Weekly Edition
Return to the Announcements page
| |||||||||||||
WebM: Missing The Assurances Open Source Needs? (ComputerWorld)
Simon Phipps expresses
some concerns about WebM in ComputerWorld UK. "Firstly, the new
license Google is using for the project is one that's not been submitted to
the Open Source Initiative for approval. As it stands it possibly can't be
approved due to Google's ironic inclusion of a 'field of use' restriction
in the patent grant (which is restricted to 'this implementation of VP8'
rather than the more general grant in the Apache license from which the
text started). That means WebM is not currently open source, despite using
a license based on the BSD and Apache licenses."
(Log in to post comments)
Patent grant is additional Posted May 24, 2010 14:17 UTC (Mon) by epa (subscriber, #39769) [Link]
The patent licence grant in the text Google uses gives you extra rights above the BSD or X11 licences (which do not grant any patent licence at all). It's hard to see how it could be considered non-free because of that.
However, I would prefer to see separate licences for copyright and patents, rather than mixing them in the same text. Then if you wanted you could disregard the patent side entirely and pretend the software was simply BSD licensed, and no different from a BSD licensed H.264 encoder or MP3 player (which are considered open source by the OSI's definition).
Patent grant is additional Posted May 24, 2010 14:59 UTC (Mon) by njs (guest, #40338) [Link]
There is also a separate license for the patents, to cover independent reimplementations. But there's an important reason why the copyright license includes patent language: if you make a patent claim against VP8, then your BSD-ish rights to the reference implementation are revoked.
Patent grant is additional Posted May 24, 2010 15:06 UTC (Mon) by mjg59 (subscriber, #23239) [Link]
Not quite. The text of the patent grant indicates that your *copyright* license terminates if you engage in litigation in which you claim that WebM infringes upon your patents. So there's definitely an opportunity to argue that this makes it non-free, but if we look at the Apple Public Source License (which is on OSI's list of approved licenses) we see:
12.1 Termination. This License and the rights granted hereunder will
(...)
(c) automatically without notice from Apple if You, at any time during
That is, the APSL terminates your copyright license even if you engage in entirely unrelated patent suits against Apple. I lean towards thinking that the Apple license is overreaching, especially from a hardware company - I can't sue Apple for patent infringement on some novel aspect of platic injection moulding without also losing my right to use any APSL licensed software. Google's is more like the MPL, which terminates the copyright license and the patent license if you allege that the MPLed work infringes your patents but only terminates the patent license if you allege that some unrelated work infringes your patents.
So while this license is in some ways more restrictive than BSD, it's less restrictive than APSL and arguably less restrictive than this aspect ot MPL.
Patent grant is additional Posted May 24, 2010 17:22 UTC (Mon) by webmink (guest, #47180) [Link] So while this license is in some ways more restrictive than BSD, it's less restrictive than APSL and arguably less restrictive than this aspect ot MPL. Except the APSL (which would not be likely to be easily approved by today's OSI by the way) at least allows you to defend yourself; Google's license removes both copyright and patent licenses in the event of a counterclaim. That strengthens Google's market power over their supporters. It also conditions patent grants more narrowly than the Apache license it is derived from, so that there would be no certainty you had patent cover in the event you had to remove some function if you yourself were attacked by an external patent holder and had to settle. And there are other questions too. In other words, there is plenty of new work that needs discussion, even if I understand what they are doing, why they are doing it and agree with them it's needed. Google's new license or may not be a free and open source license, and my view is worth roughly as much as everyone else's. To reach a consensus I would rather have a public discussion about it in the usual way on OSI's license-review list than simply have it declared Free by fiat (be that Google, FSF or whoever) or in blog comments by partisans (like me). Folk here may not hold OSI in much regard (actually neither do I, which is why I joined the Board to fix it instead of just complaining), but it's all we have for now in terms of objective public review if companies like Google choose to invent their own licenses instead of using existing free software licenses. If we decide OSI's public review and approval process is optional, however flawed it may be, we are back to the old days of vendor claims being king because anything else is just someone's dismissable view.
Patent grant is additional Posted May 24, 2010 17:45 UTC (Mon) by mjg59 (subscriber, #23239) [Link]
On the other hand, MPL has no such counterclaim language. Given that such provisions appears acceptable to the community at large, and given that licenses have traditionally not been required to contain any patent language at all to be considered free, I think that there's pretty much no plausible scenario in which Linux distributors will view the license as non-free.
The real question isn't its freeness. It's whether people feel sufficiently safe from patent suits that they're willing to distribute it. That's something that the license text may influence to a certain extent, but as you point out in your article an actual promise of indemnity (along with the infrastructure to back that up) would be more convincing. Is it necessary? I'd be surprised if it's an issue in the Linux world, although some commercial software vendors in other markets may see it as a blocker. Microsoft may well be willing to support WebM through codec plugins, but it wouldn't shock me to hear them suggest that they're unwilling to ship it with the OS due to patent concerns.
Patent grant is additional Posted May 27, 2010 1:10 UTC (Thu) by Arker (guest, #14205) [Link]
Keep in mind there is no patent indemnity offered by the competition either.
Patent grant is additional Posted May 24, 2010 17:51 UTC (Mon) by njs (guest, #40338) [Link]
> It also conditions patent grants more narrowly than the Apache license it is derived from, so that there would be no certainty you had patent cover in the event you had to remove some function if you yourself were attacked by an external patent holder and had to settle.
