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

Sat Aug 29 00:26:44 UTC 2026

On 8/28/26 15:10, McCoy Smith wrote:
> On 8/28/2026 12:00 PM, Rob Landley wrote:
>>
>> AI's need for Lebensraum is irrelevant to whether copyright 
>> termination for an assertion of copyright infringement is compatible 
>> with the open source definition.
>>
>> I would argue that it is not, nor should it be. If this was a term of 
>> GPLv2, no company would have ever touched it. It attempts to give 
>> sufficiently popular projects blanket permission to infringe any 
>> copyright.
> 
> As Richard previously established, GPLv3 allows copyright termination 
> for an assertion of patent infringement: https://lists.opensource.org/ 
> pipermail/license-review_lists.opensource.org/2026-August/006142.html. 

Patents are not copyrights. I am automatically granted a copyright for 
typing this sentence. Obtaining a patent is a long expensive process 
requiring manual approval from professionals.

If you believe they are the same thing, show me one example of a class 
action suit where more than a hundred different patentholders came 
together to complain that each one had a patent that had been infringed 
by the same individual.

Because this is common in the copyright world. Here's an example of a 
copyright class action suit from 10 years ago having nothing to do with 
"AI" or even software:

https://authorsguild.org/advocacy/accomplishments/multimillion-dollar-freelance-settlement/

> The question in my mind then, is what's the difference between copyright 
> termination for a patent assertion (OK, at least under GPLv3 and MPLv2) 
> and copyright termination for copyright assertion (several have argued 
> that that is impermissible)?

The copyright is what is BEING licensed. Most open source projects don't 
have any patents, but all of them have copyrights. Licensing copyrights 
is the primary purpose of open source licenses. A copyright suit 
_against_ the project is an assertion that some of the copyrights they 
are licensing do not belong to them, that the entity suing is the actual 
copyright holder, and that the material covered by the licenses has thus 
been misrepresented.

Copyright and patent are fundamentally different. I have literally 
thousands of copyrights. My wife has copyrights. My mother had 
copyrights. My brother and sister each have copyrights. With the 
possible exception of my brother, I believe all of us have licensed them 
professionally at one point or another.

None of us have patents. (It's possible my father's name has been _on_ 
patents over a 30 year career as an electrical engineer, but if so they 
belonged to his employer.)

Almost the entire population of the united states has copyrights. You 
would have to have never written or recorded anything to not _have_ a 
copyright.

According to https://fred.stlouisfed.org/series/PATENTUSALLTOTAL less 
than 200k patents are granted in the USA annually. I'm aware Gemini 
provided a 350k annual figure when I searched for that, before I 
scrolled down to a source I trust.) Either way even if each patent was 
given to a different individual US resident (which is not the case), 
that's less than 1% of the population receiving patents each year. Which 
then expire after 20 years. (Copyrights last at _least_ a century, 
usually longer.)

And that's before you get into domain specificity: I doubt anything I've 
uploaded to github infringes on a pharmaceutical manufacturing update 
for selecting specific isomers from an otherwise generic drug. Cuts the 
number of patents applicable to software projects down quite a bit, I 
expect.

The surface area for patent plaintiffs to object to what is 
fundamentally a copyright license to begin with is much, much smaller. 
The number of copyrights "in play" at any given time (and thus the 
potential number of plaintiffs who may have grounds to sue) are multiple 
orders of magnitude different.

> A few have invoked "software 
> freedom" (which I'm not sure how we measure that -- is it the OSD, is it 
> the FSD, or something else?) or more explicitly -- Simon in particular 
> -- Freedom Zero. Freedom Zero says  you have "[the freedom to *run* the 
> program as you wish, for any purpose" https://www.fsf.org/blogs/ 
> community/your-digital-life-is-not-yours [emphasis added] That's not a 
> copyright right.* It's probably a patent right ("use"). So I'd like to 
> understand more about how we get to the rationale that if you sue for 
> patent infringement, it's OK to terminate (and that doesn't violate 
> Freedom Zero), but it's absolutely impermissible to have the same for 
> copyright infringement.

There's no first sale doctrine for patent rights. It very different law.

> Especially since, I think, the mere act of 
> running software is probably not something you're going to be able to 
> sue for copyright infringement anyway.

