I don’t see the Zero Clause BSD license mentioned here which, due to its simplicity, is my favorite solution to this problem.
Instead of attempting to give up copyright and assign the software to the public domain, the 0BSD license allows anyone to do anything with the software with no obligations or restrictions.
This is true, but compared to CC0 it does still lack the third tier of the release: the declaration that the copyright owner will not take action in case some rights can’t be relinquished or licensed.
This may a problem at least in Europe because there are rights which can’t be abandoned, transferred, or licensed: moral rights. Authors (and their heirs) in Europe always have the right to require they are named as the author of their work, and to object to demeaning treatments of their work.
0BSD is probably acceptable for US authors, because the US does not recognize any inalienable moral rights distinct from the transferrable usage rights.
There has been some historical concern surrounding CC0’s patent clause, which unintuitively may be problematic because it’s explicitly mentioned rather than remaining implicit.
Due to that, I have also been favoring 0BSD as the most simple approximation of public domain that’s legally viable.
EDIT: I know you mention the issue, but until it’s really tested in international courts, that’s the main reason I’ve stayed away from CC0 in favor of 0BSD’s simplicity.
Has there ever been a case of a publicly licensed work causing trouble for someone because the author later decided to do something unexpected with their moral rights?
Given the rather absurd copyright limit of 70 years after author’s death, I don’t think we’ve reached the stage where enough creators are dead so that their legatees can file suits arguing that the PD assignment was incorrect.
I personally went the dual licence route for Monocypher, in the name of assuaging the fear of people who aren’t familiar with CC0. Though I didn’t quite realise at the time about the contradiction, and how doing so might render my 2-clause BSD void, or worse, void CC0 itself in some jurisdictions. My counsel at the time was right about one thing though: CC0 alone wouldn’t have cut it. It pains me to see that big open source organisation who should know what they’re doing, are being afraid over nothing.
Meanwhile, I’m still wondering whether something that dynamically links against a library is a derivative work of that library…
Last time I talked to a lawyer about it, dual licensing was a minefield. In particular, if one license grants a subset of the rights of another then a ‘reasonable person’ (test used in court) would never accept the license that gave them fewer rights and so it becomes meaningless. This makes GPL or BSD dual licensing pointless in a lot of jurisdictions because the legal assumption is that no one would ever accept the GPL (it provides them with fewer rights, in exchange for nothing) and so it is equivalent to BSD. Where the licenses have different terms (for example, GPL and CDDL) the situation gets more complex and it’s not clear what happens in a bunch of corner cases. Most of the case law is around commercial licenses where different people have different terms, or where you need to accept one of the licenses to do something specific (for example, commercial use) so doing something makes it clear which license you accepted.
I also saw a fun case some years ago where someone tried to dual license Wikipedia contributions as public domain and GFDL. You cannot simultaneously assert and disclaim copyright (CC0 has specific verbiage that says that the license applies only if the attempt to disclaim copyright fails).
I don’t think any of this applies to you, but it convinced me that talking to a lawyer about it over dinner was fun, talking to a lawyer about dual licensing because I needed to care about dual licensing was something I actively wanted to avoid.
Meanwhile, I’m still wondering whether something that dynamically links against a library is a derivative work of that library…
My bet: linkage is actually irrelevant to the question, and if courts and legislative bodies ever come to any conclusion on this topic, it won’t look anything at all like what the FSF imagines on the topic. I wouldn’t be surprised if the LGPL and GPL were rendered functionally identical in some jurisdictions.
To be fair, the FSF does not believe that linkage is relevant to the question so I’m curious in what specific way you think they will turn out to be wrong.
The FSF claims the type of linkage is irrelevant to the question, not that linkage itself is irrelevant. They think writing a program designed to link with GPL software necessarily creates a derivative work, but writing a program designed to communicate with GPL software by any means other than linking only sometimes does.
AFAIK they’ve never given any concrete examples of those times. Their interpretation seems to be motivated by (amongst other things) an intuitive idea about what a “program” is, which I’m not sure would necessarily be relevant to all courts.
SQLite is simultaneously: 1. in the public domain, 2. one of the most distributed and used pieces of software in existence. As long as this remains true I will ignore anyone who invents imaginary problems with public domain software.
SQLite’s authors are in the United States, where a public domain dedication is clearly recognized in law. This is not true worldwide, which is part of the point of my essay.
SQLite’s authors also fund themselves to develop the software in part by selling ‘warranties of title’ to companies who are concerned that using public domain software might pose some kind of legal risk. So your example is based on software which is arguably exploiting the uncertainties about ‘imaginary problems’ as part of getting used and distributed so widely.
I don’t see the Zero Clause BSD license mentioned here which, due to its simplicity, is my favorite solution to this problem.