This is what I don't see in the text -- can you explain where you see this difference between the Apache and WebM arising?
(I wouldn't mind an explanation of this "field of use" restriction you claimed in the original post, while you're at it...)
> To reach a consensus I would rather have a public discussion about it in the usual way on OSI's license-review
Whatever OSI's ideal role would be, in practice the usual way we reach consensus on these things is a combination of public discussion (e.g., what we're doing now), debian-legal, and fedora-license-list, with the OSI mostly irrelevant.
Patent grant is additional Posted May 24, 2010 18:24 UTC (Mon) by webmink (guest, #47180) [Link] can you explain where you see this difference between the Apache and WebM arising? Apache talks about "the Work", which embraces community activity in the given Apache project and plausibly maintains rights when people need to reduce function for whatever reason. Google says "this implementation of VP8", strongly implying that the only acceptable derivatives will be compliant codec implementations with the same function as the original. That's what I mean by a "field of use" restriction; you may have no patent protection if you use WebM code in anything other than a VP8 codec. I admit to using the term somewhat rhetorically to echo another case some readers know about... in practice the usual way we reach consensus on these things If "we" means "LWN readers" you're probably right. The world is larger, though. There are many entities that rely on OSI approval in their selection mechanism, including especially governments. Love it or hate it, OSI still has a role that we in the software freedom community are foolish to neglect, which is why I for one felt compelled to try to stop its death-spiral. Not that anyone here would dare agree with me on that :-)
Patent grant is additional Posted May 24, 2010 18:44 UTC (Mon) by mjg59 (subscriber, #23239) [Link] I don't think there's anything in the "Field of use" argument, at least as far as whether or not the license can be considered free or not. If the patent language were stripped entirely then it would clearly be a free license. If the patent language purely consisted of the termination clause, I think it'd be consistent with the MPL and so still be a free license. Your argument appears to be that because the grant of patent rights provided may only apply to implementations of the spec rather than any other derivative works, it makes the license non-free. Given that it's granting you rights above and beyond those provided in what would otherwise be considered a free software license, this position seems... odd.Love it or hate it, OSI still has a role that we in the software freedom community are foolish to neglect Well, we pretty much agree that OSI's positions don't have much influence on the Linux world. It's in OSI's interests to avoid there being major differences in position - if OSI doesn't view the WebM license as acceptable while Mozilla and every significant Linux distribution do, that reduces your credibility. So I think you're better served making a more active attempt to determine what the feeling of the rest of the community is before making pronouncements like "That means WebM is not currently open source" - OSI has no ownership of the term, and pretending otherwise risks you becoming even more marginalised among the groups that produce the works you seek to advocate.
Patent grant is additional Posted May 24, 2010 18:55 UTC (Mon) by webmink (guest, #47180) [Link] Your argument appears to be...non-free. No, my argument is that it merits review rather than the free pass it's uncritically getting, since reasonable people have scope to differ over its interpretation and there has been no conclusive independent review of the kind many of us prefer for things self-describing as "open source". Your view of OSI is depressingly unsurprising but I suspect it's a waste of time trying to discuss it.
Patent grant is additional Posted May 24, 2010 19:04 UTC (Mon) by mjg59 (subscriber, #23239) [Link]
It's a license that grants you permissions that are broader than the permissions you gain from a 3-clause BSD license. It's clearly an open source license as defined by the OSD, a free software license as defined by the FSF's four freedoms and almost certainly something that would be considered DFSG-compliant. Saying "It's not currently open source" isn't giving it review - it's denying reality. I think there's a good argument for requiring stronger patent language in modern license, but if OSI want to use that to decide whether something's open source or not then that's a conversation OSI have to have with the community rather than imposing unilaterally.
Patent grant is additional Posted May 25, 2010 6:49 UTC (Tue) by BrucePerens (guest, #2510) [Link]
How can you say that? The "this implementation of VP8" text prohibits modification.
Patent grant is additional Posted May 25, 2010 9:39 UTC (Tue) by nye (guest, #51576) [Link]
>How can you say that? The "this implementation of VP8" text prohibits modification.
I can't see how any reasonable interpretation could lead to that conclusion based on the phrase "this implementation of VP8", while not coming to that conclusion based on the phrase "the work", without also adding "and any derivative works". (Obviously one might reasonably conclude that in *both* cases.)
That said, I'll gladly accept that legal documents don't generally permit reasonable interpretation, however even a patent grant that narrow in scope can't be less free than *no patent grant at all*, which everyone seems to agree would make this a free licence.
Patent grant is additional Posted May 25, 2010 13:24 UTC (Tue) by BrucePerens (guest, #2510) [Link]
The reason this is less free than a license without an explicit grant is that those licenses have implicit grants.
Patent grant is additional Posted May 25, 2010 15:06 UTC (Tue) by pdundas (subscriber, #15203) [Link]
It seems to depend on whether you interpret "this implementation of VP8" as meaning on one hand the exact sequence of ones and zeros in the version of the software to which it refers, or on the other hand, that version and any derived work. Or some further refinement, or restricted set of derived works.
Google are competent to say which they mean.
If the text does not reflect what they intended, and a change is required, it would not be the first time they have made the necessary corrections to T&Cs or licences.
It may not yet be time to leap to firm conclusions.... though these are fair questions to ask.
