[License-review] For Approval: OpenMDW License Agreement, versions 1.1 (OpenMDW-1.1)

Fri Sep 11 13:41:54 UTC 2026

Joshua-san,

I had been sitting back for a few days because of an unrelated side issue,
but I think your argument remains the most important unresolved issue in
this discussion.

I agree with you that a "realistic threat" by itself cannot justify a
restriction in an Open Source license. I also agree that the "nexus and
proportionality" principle I suggested does not, by itself, fully answer
the threshold question you raised. I do not claim to have a neat or
complete threshold test either.

But I still wonder whether your account of patent termination ultimately
relies on another principle: reciprocity.

I understand your explanation that a patent grant closes a gap that
copyright permission alone cannot close, and that defensive termination can
protect the resulting patent peace. That makes sense to me.

But there still seems to be one additional normative step: why may a
license withdraw that patent grant when the licensee asserts its own patent
against the covered subject matter?

It seems to me that reciprocity is doing some of the work there. A licensee
receives patent peace with respect to the covered work, and if that
licensee then asserts its own patent rights against that same subject
matter, it may lose the benefit of that peace.

If reciprocity helps supply the justification for patent termination, I am
not yet persuaded why reciprocity must be categorically unavailable when
the asserted right is copyright and the assertion is directed at the same
licensed subject matter. Why can a comparable reciprocal relationship never
arise with respect to a copyright grant?

The current ModelGo discussion also seems relevant here. There, the trigger
is patent-only, but McCoy has raised the question whether extending
termination to patent assertions against Derivative Materials goes too far.
That suggests to me that "patent" does not automatically make a termination
provision acceptable, just as "copyright" should not automatically make one
unacceptable.

Ultimately, I suspect we have to look at several dimensions together: which
right is being asserted, what subject matter the assertion concerns, what
exactly triggers termination, what grants are terminated, and against whom
the litigation is brought. And perhaps that is unavoidable if
proportionality is part of the analysis. Proportionality is a matter of
degree rather than a binary distinction. We may have to approach the
boundary from both directions: cases that almost everyone would regard as
acceptable, and cases that almost everyone would regard as unacceptable,
and then ask where and why the line is crossed.

One unrelated note..., prompted by a recent side issue.
If anyone is using an AI Agent to participate in or analyze this
discussion, I would encourage them to read the Linux kernel's current
guidance on AI-assisted contributions and to consider what the Debian
Project's recent vote on generative AI says about human responsibility and
accountability. Whatever tools we use, someone should understand, defend,
and take responsibility for what is submitted under their name.

--
Shuji Sado
Chairman, Open Source Group Japan
https://opensource.jp/
English blog: https://shujisado.org/
Japanese blog: https://shujisado.com/

2026/9/4 12:29 Joshua Gay via License-review <
license-review at lists.opensource.org>:

