[License-discuss] Pre-submission discussion: Zero-Clause BSD with Patent Grant (0BSD-P)

[Rob Landley]

On 10/11/26 05:21, Михаил Чернов wrote:
> Subject: Pre-submission discussion: Zero-Clause BSD with Patent Grant
> (0BSD-P)
> 
> Hello,
> 
> I am considering submitting the Zero-Clause BSD with Patent Grant (0BSD-P)
> to OSI for approval. Before doing so, I would like to get feedback from the
> community.

Wouldn't you have to submit diffs to the project in order to grant any 
of your patents? (Mere use ain't gonna cut it.)

Which means this doesn't prevent patent trolls from suing the project 
for unknowing violations.

> The gap I see is this: no OSI-approved license currently combines zero
> attribution requirements with an express patent grant.

Because it's a bad idea.

Nobody who has 5-6 figures to obtain and defend a patent wants zero 
attribution requirements. (Yes 5-6 figures per patent: your writeup and 
supplementary filings have to convince a patent clerk to approve the 
thing and show you didn't invalidate your own patent with prior art, 
while you CAN do your own taxes here it's still going to be hundreds of 
hours of work including coming up to speed on the legal niceties, prior 
art search, navigating the bureaucracy... Budgeting your _own_ time at 
$50 hour means you hit 5 figures at 200 hours, ignoring office space and 
health insurance and the taxes your employer doesn't pay half of when 
you're self employed. And that's before you get into _international_ 
patents, because otherwise somebody will just do a chinese version and 
bounce it off singapore and ship a production run of inventory into the 
distribution channel and disband a shell corporation before the product 
hits shelves and you don't even have standing to complain until long 
after the fact...)

The patent grant part kinda REALLY WANTS the attribution. There's a 
chain of custody with patent grants, that's why people who actually have 
patents tend to use huge boilerplate things like the apache license.

A patent grant with zero attribution seems equivalent to abandoning the 
patent. You want to encourage somebody to suck in code to another 
project without keeping track of where it came from and who granted it 
for use in what, which means someone can always (even retroactively) say 
"oh I got that function from $PROJECT and then adjusted the code to my 
tastes but trust me it's really from that other thing so no you can't 
enforce your patent against me".

Except that's not what you did, you did:

 > The patent license shall not apply to any other combinations which
 > include this software.

Which seems to mean the last version I contributed to is frozen in time, 
nobody else can add another patch to it after that or my patent grant is 
invalidated. (Because those modified versions are another combination 
that includes this software. If that's NOT the case then what do you do 
about forks that make arbitrary changes but claim to still be the same 
thing?)

> Today, a developer who wants to release code without any attribution
> obligation has two OSI-approved options: 0BSD and MIT-0. Neither of them
> contains a patent grant.

Personally I'm against it. Even if what you did COULD work, I think it 
creates confusion. 0BSD is a copyright license. You can add your own 
patent grant without modifying the copyright license. (And you're still 
ignoring trademarks, those are usually separately licensed as well even 
in GPL projects. Why is that ok but separately licensing the patents isn't?)

> A developer who wants patent protection has options
> like BSD-2-Clause-Patent and Apache-2.0,

It doesn't protect you from being sued by someone who isn't an author of 
the project. It only protects you from being sued by someone who 
explicitly contributed to the project.

Contributing an infringement of your own patent without disclosure and 
then turning around and suing the project for using your own code seems 
like something there's probably a countersuit for? (If the only reason 
this patent is infringed by this project is because you explicitly 
created the infringement you're suing against... Seems easier to just 
find projects already infringing and sue those with clean hands?)

"This license does not require you to strictly track authorship of 
projects, code goes into a fungible pool"... is not how patents work. 
Not even for the https://openinventionnetwork.com/ people explicitly 
trying to create a big open source patent pool, who have SCRUPULOUS 
tracking of what _versions_ of what projects are and aren't covered. (I 
get emails. They have _conferences_.)

> This means that for small utilities, scripts, generated code, or similar
> artifacts where attribution is impractical or undesirable, there is no
> OSI-approved way to also grant patent protection. The 0BSD-P is meant to
> fill
> exactly that gap.

"Attribution is impractical or undesirable" plus "I'm going to license 
my patents to it".

Those two sentences do not generally go together, in my experience. The 
form to abandon a patent is 
https://www.uspto.gov/sites/default/files/documents/aia0024.pdf and I 
assume that's what you do if you want to assure everybody it will never 
be enforced again even if the company that currently owns it gets bought 
by scumlords.

> I have attached the full text. I would appreciate any feedback, especially
> on:
> 
> 1. Whether this gap is real and worth filling.
> 2. Whether the combination is correct.
> 3. Whether there are any concerns about OSD compliance.
> 
> Thank you for your time.

I don't have a vote at OSI, but personally I'm not a fan of this.

My goals with 0BSD were to give developers of the increasing amount of 
"no license specified" code on github a comfortable and easily 
understandable option, while also allowing Fortune 500 corporations to 
comfortably use and contribute to such projects.

Communicating to Fortune 500 companies that contributing code to a 0BSD 
project potentially means abandoning ALL your patents (be sure to use 
0BSD and not 0BSD-P), because you didn't even list which specific patent 
numbers were covered in which jurisdictions, and such a list would have 
been lost in explicitly allowed "no attribution" copies anyway, except 
for the above sentence that says you lose the patent grand if you modify 
the work in any way...

What happens if they contribute a patch to the mailing list and you make 
your own arbitrarily modified copy that their patch applies to your 
modified version, which says that the method of submission to the 
project becomes important, a git pull against a specified base vs a 
patch to a list, but if you can "git format-patch" to take a patch out 
of git (standard tool!) and apply it to your modified base as the last 
patch you haven't changed the license terms (or git fast-forward, 
another standard tool) and I have no idea what a judge would say about 
any of this...

There are reasons I never went there.

Rob