Patent grant is additional Posted May 25, 2010 10:58 UTC (Tue) by epa (subscriber, #39769) [Link]
Not at all. You may modify the code all you want. However Google's patent licence to you only covers the implementation they shipped.
If you consider this non-free, then you would also have to consider a plain BSD licence (whereby Google does not give you any patent rights or promises at all) to be non-free.
However, it could certainly be better; they might grant a patent licence for use in all free software (which is equivalent to 'modified versions of this implementation', since any program can be modified into any other), subject to a retaliation clause.
Patent grant is additional Posted May 25, 2010 13:29 UTC (Tue) by BrucePerens (guest, #2510) [Link] You can modify it all you want, but you'll be a patent infringer according to their terms. That's equivalent to "you can't modify at all". I think a lot of people are not getting the fact that sometimes not stating a rule leads to a more permissive legal interpretation than otherwise. This is the case with patents and the BSD license. All lawyers (and I've discussed this with lawyers a lot - I just taught a continuing legal education course for a roomful of them last week) seem to be agreed that there are implicit patent grants in all Open Source licenses that don't make explicit ones.
Patent grant is additional Posted May 25, 2010 14:21 UTC (Tue) by mjg59 (subscriber, #23239) [Link]
<i>You can modify it all you want, but you'll be a patent infringer according to their terms.</i><p>
The specification license says otherwise. When analysing a piece of software (which is, after all, the relevant thing to analyse), the relevant factors are the ability to use, modify and distribute without fear of legal attack. x264 fails this test in much of the world because, while we have full rights to do so under copyright law, it infringes upon actively enforced patents. The WebM software license gives us the full set of copyright permissions we want, so patents are the only relevant issue. The specification license gives us a patent grant for *any* implementation, not just Google's, so the patent grant in the WebM software license isn't something we have to care about at all.
Would a broader patent grant be nice? Certainly. But I think it's a bit extreme to suggest that an implicit patent grant will generally imply that all derivative works will be covered, no matter how much they differ from the original work. In the absence of that, I don't think there's any reason to believe that the Google situation is worse than the status quo.
Patent grant is additional Posted May 25, 2010 14:49 UTC (Tue) by nye (guest, #51576) [Link]
>I think a lot of people are not getting the fact that sometimes not stating a rule leads to a more permissive legal interpretation than otherwise. This is the case with patents and the BSD license. All lawyers (and I've discussed this with lawyers a lot - I just taught a continuing legal education course for a roomful of them last week) seem to be agreed that there are implicit patent grants in all Open Source licenses that don't make explicit ones.
Do you have any references to any public comment to that effect by somebody qualified in this field, especially with some explanation rather than simply a blanket statement? Has this ever been argued in court? I would love to hear more about the legal reasoning behind this, because on the face of it it is absurd.
Elsewhere in this thread you stated that 'The issue is that if you try to sue someone for using software you've distributed with a license granting the right to use, the judge will correctly assume that you made an implicit grant to use the applicable patent as well'. I see no way to come to this conclusion - for one thing because you haven't granted 'the right to use'; you've granted a copyright license. But assuming that the judge does decide that there is an implicit license, why would it be interpreted as being any broader than the minimum license required to use the provided software? The extension of a patent licence to any derivative works seems like an even further stretch.
This sounds like tacking on some unrelated right that was never given, on the grounds that 'it's useless without it'. The same argument could be made about software which includes some open parts relying on some closed parts, or about Tivo-isation. If (in the US) you break some DRM because the software provided is otherwise useless, you are still liable to prosecution under the DMCA. Surely the same line of reasoning applied to the patent case should say that the vendor gave you implicit permission to break their DRM by providing software with an open source copyright licence.
(BTW I'm not just arguing for argument's sake; I really would like to know how that conclusion is reached.)
Patent grant is additional Posted May 25, 2010 15:17 UTC (Tue) by epa (subscriber, #39769) [Link] You can modify it all you want, but you'll be a patent infringer according to their terms.No; the licence does not state particular patents held by Google which the software implements, nor does it require you to accept any such claim. Whether or not you are infringing on a patent is not something which a licence document can change. I think a lot of people are not getting the fact that sometimes not stating a rule leads to a more permissive legal interpretation than otherwise. This is the case with patents and the BSD license.That's a good point. 'Redistribution and use are hereby permitted...' might be taken to include an implicit patent licence too.
Patent grant is additional Posted May 26, 2010 1:28 UTC (Wed) by BrucePerens (guest, #2510) [Link] I am an expert witness who testifies about Open Source licenses (here is an example) and I would have a hard time making your argument and winning the case in court. It's pretty credible that the meaning of the text is that any patent grant you might otherwise have had terminates upon modification, and thus the grant is a lot more restricted than the usual implicit one and not really "additional". It's more "subtractional" :-) I advise my clients (who are generally either lawyers or customers of a lawyer I am working with) not to accept anything this ambiguous.Larry Rosen pointed out that AFL does what they want and is OSI approved. I think the real solution is to get them to use an OSI-approved license and dump this new creation.
Patent grant is additional Posted May 26, 2010 2:28 UTC (Wed) by mjg59 (subscriber, #23239) [Link]
I don't think it's obvious at all that the intent is to provide a patent grant to all derivative works. Whether that's something that matters to the open source community is also non-obvious.
Patent grant is additional Posted May 26, 2010 9:32 UTC (Wed) by BrucePerens (guest, #2510) [Link] There is no reason to believe that the intent is to provide a patent grant to derivative works from the existing text, however this is necessary. If you really claim not to understand how much this matters to the Open Source community, at this late date, I am forced to believe that you're trolling the discussion.
Patent grant is additional Posted May 26, 2010 11:53 UTC (Wed) by mjg59 (subscriber, #23239) [Link]
If there's no reason to believe that, then there's no reason to believe that AFL does what they want. But let's play a little thought game. VP8 and h264 have various fundamental similarities. If I modify the WebM implementation into one that's able to play h264, should I still be granted a right to use Google's patents? What if I'm an MPEG-LA member company who then licenses my Google-approved h264 implementation to all the other members? What if, some years down the line, an MPEG-LA dummy company sues Google for patent infringement?
It's absolutely unclear that broad patent grants are what we want here. It's certainly unclear that they're the status quo in the open source world - as you said, there's been no legal testing of whether any implicit patent rights apply to derivative works as well. Insisting on that requirement carries a real risk of finding that traditional licenses are non-free.
The real question in terms of determining whether a piece of software is free from a patent point of view is (as it has always been): can I ship this work without significant fear of being sued for patent infringement. And the answer would certainly appear to be "yes".
Patent grant is additional Posted May 26, 2010 13:50 UTC (Wed) by foom (subscriber, #14868) [Link]
I don't really see the problem. Google has given two separate patent licenses, each covering different situations. This is two more than most software come with! I really don't see the issue here. I'd say #2 is pretty much superfluous, given #1, but could give some extra comfort, just in case their implementation of VP8 doesn't actually implement the spec, or something...
1) A license that applies to any implementation of VP8
Google hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer implementations of this specification where such license applies only to those patent claims, both currently owned by Google and acquired in the future, licensable by Google that are necessarily infringed by implementation of this specification. <<retaliation clause>>
2) Separately, a license that applies to the code google distributed, regardless of whether or not it implements the VP8 specification.
Subject to the terms and conditions of the above License, Google hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer this implementation of VP8, where such license applies only to those patent claims, both currently owned by Google and acquired in the future, licensable by Google that are necessarily infringed by this implementation of VP8. <<retaliation clause>>
Patent grant is additional Posted May 26, 2010 2:32 UTC (Wed) by mjg59 (subscriber, #23239) [Link]
Further, OSI's license proliferation committee suggests that the AFL is redundant with Apache 2.0, with the implication being that new works shouldn't use it. The messaging here isn't consistent.
Patent grant is additional Posted May 26, 2010 9:23 UTC (Wed) by epa (subscriber, #39769) [Link]
What you say makes sense. However, consider a hypothetical VP8 encoder written by a single hobbyist developer and distributed under the BSD licence. Such a program comes with no patent grant at all, since the author holds no patents. Would you consider that to be a non-free licence?
Patent grant is additional Posted May 26, 2010 9:57 UTC (Wed) by BrucePerens (guest, #2510) [Link] Even if the author doesn't presently hold patents, there is an implicit patent grant in the license. Now, change this from an individual to a company. It is actually quite common for a company to not understand what patents they hold and that they apply to a particular piece of software - this was a concern during the development of the W3C patent policy, because the participating companies didn't want to be forced to search their portfolio every time they worked on a standard. Now, say that company runs into trouble and starts aggressively asserting all of its patents in any way it can to generate cash or at least litigation that promotes a perception that the company still holds something of value and is worthy of investment. The Open Source developers can be in pretty deep trouble. This is not fiction. Consider the recent plight of the JMRI project and Bob Jacobsen. He had 5 years of hardship, came close to losing his nuclear physics job due to the action of the patent holder (and the other nuclear physics labs would not have hired him if LBL fired him), and dropped a year of income in defending himself and his project. He was not fully compensated by the settlement. We should certainly expect from Google the protections we'd get automatically from an individual that wasn't capable of asserting patent rights against us, or that we would demand from a smaller company that might be capable.
Patent grant is additional Posted May 25, 2010 23:44 UTC (Tue) by vonbrand (subscriber, #4458) [Link] That is rather suspect... the BSD license gives you absolutely no explict right over any patents the licensor might hold, and I can't see any way it gives an implicit license. "Use as you wish, at your own risk, don't expect anything else from us" is what it quite explicitly says.
Patent grant is additional Posted May 25, 2010 8:35 UTC (Tue) by jpnp (subscriber, #63341) [Link]
BSD grants you a copyright license without threat of revocation; WebM allows for the copyright license to be terminated under certain circumstances. I don't think you can say WebM is broader than BSD3.
Patent grant is additional Posted May 24, 2010 21:19 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
In practice, it is declared to be open source when major free software organizations and distributions like Fedora and Debian decide to include it. OSI's view of the matter is marginal at best because it has approved several questionable licenses that major distributions avoid. If OSI wants more credibility, it needs to revoke licenses and start considering the input of organizations mentioned above. We need OSI to be more representative of the community and play a active role.
Patent grant is additional Posted May 24, 2010 22:10 UTC (Mon) by webmink (guest, #47180) [Link]
I agree with your last sentence and am actively working on it. As for the rest; that's the view in the radical Free software community, but elsewhere the view is different.