> Respectfully, I have serious concerns with where this conversation is
> going.
>
> I think the "fruit of my labor" framing obscures why patent termination
> has traditionally been accepted in open-source licenses.
>
> If the principle were simply that a licensor may demand "you may use my
> work only if you do not sue me," well then, sure, Pamela is right that
> there would be no logical stopping point. The same rationale could justify
> termination for copyright, trademark, contract, privacy, employment, or
> almost any other claim the licensor considers threatening.
>
> But that cannot itself be the open-source rationale, because it treats the
> licensor's reason for imposing a condition as sufficient reason for OSI to
> accept it. Those are separate inquiries:
>
>    1.
>
>    Every restrictive condition can be characterized as the price a
>    licensor demands in exchange for use of its work.
>    2.
>
>    Open-source review must still determine whether imposing that price is
>    consistent with the freedoms the license is supposed to carry to users.
>
> The historical justification for patent termination is much narrower. A
> copyright license alone may not provide practical freedom to make, use, or
> distribute software because patents can independently prohibit those same
> activities. . . . The patent grant closes that gap.
>
> A termination provision tied to patent litigation can protect the
> resulting patent peace by withdrawing the patent grant from a recipient
> that enjoys it while asserting patents against the covered work. Apache-2.0
> illustrates that relationship directly: the patent grant and termination
> provision appear in the same section, and covered patent litigation
> terminates the patent licenses granted for that Work. Some approved
> licenses impose broader consequences, but that shows that OSI has sometimes
> implemented the patent exception broadly. It does not establish a general
> principle that any claim concerning licensed material may terminate the
> license.
>
> Mike's clarification narrows the copyright claims covered by OpenMDW. A
> claim based only on the use of copyrighted material during training would
> not trigger termination unless the claimant also alleges that the resulting
> Model Materials directly or indirectly infringe. That is an important
> limitation, but it does not answer the underlying question. Even on that
> narrower reading, a copyright holder may be asking a court to decide
> whether the Model Materials incorporate or reproduce its protected
> expression without authorization and whether the distributor had the right
> to publish what it purported to license.
>
> How do OpenMDW responds to that claim? By terminating every grant the
> claimant received under the license! Termination occurs upon filing or
> maintaining the claim, before its merits have been determined, *and the
> claim may be brought against any person or entity rather than only the
> licensor!* (Or at least, that is how I am reading it.) The copyright
> claim remains available, but pursuing it creates additional potential
> liability for the claimant's continued use, modification, or distribution
> of the Model Materials. The provision therefore increases the cost of
> obtaining a decision about whether those materials were lawfully published
> in the first place.
>
> Calling copyright litigation a "real threat" explains why model publishers
> would want that protection. It does not establish why using an open-source
> license to provide that protection is acceptable. Many kinds of litigation
> present real threats to licensors. If the existence of a serious threat
> were itself sufficient, the licensor's interest in avoiding litigation
> would become the measure of what conditions an open-source license may
> impose.
>
> Nexus and proportionality (and I must admit, my confidence is not as
> strong on this, so please correct me if I am getting this wrong) may help
> evaluate the boundaries of an otherwise justified termination provision,
> but they do not supply the missing justification. That is, if I am
> understanding this correctly, a nexus tells us that the lawsuit concerns
> the Model Materials, but it does not tell us why bringing that lawsuit may
> properly result in the loss of every permission to use them. On the other
> hand, proportionality requires examining the consequence as well as the
> trigger, and immediate termination of every grant upon filing a claim
> against any person is not proportionate merely because the claim concerns
> those materials.
>
> The question is therefore not whether copyright litigation poses a real
> threat to model publishers. I think we all concede that it can. The
> question is: what open-source principle permits a license to respond to
> that threat by terminating every grant to the claimant?  And, patent
