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

[Pamela Chestek]

On 8/29/2026 2:41 PM, Richard Fontana wrote:
> On Sat, Aug 29, 2026 at 2:58 PM Pamela Chestek <pamela at chesteklegal.com> wrote:
>> There seem to be two remaining points of contention (correct me if I
>> missed any others):
>>
>> 1. Whether a copyright trigger for termination is appropriate for an
>> OSI-approved open source license; and
>>
>> 2. What Model Materials are included in the termination provision.
> Not sure if anyone else sees the problem I'm seeing but I think the
> "Model Materials" issue is more complex than that of mere uncertainty
> over what is actually included in "Model Materials". Even if you had
> certainty over the contents of Model Materials, but certainly if you
> don't, there is an issue that affects both OpenMDW-1.0 (though I
> didn't see it before) as well as OpenMDW-1.1 (ignoring the copyright
> trigger), which I'd state as follows:
>
> *For a license containing a patent-assertion-triggered termination to
> be approved by OSI as "open source", where assertion of a patent
> against thing A gives rise to termination of licenses covering thing
> B, are there any limits to how *unrelated* A and B can be?
>
> I know that historical precedent may suggest the answer is "no", but
> the most recent well known license to apparently raise this sort of
> issue, some 9 years ago, was the controversial Facebook React license,
> which I referred to in my response to Josh, never actually submitted
> to the OSI (though I remember that Facebook came close to doing so).
> For convenience, here was its termination provision:
>
> "The license granted hereunder will terminate, automatically and without notice,
> for anyone that makes any claim (including by filing any lawsuit, assertion or
> other action) alleging (a) direct, indirect, or contributory infringement or
> inducement to infringe any patent: (i) by Facebook or any of its subsidiaries or
> affiliates, *whether or not such claim is related to the Software*, (ii) by any
> party if such claim arises in whole or in part from *any software, product or
> service of Facebook or any of its subsidiaries or affiliates, whether or not
> such claim is related to the Software*, or (iii) by any party relating to the
> Software; or (b) that any right in any patent claim of Facebook is invalid or
> unenforceable. [emphasis added]
>
> As far as I recall, the OSI never took an official position on this
> license (for the record, inside Red Hat I insisted it should be
> considered open source and I remember finding the three-minute-hate
> being applied to Facebook over the license quite appalling), but
> plenty of people claimed to see some problem with it that was so
> severe that it shouldn't be regarded as open source.
>
> I'm not suggesting that OpenMDW-1.x raises exactly the same issue, but
> I contend that the nature of the "Model Materials" concept, which
> collects different types of materials that conventionally tend to be
> subjected to different kinds of licenses -- remember that this is the
> whole point of OpenMDW -- and which might *not* even be distributed
> together (as I have explained), raises a similar sort of issue.
>
> To address this issue, I think it might be helpful to look at some
> actual cases of "Model Materials" that have been licensed under
> OpenMDW-1.x. I'll try to do that when I have a chance.
>
>> Given the amount of controversy over the copyright trigger, I suggest,
>> as Mike offered, that we instead consider the OpenMDW-1.0 license for
>> approval and defer question (1) to a later time when the lawfulness of
>> using copyrighted works for training is more settled.
> I agree that this would be a good idea, but in part because the second
> problem as I tried to formulate it above is possibly already
> complicated enough.
>
>> That leaves only question (2). Richard commented:
>>
>>> Maybe that doesn't matter for OpenMDW-1.0, because of an uncodified
>>> "software patents are such bad things that we can go really far in
>>> disincentivizing patent assertions" principle, but OpenMDW-1.1 has to
>>> be analyzed differently because it reaches licensee copyright claims.
>> Does considering only OpenMDW-1.0 reduce the concerns about the license
>> enough that it could be approved?
> It's clearly a less difficult problem, but I don't think it should be
> seen as a sort of slam-dunk case for approval. An obvious case for
> approval would be a version of OpenMDW without any defensive
> termination provision, or, perhaps, with patent-triggered defensive
> termination that is decoupled somehow from the concept of Model
> Materials. Without the defensive termination, the nature of Model
> Materials doesn't really matter.
>
> What I would not want to see is OSI approving OpenMDW-1.*0* (let alone
> OpenMDW-1.1) without explicitly addressing the problem I described
> above. I again suggest that the concern might be grounded in OSD 9
> ("The license must not place restrictions on other software that is
> distributed along with the licensed software.").

I think this problem dovetails with a comment I made before, which is 
the fourth scenario of licensing which is $LICENSE + OpenMDW. Mike's 
response was:

> a fourth configuration: Y unmodified, under both MIT and OpenMDW-1.1. 
> I should have included it. I do not think it changes the outcomes. MIT 
> grants permission directly to any person obtaining a copy, and every 
> recipient of the distribution has obtained a copy, so B's direct grant 
> reaches each recipient in parallel with whatever narrower sublicense A 
> conveys. A's layering cannot subtract from B's own offer. The layered 
> consequence you describe, where noncompliance with the second license 
> costs the licensee the licensor's own materials but not the 
> third-party code, is how multi-license distributions already work 
> today, including in your GPL example. And no configuration changes the 
> termination analysis, because termination reaches only rights and 
> grants made hereunder, and B's MIT grants are not among them.
I don't think that's complete, though. Take Model Materials that have a 
model under only the OpenMDW license (because it's original to the 
licensor) and inference code that is MIT + OpenMDW (because it is third 
party code). A user brings a patent infringement suit against the 
inference code, with the result that the license to the model terminates 
too, even though the model is entirely separate from the MIT inference 
code. Is that an OSD9 scenario?

I suspect, though, that the model owner will still have patent 
infringement risk for its distribution of the MIT-licensed code, if 
that's relevant.

Pam

Pamela S. Chestek
Chestek Legal
4641 Post St.
Unit 4316
El Dorado Hills, CA 95762
+1 919-800-8033
pamela at chesteklegal.com
www.chesteklegal.com