Patent grant is additional Posted May 24, 2010 23:49 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
Which parts of my statement do you really think represents the "radical" community? I consider my position eminently reasonable of course :-)
Patent grant is additional Posted May 25, 2010 7:03 UTC (Tue) by BrucePerens (guest, #2510) [Link] Unfortunately, a pretty good case can be made that being "more representative of the community" would be to admit more folks from companies that would prefer to see OSI adopt an anything-goes policy. OSI's role is really to be the Lorax (a Dr. Seuss character who spoke for the trees in an attempt to remedy unacceptable damage by foresters). They have a line to hold and being indiscriminately representative might not enable them to hold that line.
Patent grant is additional Posted May 25, 2010 12:12 UTC (Tue) by mjg59 (subscriber, #23239) [Link]
Comparing the intersection between https://fedoraproject.org/wiki/Licensing#Bad_Licenses and http://opensource.org/licenses/alphabetical we find that the list of OSI approved licenses includes things like the Frameworx license (that prohibits value-added services being sold at anything other than the market rate). OSI may not have adopted an anything goes policy, but the policies it *has* adopted position it some distance away from the people it originally sought to represent.
Patent grant is additional Posted May 24, 2010 19:19 UTC (Mon) by njs (guest, #40338) [Link]
> Apache talks about "the Work", which embraces community activity in the given Apache project and plausibly maintains rights when people need to reduce function for whatever reason. Google says "this implementation of VP8", strongly implying that the only acceptable derivatives will be compliant codec implementations with the same function as the original.
So, to check I'm understanding right, your claim is: both licenses offer patent rights for the work being licensed, and neither makes any explicit mention of those rights extending to derivative works, but, because Apache is written in a generic style and uses a vaguer term like "Work", that vagueness implicitly extends those patent rights to a broader range of derivative works?
I'm not a lawyer either, but that doesn't sound right to me. "Work" is not at all vague -- the document says: "'Work' shall mean the work of authorship, whether in Source or Object form, made available under the License, as indicated by a copyright notice that is included in or attached to the work". I.e., the stuff that the rights holder attached the license to, not any derivative work of it.
So in both cases, you get exactly those patent rights which are needed to "make, have made, use, offer to sell, sell, import, and otherwise transfer" the code that was placed under this license -- no more, no less. It doesn't matter which words the license uses to refer to that code, because it refers to the same thing either way.
> I admit to using the term somewhat rhetorically to echo another case some readers know about...
FYI, I'm not sure this helps OSI's credibility. You're speaking as a Director of the OSI, an organization who's entire purpose is to judge whether licenses meet the Open Source Definition, but you're using terms that have a technical meaning within that definition (specifically, "field of use" is understood to refer to point 6) in a vague and contradictory way. The car version is my mechanic going "oh, did I say 'sparkplug' in that quote? I meant 'sparkplug' rhetorically, as in, your drive-train is easily angered".
Patent grant is additional Posted May 24, 2010 19:53 UTC (Mon) by webmink (guest, #47180) [Link]
Yes, that vagueness defining the scope of the grant is important, as is the extension of the scope of what triggers the retraction. As for OSI; I'm beginning to understand why no-one wants the job, given the irrational hatred that spouts from the ground in connection with it.
Patent grant is additional Posted May 24, 2010 22:12 UTC (Mon) by njs (guest, #40338) [Link]
As I said, to me the license text seems very careful to rule out the vagueness you see. But perhaps this is one of those issues where neither of us can say for sure without more legal training (and/or a judge arbitrarily making something up...).
By "extension of the scope of what triggers the retraction", you mean the additional language to include lawsuits that are not directly filed by the party to the license, but including lawsuits filed by agents, lawsuits that are agreed to by that party, etc.? I agree, I'd like to see more discussion of this, though I can't currently see how it's doing more than closing some potential loopholes.
Re: the OSI: I'm sorry you're feeling attacked, and I hope I haven't been one of the people giving an impression of hatred, irrational or otherwise -- certainly I feel no hatred for OSI or you. Indeed, I totally agree that careful examination of this license is necessary (obviously, I guess), and I didn't realize that Google *had* made up a new license until this article came along. It's just a shame that what sparked it off was ComputerWorld quoting you making some, at the least, poorly supported (and potentially harmful) claims.
No one wants the job? Posted May 26, 2010 6:55 UTC (Wed) by man_ls (subscriber, #15091) [Link] It is not irrational hatred; it's just that you are held to a higher standard in your communications. Nobody hates you, in fact I'm beginning to like you because you are not afraid to engage in discussion on LWN (even if you are not a subscriber). But you have to be more careful I guess.The title "Directory of the OSI" reminds me vaguely of the "Prime Minister of Cyberspace" in Doctorow's When Sysadmins Ruled the Earth: it has no advantages and it carries a great deal of responsibility (in your case resurrecting a zombie organization). Why should anyone want the job?
Patent grant is additional Posted May 24, 2010 20:59 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
"Except the APSL (which would not be likely to be easily approved by today's OSI by the way) at least allows you to defend yourself"
Only tangentially related but, it has been clear to me that ASPL 1.x and Artistic License 1.0 are considered non-free by pretty much everyone except that they are OSI approved and diminishes the credibility of OSI IMO and should be revoked.
Patent grant is additional Posted May 24, 2010 23:24 UTC (Mon) by webmink (guest, #47180) [Link]
Sounds good. Care to suggest a process for this?
Patent grant is additional Posted May 24, 2010 23:46 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]
Talk to the vendors/communities involved with the older versions of the licenses and do a press release suggesting alternatives. Talk to the communities which already consider these licenses non-free and get quotes if necessary Drop the licenses from the OSI approved list and write in detail in a blog post about why.
Patent grant is additional Posted May 24, 2010 18:42 UTC (Mon) by jrn (subscriber, #64214) [Link]
Im not a lawyer, but I didnt read it that way.
> Subject to the terms and conditions of the above License, Google hereby grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free, irrevocable (except as stated in this section) patent license
This suggests to me that there are two separate licenses that just happen to both be described at <http://www.webmproject.org/license/software/>.
The patent license would be much clearer if it were put on a separate page, though.
Patent grant is additional Posted May 24, 2010 18:58 UTC (Mon) by webmink (guest, #47180) [Link] This suggests to me that there are two separate licenses That would be plausible if it were not for the fact that the copyright license as well as the patent license is conditioned on compliance with the patent license terms:
Patent grant is additional Posted May 24, 2010 19:25 UTC (Mon) by njs (guest, #40338) [Link] Also, to verify intent, see the FAQ:"The Apache license is somewhat similar in effect to this license. The main reason it was not used is that filing patent litigation against someone using the Apache 2 license only terminates patent rights granted under the license. Whoever filed the litigation would still be able to use the software they are suing over and still be in compliance with the license. This license, however, terminates all rights when patent litigation is filed."(Emphasis added.)
Patent grant is additional Posted May 24, 2010 20:47 UTC (Mon) by jrn (subscriber, #64214) [Link]
This could have been made a lot clearer by using a fourth clause in the BSD-style list of conditions.
Thanks for the pointer. It really does sound like the GPL-incompatible interpretation is intended. Ugh.
Patent grant is additional Posted May 24, 2010 21:00 UTC (Mon) by jrn (subscriber, #64214) [Link]
> GPL-incompatible
To clarify: I should have said GPLv2-incompatible. The GPLv3 includes a patent retaliation clause that iiuc should make it compatible.
Patent grant is additional Posted May 25, 2010 6:46 UTC (Tue) by BrucePerens (guest, #2510) [Link] Matt,The DFSG, which these days we also call the OSD, was designed to fit the Free Software licenses which we thought were "free" at the time. These included the GPL 2, which contains a defensive patent clause. The particular limit, unfortunately, prohibits modification, not just use for a field of endeavor. It's completely unacceptable in an Open Source license. All licenses without explicit patent grants contain implicit ones. The issue is that if you try to sue someone for using software you've distributed with a license granting the right to use, the judge will correctly assume that you made an implicit grant to use the applicable patent as well. How far do these grants go? Probably at least as far as the explicit grant in the VP8 license, possibly farther. You'd have to litigate to make sure. In general, an organization like OSI is going to be in favor of a defensive patent clause, because software patents have such a powerful capability to completely shut down broad swathes of Open Source. The fact that they have not already done so may only be a temporary thing.
Patent grant is additional Posted May 25, 2010 11:56 UTC (Tue) by mjg59 (subscriber, #23239) [Link]
I don't buy this argument at all, since it results in weird asymmetries. WebM's license is non-free because the patent grant is limited to implementations of WebM, but x264's license is free even though it provides no patent grant at all and there are actively enforced patents covering the work?
(Misc notes: The OSD and the DFSG are trivially not the same thing, and my name is "Matthew")
BSD has implicit patent grant Posted May 25, 2010 6:31 UTC (Tue) by BrucePerens (guest, #2510) [Link] BSD has an implicit patent grant, as do all Open Source licenses. If you publish something under a license granting the right to use and then sue the folks who use it, expect that the judge will not cooperate - you either gave those folks and implicit use grant for your patent or you entrapped them. But that grant probably does not go any farther than the patent grant in the VP8 license. It's never been litigated, so we don't know. Making it explicit makes the license non-DFSG-compliant. Which is just as well, because we really do need the right to modify.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 14:52 UTC (Mon) by j16sdiz (guest, #57302) [Link]
This is not DFSG free.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 14:54 UTC (Mon) by njs (guest, #40338) [Link]
I can't see any DFSG violation. Can you elaborate?
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 6:55 UTC (Tue) by BrucePerens (guest, #2510) [Link]
The patent grant language doesn't survive modification as presently written. It's just poorly written, possibly not deliberate at all, they need to fix it and reissue the license.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 18:00 UTC (Tue) by njs (guest, #40338) [Link]
It's identical to the Apache patent grant, which also doesn't survive modification. (See section 1, "'Work' shall mean the work of authorship...", "'Derivative Works' shall mean any work [...] that is based on (or derived from) the Work...'", and then note that section 3 grants "patent license to make, have made, use, offer to sell, sell, import, and otherwise transfer the Work". Just the Work, not any Derivative Works whatsoever, and keeping in mind that copyright law does, so far as I am aware, have any concept of a "project", as in an amorphous and mutable collection of related creative works that are treated as a single unit.)
So the only possibilities I can think of are, 1) I'm missing some difference between Apache 2.0 and WebM licenses, 2) Apache 2.0 is non-free, 3) WebM is free.
Which one are you arguing for?
