Allison Randal has posted reactions to the GPLv3 draft on O'Reilly Radar (Tim has a comment there too), and I have answered a point or two. Since I never post elsewhere without letting you know, so you always know if a post is really written by me and not an imposter, I'm letting you know, and I'll repost it here, to verify.
For context, here's the piece she wrote that I responded to:
Second, this addition to the GPLv3 is based on an assumption about the nature of the Microsoft/Novell deal. The key phrase is "conditioned on the non-exercise of any of the rights that are specifically granted to recipients of the covered work under this License." The assumption is that the grant of the patent license applies to users of the particular distribution of Linux (SLES) only if "the user does not make or distribute additional copies of SLES" (from the rationale document for the third draft, page 25), so it would be a restriction on the redistribution of GPLd software. But we don't actually know yet if that's part of the agreement. So they're adding language to the license in response to a deal, when the language may not even turn out to be relevant to the deal.
A third, and possibly most important, question is whether the terms here (however they're phrased) are even what free software needs. At the core it's an attempt to stop companies who distribute Linux, for example, from making patent deals with other companies. But these distributors already have hundreds of other deals with other companies. For most companies (software or hardware) the whole point of filing patents is to use them as the basis for collaborative deals with other companies. The real threat is not patent deals, they're harmless and only give the two companies involved a small measure of comfort that their partner won't stab them in the back for the duration and scope of the deal. It's patent suits that are dangerous and destructive, whether they're frivolous or not.
For full context, you likely will want to read everything she wrote, not just this segment, but this is just so you will understand what I wrote. She and I are on the same committee, by the way, and her input is very helpful on specific language. But this discussion has more to do with the nature of FOSS development and whether the community can or should just go along with the proprietary way of handling the software patent problem.
Tim posted this:
What bothers me here is the amount of attempted micromanagement of others that I see in this license draft.
For a license that's supposedly about software freedom, it's got an awful lot of detail on what people can and can't do with the software. And as you note, it's in the hands of a group that's willing to reshuffle the cards, making the license a moving target.
For true freedom, I'll take BSD any time
I responded like this:
I think you and Tim are both missing some things:
1. Eben Moglen *has* seen the agreement under NDA, so he does know what it says. I think it is therefore safe to say the language crafted is designed to match what he read.
2. Patent agreements may not be unduly harmful for proprietary companies, or so you posit. I think we all know that Bill Gates said if patents had been allowed on software when he started, the entire industry would be shut down by now. But leaving that aside, they are harmful for GPL'd code, in my view. Even GPLv2 has an implied patent license, the goal being to prevent patent infringement lawsuits over GPL'd code *if* you have taken the code and benefited from it by copying, modifying, redistributing, etc. You don't have to use GPL code. You can write your own code or buy some from someone. But the idea is to create a patent-free zone for GPL code, where innovation is not dependent on payments, further restrictions can't be placed on the code, and you can't benefit from the code and then turn around and sue the author.
A lot of GPL'd code is written by individuals, not companies, and they mostly can't afford to pay for patent licenses, even if they were willing, so if enough patent agreement arrangements get set up, it can actually shut down FOSS development as we know it. Even if it didn't have that radical effect, it certainly would change it, of necessity, since currently you can freely download such software for free.
I think the current Microsoft-Novell agreement, had it been in place in the early nineties, would have been sufficient to shut down Linux development before it could get off the ground. Most of the contributors then were individuals, not companies or employees of companies, who wrote code as volunteers. Under the agreement, they would not, as I understand it, have been able to write code ever used in a commercial system without threat of being sued for patent infringement. Who would wish to take such a risk?
So, shall we say we need no further Linux-like development projects? Unless we decide that volunteer coders are no longer needed and the Linux experiment is the last of its kind, then patent agreements are indeed harmful, since they would have that damaging effect on the open development method.
FOSS isn't developed the way proprietary software is, so what works in one environment doesn't work in the other, so extrapolating from one to the other does not work. You don't have to be anti-proprietary software to see that it's two different types of development.
3. BSD is a fine license for certain purposes, but not for ensuring freedom for the code. The whole idea of the GPL is to make sure the code does not get taken proprietary. The BSD license doesn't address that or block it. Companies can take BSD code proprietary, and then users can't view the code or fix it or modify it or do much of anything with it. Ironically, then it is possible for the company to make one small change and then charge the actual authors of the BSD code for the privilege of using the slightly modified version and to compete against them in the market. That's why JBOSS, prior to its acquisition, used predominantly LGPL. If they used the BSD, they did the development, then a company could take their code, make a small change, and then sell it in competition, without the development costs, and if they were a larger company, they likely could win the competition. Does that feel like freedom to you? Sensible even? Surely another word would be more appropriate.
The BSD license does nothing to ensure that code remains nonproprietary. If that is your definition of "freedom" -- maybe you are thinking from the point of view of the programmer, and maybe not about the JBOSS situation, and not about the end user, and certainly not about the code.
I have also added a comment to a new article she wrote on GPLv3 and 2-only compatibility, although she wrote GPLv2 instead, which makes no sense since v2 is compatible with v3. She wrote in a comment that GPLv3 would have no effect, because FSF couldn't sue on aggregated works:
This position means that there is no way for the FSF or anyone else to enforce any license on the Linux distribution as a whole, or on Linux as an operating system. Aggregates aren't subject to the GPL terms. Licenses can only be enforced on individual components separately, and lack of compliance on one component has no impact on other components.
The SCO litigation demonstrates how the GPL
license *does* have an effect, even though the
FSF isn't the one bringing the claim and
neither is Linus, the copyright holder on
the aggregate work. It's the author of the works infringed, the copyright holder on the pieces
inside the aggregated work, that is doing so, namely IBM.
GPLv3 won't be any different. And it will be
just as effective, hopefully more so.
By the way, if the Linux Foundation were to
follow the path you mention, not that it
plans to, it would be just as
difficult to make the switch as it will be
to switch to GPLv3, because all the authors
would have to agree, and some of them are
dead, disappeared, etc.
I wonder why so much energy goes into trying
to subvert GPLv3. There must be a piece I
don't see. The same thing happened with
GPLv2. It's a genuine puzzlement to me. It's
just a license, folks. And once it's out there,
it's not the FSF that controls how you use
it and in general it isn't the FSF that
enforces it. Enforcement is by means of
copyright law, which is powerful enough to
do the job. You don't have to be the
holder of the copyright on the aggregate
work to enforce your piece.