPLv2 says that you need to include Makefiles or other compilation scripts, if you used them. Which makes sense. The Linux kernel would be a right pain to build without its Makefiles. This provision about compilation scripts is a condition on distribution: "You may distribute provided that you do the following..." Just as, under the GPLv3 draft, you may distribute Tivoized software provided that you give users the authorization key.

The GPLv3 [tries]...to control the handling of a piece of information that was created independently of any GNU software. That piece of information is Tivo's private key, which is only in the hardware.

It is not only in the hardware. It is also, like a compilation script, in Tivo's software somewhere so that they can install new kernels. The GPLv3 merely says that they must include that information, that software, as part of the source code, just as they must include the Makefiles.

GPLv3 proponents might argue that this violation of freedom 2 is done to strengthen another, more important freedom, but this brings in two new problems.

It brings in new problems only if it is a violation of freedom 2. But people are still free to redistribute copies. They need only to satisfy a few conditions and none of those conditions says anything about how Tivo manufactures hardware. With those conditions satisfied, other people can distribute improved, useful versions of Tivo software and help their neighbors, the purpose of freedom 2. So conditions on distribution are not offensive intrinsically; their moral weight depends what they are. As an example of a good restriction, the GPLv2 requires distributers to provide source code.

The GPLv3 is being consistent: It guarantees user freedom.

Suppose, under the GPLv2, you take the Linux kernel, make a version that will not boot without first getting an authorization key from Ovit Inc. You distribute that version in your hardware. Fine! Since recipients can get the source code, thanks to the GPLv2, they can undo the damage and install a less ridiculous kernel. Freedom prevails.

Now suppose Ovit Inc says: "By signing this contract, you agree not to install such a changed kernel. We will sell you an Ovit device once you sign on the dotted line." That trick too is foreclosed by the GPLv2, which says (in section 6): You may not impose any further restrictions on the recipients' exercise of the rights granted herein. Ovit could claim that it never agreed to the GPL and is therefore free to impose any contractual obligation on you (that is not against the law). However, that choice would make Ovit a copyright infringer, a ruinous proposition thanks to how draconian copyright law has become.

So Ovit Inc tries a third approach. It builds hardware with an embedded private key, and only if the keys check out with Central Command will the kernel boot. Here probably the GPLv2 has no solution (although the "compilation scripts" provision may help, but only a judge would know, or more precisely, could create an answer by making a ruling).

The GPLv3 says: "No. Just as you cannot lock the code down with boot keys or with legal provisions, you also cannot lock it down with hardware. Make whatever hardware you like, we have no opinion on that. But if you can modify the free software you are distributing, so can the people who receive it: You must give them the magic keys."

Too bad the kernel won't switch to GPLv3, due to logistical difficulties (pain in the neck to contact all copyright holders) and to ideological objections. But that doesn't mean the GPLv3 isn't a reasonable, consistent solution to a real problem.

Posted Oct 4, 2006 12:29 UTC (Wed) by stijn (subscriber, #570) [Link]

I actually have problems just understanding your Tivo example (also used elsethread) and I would like to grok it - this kind of scenario is what is needed most in this type of discussion. It is likely due to the fact that I do not have a full grasp of the licensese in GPL v2 and draft v3 and its implications in the scenario. More specifically, you seem to focus on userspace (significance?), binaries (source?), and redistribution (arrival in the first place?). This pertains to your point of freedom 2) being violated.

Freedom to tinker is an essential part of the famous printer driver story. I am not sure I follow.

About the fundamental moral question now. I wanted to establish that the FSF and Richard Stallman and Eben Moglen are acting in good faith and according to a consistent view of the world. The divide now is whether v3 is 'similar in spirit' or not. I tentatively think it is although I am not sure the DRM clause can be formulated in a sufficiently clear way (or, whether it can find an unambiguous border line). I am dead certain however that the FSF *intent* is to be 'similar in spirit'. I think there has been a lack of acknowledgement regarding this.

The moral question then pertains to all the people who licensed with the famous 'or (at your option) any later version', and it pertains to whether they judge the DRM clause to be similar in spirit. I am one of them, albeit with some very small impopular user-space projects. The kernel developers are not as far as the kernel is concerned. The busybox case is not so clear cut.

Linus Torvalds has done well to do without the 'later' clause, given his liking and interpretation of 'tit for tat'. The concerns raised by the kernel developers and the rationale are genuine and your many contributions in the various LWN threads are highly valued. The general stance has been rather unforgiving, emotionally charged, and one of abandonment. In this last sentence I was going to denounce that, but who knows, sometimes conflict is necessary.

Posted Oct 4, 2006 21:49 UTC (Wed) by roelofs (subscriber, #2599) [Link]

I believe your point is also missing a fundamental moral question: does RMS have the moral right to unilaterally and retrospectively change the contribution dynamics of a codebase that is roughly 1 billion lines of code written by hundreds of thousands of contributors, via a 340 lines license RMS wrote 15 years ago, without considering the motivations and moral background of the many contributors that wrote that codebase?

Insofar as those contributors granted him/FSF that right, whether explicitly or implicitly, why wouldn't he have it? How naive do you have to be to grant the "or any later version" power to a third party and not expect that it might result in license changes not entirely to your liking?

If you truly want to preclude such changes affecting your code, then you don't use the "or any later version" language--just as, in fact, Linus and you and the other kernel developers did. But for everyone else--everyone in that 1-billion-LoC, 100,000-plus-contributors category, the numbers of which I'll take your word for--they ceded both legal and moral authority to the FSF when they chose that wording on their license grants.

The only vaguely comparable examples I can think of--but where the "legal" and "moral" rights arguably did diverge significantly--are the Unisys/GIF/LZW promise of a patent exception for free (not Free) software, and the amazingly similar Mozilla/Firefox promise of a trademark exception for Debian. In neither case was the promise ever a legal agreement, at least under US law, and in both cases it was later revoked unilaterally. To put it another way, "moral" rights were (temporarily) granted, but legal rights never were. However, I see no similarity to the current GPL v2/v3 issue beyond the most basic level of all three examples having to do with legal rights and additionally having a "moral" or ethical dimension. The GPL case, unlike the others, seems legally, morally, and ethically self-consistent to me.

Greg