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 14:53 UTC (Mon) by njs (guest, #40338) [Link] The only thing I can see in this comments that is at all accurate is his claim that the WebM license hasn't been submitted to OSI. (Or at least, I assume he would know if it had, since he's an OSI board member.) Is he just pissy about that, or... what?The WebM license is 3-clause BSD + a patent grant. The patent grant is taken from Apache 2.0, except that they've replaced "Contributor" with "Google" and "Work" with "this implementation of VP8". I do see a few interesting differences. The language has been strengthened in a few places. Specifically, the actions that can trigger the rights takeback clause in Apache are: "If You institute patent litigation..."; for WebM, they're "If You or your agent or exclusive licensee institute or order or agree to the institution of patent litigation...". And in Apache, a triggering lawsuit is one which alleges "direct or contributory patent infringement", while WebM adds "inducement of patent infringement" to the list. More importantly, in Apache, such a lawsuit causes you to lose the relevant patent rights; in WebM, both patent rights and the copyright license itself are revoked. Someone more up on licensing debates than I can probably say more intelligent things about the free-ness implications of this, but it seems pretty close to e.g. the GPLv3 rules. What I don't see is anything that resembles what Simon Phipps is saying. First of all, because a 'field of use' restriction is something else entirely (right?), and second of all, because the grant appears identical in this respect to the Apache license. The Apache license just grants you the necessary patent rights to use, make, etc. "this Work", not 'this Work and any other implementations of the same specification that you might have lying around'; when the WebM license says "this implementation of VP8", it's just saying the same thing in more specific terms. It doesn't give you any fewer rights. And then, if you check the specification license, it *does* grant you those same patent rights for 'any other implementations of the same specification that you might have lying around', so the patent grant is actually quite a bit stronger than Apache's. So he's just... wrong. Or am I missing something?
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 15:05 UTC (Mon) by spot (subscriber, #15640) [Link]
I think a notable point is that Simon is saying that the license isn't "Open Source" because the OSI hasn't approved that specific set of text.
Not that it matters if you submit a license for review to the OSI, they have a terrible track record of actually approving licenses. Technically, the Latex Project Public License isn't officially "open source" either, for the same reasons (although, to be fair, not for a lack of trying on the Latex Project's side).
It's a bit petty, and stupid. Which is one of the reasons we stopped caring about OSI's licensing decisions in Fedora.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 16:31 UTC (Mon) by mjw (subscriber, #16740) [Link]
Maybe OSI isn't a good forum to discuss new Free Software (or Open Source) licenses. But Simon does have a point that this is a new license, which would be good to discuss the details of. Is there a forum or organisation you would recommend instead of OSI?
As other people commented already there is a new interaction between the patent grant and the termination of the whole (copyright) license. One not previously seen in other Free Software licenses.
Wouldn't it be good if this was properly and openly discussed? Even the FSF these days holds public commenting periods when they introduce new (versions) of a license. Which in general seems to be regarded as a good thing. To prevent unintentional interpretations of the wording.
Personally it isn't immediately clear to me what the patent grant and revocation of copyright grant cover. It says "this implementation", does that mean only the exact implementation as distributed by Google, what about any derivatives? What about a theora encoder/decoder incorporating parts of the implementation? etc.
It would be good to have the discussion on what the actual intention was, and maybe the community can help strengthen the text to make it less ambiguous.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 16:52 UTC (Mon) by mjg59 (subscriber, #23239) [Link] Maybe OSI isn't a good forum to discuss new Free Software (or Open Source) licenses. But Simon does have a point that this is a new license, which would be good to discuss the details of. Is there a forum or organisation you would recommend instead of OSI?I don't think there is one. OSI definition and DFSG notwithstanding, the only relevant thing about a license is whether people you care about feel that they can ship it or not. There's a number of OSI approved licenses that Debian won't ship, for example. As other people commented already there is a new interaction between the patent grant and the termination of the whole (copyright) license. One not previously seen in other Free Software licenses. Not really. MPL has almost identical provisions and that didn't seem to prevent anyone shipping Mozilla before it was relicensed.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 17:36 UTC (Mon) by drag (subscriber, #31333) [Link]
Yeah...
Is there patent language stuff in GPLv3, Apache, CDDL and a few other licenses?
You might have a problem with compatibility with the GPLv2 -only, but for most other stuff it shouldn't be a big deal.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 19:08 UTC (Mon) by Trelane (subscriber, #56877) [Link]
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 6:58 UTC (Tue) by BrucePerens (guest, #2510) [Link] There was never anything about the MPL patent license that failed to survive modification, as the language in the VP8 grant fails.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 24, 2010 20:03 UTC (Mon) by pixelpapst (guest, #55301) [Link]
> Is there a forum or organisation you would recommend instead of OSI?
The debian-legal mailing list has a good track record of finding the possible non-obvious interpretations and the hidden interactions between clauses.
> As other people commented already there is a new interaction between the patent grant and the termination of the whole (copyright) license. One not previously seen in other Free Software licenses.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 6:56 UTC (Tue) by BrucePerens (guest, #2510) [Link] Approving licenses is a bad thing to do to the community. There are too many of them now, and each one doubles the combinatorial problem.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 17:51 UTC (Tue) by martinfick (subscriber, #4455) [Link]
There has been a lot of questioning in these threads as to why there is so much contempt for the OSI. One of the primary reasons suggested was: the OSI approved licenses that are obviously not viewed as free by others. I however think that this is merely a symptom of the real problem with the OSI. I believe the primary problem with the OSI is an unsolvable one. In other words, much of the contempt for the OSI is due to the nature of what the OSI is: primarily a political organisation.