Instead of attempting to give up copyright and assign the software to the public domain, the 0BSD license allows anyone to do anything with the software with no obligations or restrictions.
https://opensource.org/license/0bsd
This is true, but compared to CC0 it does still lack the third tier of the release: the declaration that the copyright owner will not take action in case some rights can’t be relinquished or licensed.
This may a problem at least in Europe because there are rights which can’t be abandoned, transferred, or licensed: moral rights. Authors (and their heirs) in Europe always have the right to require they are named as the author of their work, and to object to demeaning treatments of their work.
0BSD is probably acceptable for US authors, because the US does not recognize any inalienable moral rights distinct from the transferrable usage rights.
There has been some historical concern surrounding CC0’s patent clause, which unintuitively may be problematic because it’s explicitly mentioned rather than remaining implicit.
Due to that, I have also been favoring 0BSD as the most simple approximation of public domain that’s legally viable.
EDIT: I know you mention the issue, but until it’s really tested in international courts, that’s the main reason I’ve stayed away from CC0 in favor of 0BSD’s simplicity.
Has there ever been a case of a publicly licensed work causing trouble for someone because the author later decided to do something unexpected with their moral rights?
Given the rather absurd copyright limit of 70 years after author’s death, I don’t think we’ve reached the stage where enough creators are dead so that their legatees can file suits arguing that the PD assignment was incorrect.
I was also about to suggest that. I only use the ISC or 0BSD licenses. They are suckless.
I personally went the dual licence route for Monocypher, in the name of assuaging the fear of people who aren’t familiar with CC0. Though I didn’t quite realise at the time about the contradiction, and how doing so might render my 2-clause BSD void, or worse, void CC0 itself in some jurisdictions. My counsel at the time was right about one thing though: CC0 alone wouldn’t have cut it. It pains me to see that big open source organisation who should know what they’re doing, are being afraid over nothing.
Meanwhile, I’m still wondering whether something that dynamically links against a library is a derivative work of that library…
Last time I talked to a lawyer about it, dual licensing was a minefield. In particular, if one license grants a subset of the rights of another then a ‘reasonable person’ (test used in court) would never accept the license that gave them fewer rights and so it becomes meaningless. This makes GPL or BSD dual licensing pointless in a lot of jurisdictions because the legal assumption is that no one would ever accept the GPL (it provides them with fewer rights, in exchange for nothing) and so it is equivalent to BSD. Where the licenses have different terms (for example, GPL and CDDL) the situation gets more complex and it’s not clear what happens in a bunch of corner cases. Most of the case law is around commercial licenses where different people have different terms, or where you need to accept one of the licenses to do something specific (for example, commercial use) so doing something makes it clear which license you accepted.
I also saw a fun case some years ago where someone tried to dual license Wikipedia contributions as public domain and GFDL. You cannot simultaneously assert and disclaim copyright (CC0 has specific verbiage that says that the license applies only if the attempt to disclaim copyright fails).
I don’t think any of this applies to you, but it convinced me that talking to a lawyer about it over dinner was fun, talking to a lawyer about dual licensing because I needed to care about dual licensing was something I actively wanted to avoid.
My bet: linkage is actually irrelevant to the question, and if courts and legislative bodies ever come to any conclusion on this topic, it won’t look anything at all like what the FSF imagines on the topic. I wouldn’t be surprised if the LGPL and GPL were rendered functionally identical in some jurisdictions.
To be fair, the FSF does not believe that linkage is relevant to the question so I’m curious in what specific way you think they will turn out to be wrong.
The FSF claims the type of linkage is irrelevant to the question, not that linkage itself is irrelevant. They think writing a program designed to link with GPL software necessarily creates a derivative work, but writing a program designed to communicate with GPL software by any means other than linking only sometimes does.
AFAIK they’ve never given any concrete examples of those times. Their interpretation seems to be motivated by (amongst other things) an intuitive idea about what a “program” is, which I’m not sure would necessarily be relevant to all courts.
SQLite is simultaneously: 1. in the public domain, 2. one of the most distributed and used pieces of software in existence. As long as this remains true I will ignore anyone who invents imaginary problems with public domain software.
SQLite’s authors are in the United States, where a public domain dedication is clearly recognized in law. This is not true worldwide, which is part of the point of my essay.
SQLite’s authors also fund themselves to develop the software in part by selling ‘warranties of title’ to companies who are concerned that using public domain software might pose some kind of legal risk. So your example is based on software which is arguably exploiting the uncertainties about ‘imaginary problems’ as part of getting used and distributed so widely.
Publish under a pseudonym.
It doesn’t really solve the problem described in the article, does it? The user is still at risk if: