The legal doctrine of promissory estoppel is generally considered to protect open source licensees. If you license something for free, and people come to rely on that free licensing, they generally have a right to keep using it, even if you change your mind and try to revoke it later. You can, however, stop licensing the software to new parties.

Novus seems to be trying to get around this by claiming that the license was never valid to begin with, because it was issued by a rogue employee. However, I would argue that the doctrine of apparent authority applies here. That is, to a potential licensee, there was no reason to believe that the open source licensing was anything but company-sanctioned. (The rules for apparent authority are actually a bit more nuanced than that, but the main point is the same.) Thus, even if the employee did indeed act without authorization, I think Novus would still be bound by the license.

Novus seems to be on shaky legal ground, and I find its cease-and-desist questionable. Unfortunately, it would appear that the recipients of the cease-and-desist opted to comply rather than risk a fight. So the scary thing is not that companies can arbitrarily revoke an open source license--in fact, they can't. Rather, it's that a letter containing vague legal threats can have such a strong chilling effect.

It seems like if the code is considered "stolen" there must be some legal common sense. I would also imagine the longer the code stays as open source, the less likely you'd be able to claim theft. If you immediately took it down claiming copyright that would be one thing. If you knowingly left it up for a year, though, that would certainly be a different situation.

While I think there is a lot of legal subtlety at play in the nv3d case, I think your case is a lot more clear cut. You never had the legal authority to make the Windows 8 source code GPL in the first place, so the code isn't GPL, no matter what the README you attached to it says and people continuing to distribute it would be in clear violation of copyright.

Your Windows 8 example is pretty clear cut, sort of like if I steal your car and resell it to someone else. Just because they bought the car doesn't mean they own it, it still belongs to the original owner though the person who bought it clearly has a legal case for restitution against the person who sold it to them without owning it.

It would be different but it would still be utterly unlike what jarrett describes.

1. Estoppel is only going to get you a very limited set of rights. It's highly unlikely a judge is going to find you have the same rights as the open source license, unless you were actually relying on all of those rights. You certainly would not be able to sublicense those rights further, except to those you had already. This is not the same as Novus's ability to "stop licensing the software to new parties", it means you would not be able to give others the rights you had gotten through estoppel.

In effect, estoppel mainly going to protect you from damages, not give you the right to use it as open source.

2. It certainly depends, but the apparent authority question is a lot closer than you make it out to be. There are plenty of cases in the US where "the nature of the transaction" should have caused one to question whether the employee had authority, etc. Not just that, but in New York state, where Novus Partners is, the law is nowhere near like you make it out to be. AFAIK, under New York State law, the apparent authority doctrine will hold a principal responsible for its agent’s actions as long as the principal clothed the agent with apparent authority. Novus Partners would have had to have done something explicit to make you believe this person had authority to open source.

See Hallock v. State, 64 N.Y.2d 224, 231 (1984).

  “Essential to the creation of apparent authority are words
  or conduct of the principal, communicated to a third party,   
  that give rise to the appearance and belief 
  that the agent possesses authority to enter into a  
  transaction,”

  An agent can never “by his own acts imbue himself with
  apparent authority,” Id.

  “[T]he existence of ‘apparent authority’ depends upon a   
  factual showing that the third party relied upon the
  misrepresentation of the agent because of
  some misleading conduct on the part of the principal — not 
  the agent,” Id.

  “Moreover, a third party with whom the agent deals may rely 
  on an appearance of authority only to the extent that such   
  reliance is reasonable,” Id.

Source: I'm a registered patent attorney and corporate IP lawyer who has been doing open source lawyering for many many years now.

"It's highly unlikely a judge is going to find you have the same rights as the open source license, unless you were actually relying on all of those rights. You certainly would not be able to sublicense those rights further, except to those you had already. This is not the same as Novus's ability to "stop licensing the software to new parties", it means you would not be able to give others the rights you had gotten through estoppel."--I agree 100%. My apologies to other readers if I implied the opposite. I do still think that estoppel would protect licensees themselves insofar as they had relied on the software in, e.g., the development of their businesses.

Your point about sub licensing is well taken and deserves further exploration. It would appear that, estoppel or not, those who are now rereleasing the software may be on shaky legal ground. As you said, estoppel would probably not give you the right to sublicense, which is effectively what these folks are doing.