> precedent cannot answer that question unless its patent-specific
> justification also applies to copyright.
>
> And, I have not yet seen an explanation of why it does. Unless that
> explanation can be supplied, the fact that copyright litigation threatens
> model publishers remains a reason why publishers want the condition, not a
> reason for OSI to accept it as a reasonable open-source condition.
>
> Best,
>
> Josh
>
>
> Joshua Gay
> Sr. Manager, Open Source Community and Infrastructure
> IEEE SA
>
> On Thu, Sep 3, 2026 at 5:56 PM Shuji Sado <shujisado at gmail.com> wrote:
>
>>
>> Pamela-san and All,
>>
>> I have been translating almost all of this discussion into Japanese as I
>> work through it, and it has been a genuinely deep discussion. It has also
>> been costing me a little sleep lately. :)
>>
>> I think Pamela-san's latest message usefully identifies the philosophical
>> divide, but Mike's clarification of the scope of the trigger makes me
>> wonder whether the choice is quite as binary as it first appears.
>>
>> Mike has now clarified that "indirectly" in Section 5 is intended to
>> refer to secondary liability, such as a claim that making, using, or
>> distributing the Model Materials contributes to infringement by others. He
>> also clarified that OpenMDW-1.1 does not reach back to artifacts used in
>> creating the model unless those artifacts are themselves included in the
>> subsequently published Model Materials.
>>
>> That distinction matters to me.
>>
>> A claim that copyrighted material was unlawfully acquired or copied
>> during training is not, by itself, the same as a claim that the resulting
>> Model Materials directly or indirectly infringe copyright. So, on Mike's
>> stated interpretation, the picture in which an ordinary creator sues merely
>> over unauthorized use of their work in training and therefore loses the
>> OpenMDW license seems broader than the intended scope of the trigger.
>>
>> The trigger is instead directed at claims against the licensed Model
>> Materials themselves, whether as direct infringement or under a theory of
>> secondary liability. And when I look at that narrower relationship, I still
>> have difficulty finding a reason why copyright should be categorically
>> different from patent.
>>
>> Suppose the copyright holder of novel A is using model X under OpenMDW,
>> and then sues the provider asserting that model X itself infringes the
>> copyright in novel A. The copyright claim does not disappear. The plaintiff
>> remains free to pursue the claim and seek whatever remedies are available.
>> What terminates are the OpenMDW grants the plaintiff received with respect
>> to those Model Materials.
>>
>> In that situation, I am not sure Open Source principles require the
>> plaintiff to continue receiving permission to copy, modify, and distribute
>> the same Model Materials that the plaintiff is asserting are infringing. I
>> also do not think recognizing this necessarily leads to the broader
>> position that any legal claim involving licensed materials may properly
>> trigger termination.
>>
>> Yokoi-san's question and Mike's answer help here as well. OpenMDW grants
>> rights relating to copyright, patent, database rights, and trade secrets,
>> but database-right and trade-secret claims do not trigger termination. Mike
>> explained that the trigger was not designed simply to mirror every right in
>> the grant, but rather around the realistic litigation threats faced by
>> openly published model materials.
>>
>> So perhaps the limiting principle need not be either "patents are
>> historically special" or "any claim against licensed materials may trigger
>> termination." Richard-san's historical point may explain why patent
>> termination developed first, but I am not sure that history by itself
>> establishes a categorical boundary for what may be acceptable today.
>>
>> I am increasingly inclined toward a more general principle:
>>
>> A defensive termination provision may be acceptable where the assertion
>> is directed at the licensed subject matter itself, provided that the scope
>> and consequences of the termination are reasonably related and
>> proportionate to that assertion.
>>
>> This would not make every type of litigation a permissible trigger. It
>> would instead ask whether there is a sufficiently close nexus between the
>> assertion and the licensed subject matter, and whether the termination goes
>> no further than is reasonably justified by that assertion. To me, that
>> seems more durable than drawing the line solely according to whether the
>> asserted right is a patent or a copyright.
>>
>> --
>> Shuji Sado
>> Chairman, Open Source Group Japan
>> https://opensource.jp/
>> English blog: https://shujisado.org/
>> Japanese blog: https://shujisado.com/
>>
>>
>> 2026/9/3 8:46 Pamela Chestek <pamela at chesteklegal.com>:
>>
>>> On 9/2/2026 3:39 PM, McCoy Smith wrote:
>>>
>>> On 9/1/2026 3:34 PM, Richard Fontana via License-review wrote:
>>>
>>> On Tue, Sep 1, 2026 at 5:56 PM Josh Berkus <josh at berkus.org> wrote:
>>>
>>>> I'll stand by my claim, though, that any argument that can be made
>>>> against the principle* of a copyright antisuit provision applies
>>>> equally
>>>> to a patent one.  And that if OSI decides that the former isn't
>>>> OSI-compliant, then the latter isn't either, and we're going to need to
>>>> revisit a lot of licenses.
>>>>
>>>
>>> This is clearly one of the questions the OSI needs to resolve: is there
>>> something special about copyright? Or is there something special about
>>> patent?
>>>
>>> In terms of historical precedent, I think we can leave the
>>> long-abandoned OCLC license aside as a likely case where the early OSI
>>> wasn't thinking about things as carefully as the present-day OSI would
>>> (that's about as charitable as I'm willing to be here). Having looked at
>>> the OCLC license-discuss thread the other day, I don't think anyone paid
>>> any attention to the litigation termination provision. I wonder actually if
>>> some of the license-discuss participants noticed it but thought
>>> "intellectual property" was synonymous with "patent", but I don't think any
>>> of the participants were on the board at the time other than Russ Nelson. I
>>> seriously think it's not beyond the realm of possibility that the OCLC
>>> lawyers who *drafted* the license sincerely thought "intellectual property
>>> infringement litigation" was synonymous with "patent infringement
>>> litigation".
>>>
>>> The OSI should also think about the other side of the question. If a
>>> copyright trigger for license termination (including copyright license
>>> termination) is okay in addition to a patent litigation trigger, what about
>>> other kinds of claims? Trademark infringement? Trade secret? What about
>>> claims concerning things having no connection to the software (or model,
>>> etc.) being licensed? Shareholder litigation? Personal injury? Employment
>>> discrimination?
>>>
>>> The IBM Public License says:
>>>
>>> "If Recipient institutes patent litigation against a Contributor with
>>> respect to a patent applicable to software (including a cross-claim or
>>> counterclaim in a lawsuit), then any patent licenses granted by that
>>> Contributor to such Recipient under this Agreement shall terminate as of
>>> the date such litigation is filed."
>>>
>>> Assuming the OSI would consider this acceptable in an open source
>>> license today, what if instead it said:
>>>
>>> "If Recipient institutes copyright litigation against a Contributor
>>> with respect to a copyright applicable to software (including a cross-claim
>>> or counterclaim in a lawsuit), then any copyright licenses granted by that
>>> Contributor to such Recipient under this Agreement shall terminate as of
>>> the date such litigation is filed."
>>>
>>> Surely that would not be okay ... would it?
>>>
>>> Litigation termination triggers seem to be the one area where the
>>> discussion is converging on: what sort of litigation can properly terminate
>>> grants in a license, and what grants can be terminated?
>>>
>>> So: can patent litigation trigger termination? That's a fairly common
>>> trigger. Against what? The example above from the IBM public license
>>> (OSI-approved https://opensource.org/license/IPL-1.0) the trigger
>>> applies to any patent litigation "applicable to software." I think this
>>> sort of broad trigger is, to put it mildly, disfavored. What about what's
>>> in the currently pending ModelGo-Attribution-2.0 license:
>>> "This License shall terminate immediately ... if You initiate any legal
>>> action against the Licensor alleging that the Licensed Materials and/or
>>> Derivative Materials infringe any patent worldwide."
>>> I believe many feel that extending the trigger to Derivative Materials
>>> (i.e., things beyond that which the patent holder has licensed under the
>>> license) is also too broad (and most of the popular licenses explicitly or
>>> implicitly are not triggered by patent claims against derivative works.
>>>
>>> As to Richard's question, I don't see where the trigger in OpenMDW-1.1
>>> is "concerning things having no connection to the software (or model, etc.)
>>> being licensed." OpenMDW-1.1's trigger says:
>>> "If you file, maintain, or voluntarily participate in a lawsuit against
>>> any person or entity *asserting that the Model Materials directly or
>>> indirectly infringe any patent or copyright,* then all rights and grants
>>> made to you hereunder are terminated, unless that lawsuit was in response
>>> to a corresponding lawsuit first brought against you."
>>>
>>> The license grants copyright and patent rights, and the trigger is a
>>> reciprocal assertion of those same rights, and both apply to the Model
