@zkat My main concern with removing freedom zero has less to do with discrimination, and more with the swiss-cheese problem you get in licensing restrictions, once dependencies get involved - as your dependency tree grows, it's going to pull in more and more different licenses, all of which may introduce different restrictions, many of them likely to be vague enough that they 'overcatch'.

So in the end you'd likely end up with a very narrow set of actual allowed usecases, with different usecases being ruled out by different dependencies in the tree. This creates a direct incentive against adding dependencies, which I'd argue goes against the idea of building up a collective public commons, as it gets more difficult with every step.

How would you address that?

@zkat I've long felt that, just like we have a plethora of standard licenses to pick from, we should also have more named and documented governance models, because these two are separate dimensions that tell you something about the "openness" of a project.

You can be closed source but with community (led or involved) governance, you can be (corporate) open source without any community involvement at all, and many shades between. This wouldn't be captured in the license, but in some kind of governance contract with the community.

@zkat have you read https://www.gnu.org/philosophy/programs-must-not-limit-freedom-to-run.en.html :P

@zkat This is good. This is the right kind of thing to do, and the right direction to take it. But I continue to think that licenses are the wrong tool, and the four freedoms are the wrong starting point.

I would start with ideas like the right to repair and the right to (not) associate.

@zkat My initial thought is that a freedom has to be universal. If the issue is capitalistic greed and overreach, and the goal is regulation or restitution/recontribution, we have other societal levers we can (and should) pull for that.

@zkat @thephd I’m not sure whether it’s the same thing, but it certainly seems within the same ballpark, but have you looked into Ethical Source? https://ethicalsource.dev

@zkat Could you elaborate a bit on why it is bad when commercial entities use community software? I can think of reasons, but I'm interested in your reasons.

@zkat the problem with "non-commercial" licenses is everyone has a different definition, and the broadest, most restrictive interpretation usually does not overlap with anyone's actual intent. it always leads to surprises when people actually read the license. mostly what happens in practice is people either avoid NC stuff on principle because it limits the use of their own project too much, or the use it anyway and often end up out of compliance on accident

@zkat liberationware – free software without the Californian ideology, yes please.

@zkat I've seen this pop up in various forms (eg during FOSDEM 2020, I think it was https://archive.fosdem.org/2020/schedule/event/legal_organizers_panel/), but always fail because not allowing to run for any purpose means that the license (or, worse, author) picks what is OK to use it for, and no big enough chunk of the ecosystem agrees on what is OK.
Put commercial use or any concrete business in it, and you lose anyone who gets paid for their work on FLOSS. Put in any ethical criterion, and you'd need millennia of philosophy to conclude.

@zkat I've been thinking along those lines as well, not as carefully or as deeply, to be sure, but one of the pieces that got me down that road is that there's no way with AGPLv3 to express the constraint that "if you scrape this work for AI training, copyright can't stop you, but you lose the benefits of the license." That feels like it points to a pretty fundamental limitation with the intellectual framing of *GPL licenses — Freedom Zero was a mistake, as you say.

@zkat I've been thinking a lot about how to revive copyleft, without the poisoned entanglement of FSF. It's a difficult balance, especially whether to have freedom-zero or not, but I think it's important to put it in the context of today's power dynamic: commercial interests are 10 orders of magnitude richer and more powerful than us.

When the copyleft "community" makes a large body of useful software, the license restrictions are an equalizer. This worked up until TiVo, basically.

@zkat I have also had a lot of similar thoughts of late. The intellectual commons is under attack and it ought not to be surprising that people that contribute to it also act to preserve it for future generations. Defensive licensing seems like a good first step on this road.

@zkat i am opposed to getting rid of freedom zero enough to personally commit to reimplementing anything i see under such a license under my usual choice of 0BSD.

i do not think you can build a functioning labor movement on top of Cop Stuff. and licenses are Cop Stuff. these are the tools of the enemy.

@zkat I agree that there is a problem of exploitation that's bad for the health of the FOSS ecosystem, but I'm not sure getting rid of freedom zero is the way to do it.

A worrying recent example from open source hardware is that Prusa proposed a new license that basically doesn't allow commercial use. Their target is to prevent companies in Shenzhen from undercutting them, but in doing so, they would also prevent people from selling parts for repair or selling improvements to existing designs, which is a hugely important part of what got 3D printing to where it is today. IMO, if this license catches on, the 3D printing community is going to collapse and it'll just be corporations in the space.

While there's definitely a need to change the structures and tools we use to defend against corporate capture, I worry your proposed solution could backfire, especially if the solution entails disallowing specific uses in the license, since that's one area where corporate trends move too quickly (notably, how short-lived the NFT trend was, and how quickly AI became the "next big thing").

My other concern is enforceability, i.e. writing an actually valid legal document and setting precedent in courts, but that seems like a solvable problem.

@zkat i used to like the term "public good" but i think most definitions of it (eg https://en.wikipedia.org/wiki/Public_good ) still have a lot of propertarian (thus exploitable) baggage. software isn't just information obviously, it performs functions, and very different consequences flow from who is able to access those functions. so "common" is still probably a better fit and i appreciate your ideas and reasoning here.

@zkat we have several open source organizations in France around the idea of the Commons. Would that fit your idea of another vision of open source ?

@zkat A variant of the idea: if you do not (clearly) meet the requirements of the license, you must notify the author of the software of your usage of it and provide access to your source code (publicly or privately?), and the author may choose to void/veto your license for use.

This notification could conceivably be some standardized/(semi-)automated process, something built on open standards, to make it actually viable in practice.

@zkat Hmm, that's an interesting idea. I'll have to think about that for a while!

The only concern that comes to mind right now is how to remove the ability for said actors to section off their usage of the thing into some separate project with a HTTP API or whatever while keeping the rest of their system proprietary and secret. But that's a more general virality question, I suppose.

@neil @zkat I think a big part of why freedom zero exists is because removing it creates a wealth of problems that nobody wants to deal with. Every intention can be flipped on its head and ultimately you can't win from that.

But even in the realm of existing FOSS, we already know that there are things they won't touch: software under GNU licenses, software that is demonstratively social, and software that is architected to maximize reciprocity.

Use those in anger if you wish. 😈

@zkat @joepie91 right now, half* the hacker circles are explicitly in favor of OSS being used by the MIC to the point where talks about Kubernetes on a [vehicle]** is considered normal or actively desirable. this is a problem! until we can change that norm, i don't think that enforcing it will result in any meaningful difference

* not literally half. a very significant amount
** there's apparently quite a few of these deployments. i think the talk i remembered was by SonarQube...

@whitequark @joepie91 this all feels like barely doing anything too, though. They can just maintain their own forks, pretty trivially, with their own patches. Something you could do even if you had the additional stumbling block of unfriendly licensing.

@joepie91 idk that sounds like a pain. I have a lot of users.

@zkat @joepie91 I mean same, my position is "you will not receive support, you will not be accepted in the community, and your PRs will not be merged" (which turns out[1] to be incredibly controversial even though it's like the lowest bar imaginable)

I think the really important change here is cultural: to kick these people out of our shared communities. licenses are a distraction from that.

[1]: https://github.com/povik/yosys-slang/pull/248

@zkat @joepie91 how are you so certain the MIC are compliant? why would an entity that by definition produces classified work, inexaminable by the general public, care about what license it's under?