Launching a binary means copying the software from storage into memory, 
and often saving it to swap space. (Yes this got relitigated back in the 
90s, I think it was settled rather than forming a new precedent but only 
because 
https://en.wikipedia.org/wiki/Sony_Corp._of_America_v._Universal_City_Studios,_Inc. 
was waved prominently around. I think I wrote a Motley Fool column 
mentioning it but am not digging for it.)

Since then the DMCA got passed with anti-circumvention provisions and 
they threw a TPM and signing chain in there so arbitrary DRM enforcement 
could extrajudicially create felonies if you tried to get time shifting 
or first sale doctrine back. It's terrible and complicated and handing 
the people who want to extend it more weapons seems like a terrible idea.

> FWIW, if your answer is "it's not really grounded in the OSD, or FSD, 
> but is more of a gestalt principle around my sense of what is 'software 
> freedom,'"

In 2006 I filed GPL enforcement lawsuits on behalf of busybox. (Well, 
Eben Moglen filed them, I got the ball rolling and signed paperwork.)

If GPLv2 had this copyright termination clause in it, and any of the 
companies we sued had ever contributed code to a GPL project (kernel 
etc: some of these were big companies with a lot of employees touching a 
lot of projects), they could say "you have filed a copyright enforcement 
suit against us: you OWN right to use busybox is terminated". (Because 
we took external contributions, and could no longer distribute THAT code 
without a license to it.)

Or would the judge merely have terminated our right to use the kernel? 
Do you consider either case compatible with the open source definition? 
(How about Freedom 9? If I filed suite against the copyright holder, is 
the termination limited to JUST that package, or any other packages 
under the same license they have code in?)

> that's OK, but you have to understand that OSI is going to 
> need to understand the metes and bounds of that principle as we evaluate 
> license submissions (both plain old software licenses, and AI licenses) 
> going forward, and reasonable minds will likely differ on what is that 
> gestalt.

If you want the headline to be "Hugging Face convinced OSI to approve a 
license that automatically terminates for anyone who tries to enforce a 
legitimate copyright claim", go for it.

> *I understand there is some case law in the USA about ephemeral copies 
> potentially being the subject of copyright infringement under the 
> copyright right to control copying/reproduction, but those tend to be 
> ones that fall outside the bounds of 17 USC 117(a)(1), which would 
> likely be implicated in the specific scenarios invoked here.

Oh there's YEARS of back and forth on that. Stallman was originally 
trying to do GPLv3 to close what he called the "application service 
provider" loophole where GPL code would run on a web server and the code 
itself never be distributed outside it, but data was processed by it and 
he wanted to change GPL so he could not get THAT code, but had the right 
to install modified versions of it. (And I pointed out that if the World 
of Warcraft servers were running busybox, and I had a WoW subscription, 
they did not owe me root access to the servers to run arbitrary code there.)

All this was 20+ years ago. Very old well-trodden ground even before the 
big GPLv3 push circa 2006. The "tivoization" stuff was an early version 
of leveraging DMCA anti-circumvention provisions to attach a signing key 
to authorized binaries so "distributing tools" to let you run different 
code on your own device potentially becomes a federal felony. Old old 
old: 
https://www.eff.org/pages/unintended-consequences-fifteen-years-under-dmca

Most of my vagueness about this issue is "wow, the last time people 
argued about this was so LONG ago, this was settled what-a-terrible-idea 
from the dot-com era". In theory, copyright termination means you can't 
launch a _countersuit_. (The person claiming you've mis-licensed their 
code means they're claiming to have code under your license, and 
therefore if you countersue for legal fees or some such: boom, you've 
terminated your own license. If you didn't 100% author all the material 
you're distributing, you can't redistribute anybody ELSE'S code under 
that license now, because you sued.)

There's just all sorts of landmines here from "should a putative GPLv3 
do THIS" discussions at LinuxWorld Expo and so on. (The hallway track 
after Stallman's panel in... I want to say 2000? California I think? I'm 
pretty sure it was the one mentioned at 
https://www.fool.com/archive/portfolios/rulemaker/2000/02/25/inside-intel-again-merced-vs-crusoe.aspx 
not the New York one the same year because I got all the casette tapes 
from that one and don't remember Stallman's speech being among them. But 
it was a quarter century ago now.)

There's a reason people haven't made this common practice before now.

Rob