>>> Materials, which is what the licensor licenses and the licensee is licensed
>>> to. And unlike, say, ModelGo-Attribution, the assertion in OpenMDW-1.1 must
>>> be against the "Model Materials" which are defined as "the materials
>>> provided to you under this agreement," not what you were licensed plus
>>> anything else that is subsequently created.
>>>
>>> If people think that patent triggers are OK (and possibly that patent
>>> triggers can, as in MPLv2 & GPLv3, terminate other rights, like copyright),
>>> and copyright triggers are not, that's OK, and we've had people try to
>>> articulate why they believe it is philosophically different, but I'm not
>>> sure it leads to a slippery slope where *any* sort of trigger must
>>> therefore be permissible. And I wonder how that same rationale would deal
>>> with the ModelGo-Attribution-2.0 trigger.
>>>
>>> I believe we are seeing a philosophical divide. I wouldn't consider the
>>> more expansive view a slippery slope, but it is more generous to the
>>> licensor because of the philosophical underpinnings.
>>>
>>> On one side we have the view that, if you use the fruit of my labor,
>>> it's only fair that in exchange you not sue me. If that's the accepted
>>> rationale for the trigger, then as a matter of logic it should be a valid
>>> position for any kind of claim against the licensed materials, whether
>>> copyright, trade secret, patent, database, moral rights, etc. The further
>>> extension that Richard identified was that any claim against the licensor,
>>> i.e., what Facebook's BSD + Patent license tried to do, could also be fair
>>> game. That could be a slippery slope, but I don't think anyone is
>>> advocating for this expansion.
>>>
>>> The other side of the divide is the one that believes patents are a
>>> special case with logic that doesn't extend beyond patents, as articulated
>>> in different ways by different people for different reasons.
>>>
>>> So the question is -- which is the better path?
>>>
>>> Just to clarify a statement you made, the OpenMDW grants patent,
>>> copyright, trade secret and database rights, but only infringement of the
>>> patent and copyright rights are a trigger.
>>> On the ModelGo license question about Derivative Works, I don't think
>>> that's particularly relevant to this license, because the OpenMDW license
>>> doesn't take that position. It might be easier to just stick to the issues
>>> presented in the OpenMDW license.
>>>
>>> Pam
>>>
>>>
>>> Pamela S. Chestek
>>> Chestek Legal
>>> 4641 Post St.
>>> <https://www.google.com/maps/search/4641+Post+St.+%0D%0A++++++Unit+4316+%0D%0A++++++El+Dorado+Hills,+CA+95762?entry=gmail&source=g>
>>> Unit 4316
>>> <https://www.google.com/maps/search/4641+Post+St.+%0D%0A++++++Unit+4316+%0D%0A++++++El+Dorado+Hills,+CA+95762?entry=gmail&source=g>
>>> El Dorado Hills, CA 95762
>>> <https://www.google.com/maps/search/4641+Post+St.+%0D%0A++++++Unit+4316+%0D%0A++++++El+Dorado+Hills,+CA+95762?entry=gmail&source=g>
>>> +1 919-800-8033 <(919)%20800-8033>
>>> pamela at chesteklegal.com
>>> www.chesteklegal.com
>>> _______________________________________________
>>> The opinions expressed in this email are those of the sender and not
>>> necessarily those of the Open Source Initiative. Communication from the
>>> Open Source Initiative will be sent from an opensource.org email
>>> address.
>>>
>>> License-review mailing list
>>> License-review at lists.opensource.org
>>>
>>> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>>>
>>
>>
>> _______________________________________________
>> The opinions expressed in this email are those of the sender and not
>> necessarily those of the Open Source Initiative. Communication from the
>> Open Source Initiative will be sent from an opensource.org email address.
>>
>> License-review mailing list
>> License-review at lists.opensource.org
>>
>> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>>
>
>
> --
>
>
>
> *Joshua Gay*
>
> Sr. Manager SA Open Source Community and Infrastructure
>
> +1 617-966-9792 <(617)%20966-9792> |  standards.ieee.org
>
>
>   <https://www.facebook.com/ieeesa>  <https://www.instagram.com/ieeesa/>
> <https://www.linkedin.com/company/2596931/>  <https://twitter.com/ieeesa>
> <https://www.youtube.com/ieeesa>
>
>   <https://standards.ieee.org/>
>
>
>
> _______________________________________________
> The opinions expressed in this email are those of the sender and not
> necessarily those of the Open Source Initiative. Communication from the
> Open Source Initiative will be sent from an opensource.org email address.
>
> License-review mailing list
> License-review at lists.opensource.org
>
> http://lists.opensource.org/mailman/listinfo/license-review_lists.opensource.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-review_lists.opensource.org/attachments/20260911/0dc5ac79/attachment-0001.htm>