Are old copies of "mistakenly licensed" OSS still subject to their previous licenses, even after "fixing"?

Question

I want to publish the source of a program which borrows some code from project A. Years ago, A's maintainers forked it from project B (which has no apparent license) and licensed it as MIT. Several months of development later, B's author approached A and and demanded that A's code be licensed with non-commercial requirement. A's maintainers complied by changing the license of B to CC-BY-NC-SA 4.0. However, a range of commits with the MIT license are still available online.

Question: Are the old commits in project A with MIT still considered MIT? If so, I'd be able to publish my codebase under MIT instead of using CC-BY-NC-SA.

Answer 1

Are the old commits in project A with MIT still considered MIT?

No, that code was never licensed under the MIT license: the individual(s) that made that offer to you never hard the right to make it in the first place. You may as well ask if the deed to the Brooklyn Bridge that you bought in an alleyway is still valid, even if the state of New York told you it isn't.

Of course, the final answer here depends on the terms of the agreement between B and A. I suppose it's not impossible that B said to A, "If you license your code under CC BY-NC-SA, then I'll allow you to offer my code under the MIT license," but I would not assume that was the case until I'd verified that agreement with B. It's possible (and I'd wager more likely) that B said something closer to, "I'll allow you and only you to distribute my code, under terms that don't allow downstream redistribution, if you license your code under CC BY-NC-SA."

If the most recent revision of the code doesn't indicate that the code is under an MIT license, I would assume B has never allowed any version of their code to be used under that license.

Answer 2

A had “no apparent license”. Without a license, you can do what copyright law allows you to do, which is very little. So It was not legal for B to publish A under the MIT license. Legally, it was never published under that license. Instead, you were duped into thinking it was.

Now A is offering a license. B can now be published with a valid license, but not with the MIT license. What it means to you is that you used B without a license (without knowing) and now you can use it with a license. You may need to change how you use B obviously. But then you didn’t have a license to use it at all n

Answer 3

Material that is published under any license at all is copyrighted, and that copyright ought to be formally registered with the government. Whether or not it has been registered, however, it is "owned." And, "ownership has its privileges."

One of those privileges is that the owner can choose to license it any way he likes, and he can change that license at any time. When someone "forks" the code and does something else with it, they are creating what is called a derivative work. The creator of a derivative work is entitled to obtain his own copyright on the derivative he has created, but it is subordinate to the original work. There are certain things that he can do, and others that he can't. For example, he can't do what this person attempted to do: namely, to replace the copyright terms with a less-restrictive (or even, "meaningfully different") one, thereby compromising the legal rights of the owner of the intellectual property from which the derivative work was derived. The owner had a legal right to demand the change, and the offender – who was in the wrong – had no choice but to promptly comply.

Although these copyright licenses are designed to be "open," they have sharp legal teeth. They have been tested and upheld in courts all around the world. When you own something, you can be as "open" or as restrictive as you care to be. You still have enforceable rights.

See this article by an IP attorney for a much better discussion.