4 ms·

> The notion is that the software they build belongs to them. They possess a sense of ownership over the software, which comes with a set of moral and perhaps l

> The notion is that the software they build belongs to them. They possess a sense of ownership over the software, which comes with a set of moral and perhaps legal rights to the software, which, importantly, are withheld from any entity other than themselves. The “developers” enjoy this special relationship with the project – the “developers” being the special class of person entitled to this sense of ownership and the class to whom the up-and-coming source-available movements make an appeal, in the sense of “pay the developers” – and third-party entities who work on the source code are merely “contributors”, though they apply the same skills and labor to the project as the “developers” do. That's... literally just proprietary software, with "source code available on request"?

abetusk 2y ago
This is talking about models that publish their code ("source available") but only allow sub-licensing through explicit negotiation. Source available is a type of proprietary software but a specialization thereof that DeVault is highlighting.
TeMPOraL 2y ago
That's every OSS product, though. It's like with Git - theoretically, the system is designed around a mesh topology, where there's an amorphous blob of clones exchanging patches, without any designated special repo; in practice, everyone pulls from and pushes to a designated "origin", creating a tree in which the root owns the canonical version of the source code. Same with any OSS project - there's always one designated organization or group that holds the moral, legal, and/or assumed right to project identity, which includes both the name and the source code. Simple proof: fork any OSS project you like, make some changes, and start promoting it under the same name. See how long you'll last before first accusations of theft roll in. OSS is a strange beast - the source code itself isn't something scarce you can make money off, but the identity of a project and associated reputation very much is. Which gives you the "first party"/"third party" developer split, regardless of whether the payment is in dollars or (with most OSS) just Internet clout.
- armchairhacker 2y ago
  In OSS projects the source code is available in general, not “on request”. On request means they can deny you the code, or provide you with a version that isn’t what they distribute. Also, if you fork an OSS project and promote it under the same name you’d be (IMO rightfully) criticized, but if you fork and promote it under a different name, and clearly credit the original, I believe most people wouldn’t care. At worst your project gets no attention, or maybe the owner cares, but if your clone is superior people will move to it. This has precedent, albeit with dead projects, like tui-rs becoming ratatui; with projects that became closed-source, like MySQL; with small patches of big projects, like linux and firefox; and where the fork has a small audience. But active projects that aren't becoming closed-source probably tend not to have substantially different, popular forks, simply because the effort to maintain such a fork to par with the original is too much, and not adequately justified by the motive. Moreover, it’s legal to fork an OSS project and start distributing it on your own. If the project is truly open-source and not copyleft, it’s even legal to close-source and sell your fork. It’s legal to directly copy parts of the open-source project into your competing project, with attribution, rather than having to “clean-room engineer” or find a different algorithm to avoid being accused of copyright. It’s legal to provide “third-party” support to the open-source project that is superior to the owner’s first-party support they are trying to monetize. This seems to me the crux of DeValt’s argument: if you create an open-source project you don’t own the project identity. You own the name, but that doesn't mean much, because someone can fork the project with a similar name and everyone can informally know it as “new [your project’s name]”. You absolutely don’t own the code legally. And people will say and act as if you own the project, but only if you don’t abandon the project or try to make it closed-source; because it’s a lot more effort to hard-fork a project with an active maintainer, which will have no initial audience and constantly need to be synced or fall behind, especially when you can just create a pull request. It’s true that most open-source projects have one repository and an “in-group” that are de-facto “owners”, but ultimately their ownership is conditional on people’s acceptance. If the people switch to a competitor who literally copies their code, the original owners have no power to stop them, as has happened to MySQL/MariaDB, Terraform/OpenTofu, and Redis/Redict. This is functionally different than legal ownership of “source available” code, which may be legally protected from derivation, and especially “open core” where there are parts of the code (blobs) you physically can’t modify or in some cases even use outside the main distribution.

abetusk 2y ago

This is talking about models that publish their code ("source available") but only allow sub-licensing through explicit negotiation. Source available is a type of proprietary software but a specialization thereof that DeVault is highlighting.

TeMPOraL 2y ago

That's every OSS product, though. It's like with Git - theoretically, the system is designed around a mesh topology, where there's an amorphous blob of clones exchanging patches, without any designated special repo; in practice, everyone pulls from and pushes to a designated "origin", creating a tree in which the root owns the canonical version of the source code. Same with any OSS project - there's always one designated organization or group that holds the moral, legal, and/or assumed right to project identity, which includes both the name and the source code. Simple proof: fork any OSS project you like, make some changes, and start promoting it under the same name. See how long you'll last before first accusations of theft roll in. OSS is a strange beast - the source code itself isn't something scarce you can make money off, but the identity of a project and associated reputation very much is. Which gives you the "first party"/"third party" developer split, regardless of whether the payment is in dollars or (with most OSS) just Internet clout.

armchairhacker 2y ago
In OSS projects the source code is available in general, not “on request”. On request means they can deny you the code, or provide you with a version that isn’t what they distribute. Also, if you fork an OSS project and promote it under the same name you’d be (IMO rightfully) criticized, but if you fork and promote it under a different name, and clearly credit the original, I believe most people wouldn’t care. At worst your project gets no attention, or maybe the owner cares, but if your clone is superior people will move to it. This has precedent, albeit with dead projects, like tui-rs becoming ratatui; with projects that became closed-source, like MySQL; with small patches of big projects, like linux and firefox; and where the fork has a small audience. But active projects that aren't becoming closed-source probably tend not to have substantially different, popular forks, simply because the effort to maintain such a fork to par with the original is too much, and not adequately justified by the motive. Moreover, it’s legal to fork an OSS project and start distributing it on your own. If the project is truly open-source and not copyleft, it’s even legal to close-source and sell your fork. It’s legal to directly copy parts of the open-source project into your competing project, with attribution, rather than having to “clean-room engineer” or find a different algorithm to avoid being accused of copyright. It’s legal to provide “third-party” support to the open-source project that is superior to the owner’s first-party support they are trying to monetize. This seems to me the crux of DeValt’s argument: if you create an open-source project you don’t own the project identity. You own the name, but that doesn't mean much, because someone can fork the project with a similar name and everyone can informally know it as “new [your project’s name]”. You absolutely don’t own the code legally. And people will say and act as if you own the project, but only if you don’t abandon the project or try to make it closed-source; because it’s a lot more effort to hard-fork a project with an active maintainer, which will have no initial audience and constantly need to be synced or fall behind, especially when you can just create a pull request. It’s true that most open-source projects have one repository and an “in-group” that are de-facto “owners”, but ultimately their ownership is conditional on people’s acceptance. If the people switch to a competitor who literally copies their code, the original owners have no power to stop them, as has happened to MySQL/MariaDB, Terraform/OpenTofu, and Redis/Redict. This is functionally different than legal ownership of “source available” code, which may be legally protected from derivation, and especially “open core” where there are parts of the code (blobs) you physically can’t modify or in some cases even use outside the main distribution.