"Essential to the creation of apparent authority are words or conduct of the principal, communicated to a third party, that give rise to the appearance and belief that the agent possesses authority to enter into a transaction"--Agreed. I was under the impression this had transpired in the present case. I'm not sure about New York, but in many jurisdictions a principal's failure to act (given some other criteria) can be enough. I was thinking that would come into play here. Naturally it would depend on the exact facts, of which I am uncertain.

I would also raise the question of jurisdiction. Novus may be in New York, but what of licensees in other states? They could potentially make the argument that their states' laws apply, because part of the "transaction," i.e. the downloading and licensing of the software, occurred there. Would this argument succeed? I dunno.

  and people come to rely on that free licensing

  there was no reason to believe that the open source
  licensing was anything but company-sanctioned

Estoppel

Estoppel in English law is a doctrine that may be used in certain situations to prevent a person from relying upon certain rights, or upon a set of facts (e.g. words said or actions performed) which is different from an earlier set of facts.

Estoppel could arise in a situation where a creditor informs a debtor that a debt is forgiven, but then later insists upon repayment. In a case such as this, the creditor may be estopped from relying on their legal right to repayment, as the creditor has represented that he no longer treats the debt as extant. A landlord may tell his tenant that he is not required to pay rent for a period of time ("you don't need to pay rent until the war is over"). After the war is over, the landlord would be "estopped" from claiming rents during the war period. Estoppel is often important in insurance law, where some actions by the insurer or the agent estop the insurer from denying a claim.

http://en.wikipedia.org/wiki/Estoppel_in_English_law

This smells like the whole Twitter Bootstrap thing a few weeks ago, but the bootstrap guys had enough pull to take the brand with them after they left Twitter.

> What prevents other open source projects from being taken down with a "management did not authorize this" notice?

Retracting an open-source product is a move without a lot of upside. What business goal is promoted by such a retraction? It seems like it will just generate controversy, tarnish the company's reputation, and lead to endless ownership fights with contributors.

In addition, I suspect that major open-source projects usually actually do have the approval of people who have the authority to make that decision.

> What happens to the commits by other authors to the source tree? Do they own the copyright to their commits, even if they modify invalid open source code?

My understanding is that a contributor (or his employer) owns the copyright to his own patches when they are written. Larger open-source projects often require contributor agreements before they'll accept patches; the contributor must legally give the copyright to the project as a condition of their patch being incorporated into the official tree. If there's no contributor agreement in place, the patches continue to belong to the contributor.

You can think of the pre-patch tree and the patch as two parent nodes of the patched version. Novus owns the pre-patch tree; the contributor owns the patch; the post-patch tree is a derivative work of both of them, and can only be distributed with permission of both owners.

The contributor's patches may be useless without the parent tree to patch against. But if the contributors own the copyright to their patches, they can still use that copyright to forbid Novus from using or distributing the patched child tree.

> How does the open source community react when this happen?

Read the Google group and see. My feeling of how they should react is by the contributors banding together and telling Novus the following:

We contributed patches to Novus based on the understanding that the patched software would be released publicly as open source.

As soon as Novus became aware of the situation, it made a clear, unambiguous statement that Novus is not, and never was, willing to agree to these terms.

Therefore, since Novus does not accept the terms under which we gave them the patches, we revoke all permission for Novus to use these patches, or any version of the software which includes them.

If the contributors do this, and Novus is using the project internally, then Novus will have to either (1) back down and say that they're okay with open-source after all, (2) spend engineering resources on proprietary reimplementation of the features the community gave them for free, or (3) live without those features. Only option (1) lacks significant cost and/or risk from Novus's point of view.

> Perhaps there are reasonable solutions to these

This suggests that the more contributors an open-source project has, the stronger it is against any one person or company claiming ownership in this way. The remaining contributors can band together in response and pull out their patches, leaving the proprietary project at a feature-poor, ancient version -- especially compared to people's still-fresh memories of the open-source version -- if not making it entirely nonfunctional. The contributors could even attempt to make their patches useful again with an independent implementation which presents the same interface as Novus's now-proprietary code. Or they could toss their patches and rewrite the library from scratch. It would presumably take much less effort because, while they can't re-use the proprietary code from the Novus version, it should be okay to re-use the design decisions and API that may have been a big reason that the Novus version was so successful.