Naturally, those who think that "using force to coerce others is OK" will have a hard time seeing the point of view of those who think that it is not OK. And, at heart, the OSI started as an organisation attempting to use force to dictate behaviour to free software communities. And so, it does not really matter what its decisions are, or how well they are made, if the outcome of those decisions has the potential to dictate rules on the communities that it endeavours to help. But, even without attempting to coerce us through language restrictions, why would free software developers enjoy being told which licenses to use any more than they would enjoy being told how they may write software?
There are other venues which have been mentioned here also which create policies about free software licenses, for example: the Fedora and Debian ones. But these organisations, no matter how they act, merely attempt to curb/control their own behaviour, not that of other organisations/communities. These organisations also DO something, they have a purpose outside of their policy making and therefore they are judged by the outcomes of their decisions as opposed to the rules they attempt to foist onto others.
...They attempt to lead instead of ruling...
They may often be ridiculed by their naysayers, but rarely will they be held with contempt. One might even say the same of the FSF. The FSF seeks to be a role model for FLOSS, they publish opinions and licenses. They even rule over the software they release, but they do not attempt to rule other FLOSS communities. So, once again, agree with them or not, applaud them or ridicule them, but rarely does one hate them.
Is it not ironic that the very person who wrote "The Cathedral and the Bazaar" was one of the original people founding an organisation attempting to be a Cathedral to rule the Bazaar? Perhaps this hypocrisy does not go subconsciously (if even consciously) unnoticed in the many communities? It should be no surprise then, that many in the communities will actively be angered by the attempts to rule them from this Cathedral. This statement is an example of an attempt to rule them if used in the context of the OSI:
"Approving licenses is a bad thing to do to the community. There are too many of them now, and each one doubles the combinatorial problem."
This does not even attempt to be an ethical guideline, it is merely an opinion about what is the best (most efficient) way to do something. Whether true or not, using it to dictate what sort of licenses are OK or not (even with good intentions) can only breed contempt.
Such a rule suggestion is unlikely to pass scrutiny of any of the distribution licensing lists, because they have to actually produce something! If someone submits a piece of software for inclusion in Debian which has a sub optimal (but libre) license, they will not likely deny the world its utility simply because it does not conform to some authoritarian view about the best way to advance FLOSS. They have their own ideas of what is the best way to advance FLOSS (and I suspect it has something to do with getting it into the hands of users). So, good intentions or not, I hope you can maybe understand why such a statement, foisted onto those unwilling to simply accept the rule of some self-appointed opinionated body (aimed primarily at influencing the big "force" organisations,) may breed contempt from a few.
And, to those who have asked here "where is a good place to discuss FLOSS licenses?", there have been many answers given: mostly licensing lists. These are good places, but, I have a better answer: in the bazaar! Yes, in many places, including here on LWN. That is the strength of the bazaar. If the only place that important concerns were discussed was in a single forum (whether it be an OSI one or not), much of the strength of the communities will have been lost.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 10:08 UTC (Tue) by sylware (guest, #35259) [Link]
It should be public domain.
If not, it should be GNU (A)(L)GPL protected.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 18:02 UTC (Tue) by njs (guest, #40338) [Link]
> If not, it should be GNU (A)(L)GPL protected.
RMS would probably disagree... http://lwn.net/2001/0301/a/rms-ov-license.php3
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 25, 2010 23:44 UTC (Tue) by sylware (guest, #35259) [Link]
What I meant is that BSD-like licences are useless, they protect near nothing and yet are not free as public domain.
Indeed, useless.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 26, 2010 2:47 UTC (Wed) by njs (guest, #40338) [Link] Not so -- they're effective in jurisdictions where one cannot place work in the public domain (the CC public domain dedication is longer and more complicated than the BSD license!), and they include (potentially very important) warranty disclaimers.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 28, 2010 9:38 UTC (Fri) by sylware (guest, #35259) [Link]
That licence pretends to be public domain.
I can write right now a domain public license:
"that code is public domain: you can do whatever you want with it at your full own responsability."
Easy. And as far as I'm concerned, till the "intellectual property" exists, I do believe that the best balance is the GNU (A)(L)GPL regarding open source interest. I barely can stand the fact that apple Oses are based on BSD code and they lock their code. BSD-like licenses are a failure to protect open source software and promote it in the current "IP" world. It's a fact.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted May 28, 2010 17:20 UTC (Fri) by njs (guest, #40338) [Link]
I don't have anything to say that I haven't already said, but please re-read what I said about public domain "licenses" above. The way law actually works (as opposed to the way it *should* work), there are real problems with the sort of public domain dedication you suggest, and using it is dangerous to you and those who listen to you. BSD-like licenses are one way to get the practical effect of a public domain dedication while avoiding those dangers.
WebM: Missing The Assurances Open Source Needs? (ComputerWorld) Posted Jun 3, 2010 10:57 UTC (Thu) by sylware (guest, #35259) [Link] BSD-Licenses are not public domain: A public domain license allows you to relicense the code as-is the way you (evily closing it) want without refering to it. BSD-Licences do not allow that. As a person pushing open source optimal software on computers, my choice in an "Intellectual Property" world is the set of GNU GPL licenses.
|
|||||||||||||