Exactly. I’m all for AGPL and have no problem with this license. But the CLA gives them propriety of the entire code base, so I’m asking: what prevents them from pulling an ElasticSearch move and say “we’re going ‘source available’ now, it’s kinda like FLOSS, but not really”
If the code was own by multiple people and covered by AGPL, this would prevent such a move.
The most likely use for the CLA is to sell/use exceptions to the AGPL terms. Purely as one data point, the FSF finds selling exceptions to be an acceptable practice: https://www.fsf.org/blogs/rms/selling-exceptions
The FSF never sells exceptions; whatever license or licenses we release a program under, that is available to everyone.
But we need not insist that companies follow that principle. I consider selling exceptions an acceptable thing for a company to do, and I will suggest it where appropriate as a way to get programs freed.
I typically don’t object to seeing a CLA, but a CLA sure does make me far more hesitant to contribute upstream. I think if the project maintainers are fine with getting fewer code contributions, then that is fine too. If they take their project proprietary at some point, then folks can still fork the most recent AGPL version.
EDIT: A CLA could also let them easily relicense to a better license down the road, if one happens to appear.
I do agree with you in spirit though, as I find the sqlite approach to be more virtuous in my mind.
As a reminder, freedom from competition is not one of the core freedoms of any FSF/GNU manifesto. Nor is suppression of competition a pragmatic goal of the Open Source movement.
Yet all of these re-licensings have in common that their main and obvious goal – more or less openly admitted in this post – is to ensure that a core person or team get to have freedom from competition, by imposing burdens on potential competitors that put them at too much of an effective disadvantage.
No amount of talking about “freedoms for our community” or “open source will always be at the core of who we are” can get around this. Nor can it change the fact that when these relicensings happen, people who previously were participants in an open project effectively become unpaid interns working on improving the “open source” company’s product.
A reminder that GNU is not open source but free software, that open source was a word used commonly at the time and redefined to mean something new and that the point of the GNU license has been and always will be freedom of the end user, not the person who owns the machine or source code.
The amount of revisionism from people who want to make money off the backs of software others made would be disappointing if not completely expected. In todays world the AGPL is the minimal license that any software which pretends to care about users freedom should be.
In todays world the AGPL is the minimal license that any software which pretends to care about users freedom should be.
Maybe the issue is that none of the participants really care about the end user. To me it sounds more like tool vendors who want to make money off selling support for software tooling to companies. Nothing wrong with that, but the incentives are slightly different.
One (unintended?) consequence of removing monetary considerations from FLOSS software is that a signal path is removed. If you make software for pay, and you’re losing money, you can check what your paying customers are unhappy about and adjust from that.
A user’s preference for Free Software as opposed to Open Source software is much less pronounced. In fact, I’d argue that the vast majority of users don’t even know the difference (as evinced by the comment I’m replying to). This leaves the next level - producing FLOSS software but charging for consulting and support (as suggested by the FSF). And then the signal is there - companies using FLOSS software and who might be prepared to pay for these services avoid GPL licenses and prefer permissive ones.
A reminder that GNU is not open source but free software
A reminder that the very first sentence of my comment began by pointing out:
freedom from competition is not one of the core freedoms of any FSF/GNU manifesto.
Meanwhile:
The amount of revisionism from people who want to make money off the backs of software others made would be disappointing if not completely expected.
A reminder that the essentially ideal software business, as advocated by the Free Software Foundation, Richard Stallman, and the Free Software movement, consists of selling the effort of software, which includes not just new greenfield development projects, but also maintenance and improvement of existing software which, yes, often will have been written by others (since in the ideal Free Software world, a recipient of software always has the freedom to perform or contract out such maintenance and improvement).
And a reminder that this wave of recent relicensings is entirely about ensuring that the VC-backed SaaS startup of the original authors of (insert project name) get to monetize the work of everyone who contributed under the prior more-Free license, while nobody else does.
In todays world the AGPL is the minimal license that any software which pretends to care about users freedom should be.
A further reminder that if you check my history, I do not personally consider the AGPL to be a Free Software license, and point out at every opportunity that it only is considered one by the FSF due to a combination of fiat (rather than adherence to principle) and an explicit AGPL-specific exception clause added to the GPLv3. The plain GPL does not allow passing on less freedom to downstream recipients than you yourself received from upstream. The AGPL requires that you pass on less freedom in certain circumstances. The moment that principle of conservation of freedom is compromised, the entirety of the Free Software moral position is not merely compromised, but utterly collapses beyond repair.
I do not personally consider the AGPL to be a Free Software license
Any further discussion is moot, and anything you’ve said has been pedantry at best and pointless at worst.
I am reminded of the discussions in the 90s as to why the GPLv2 was not free software and why the only real free software license was the MIT license for exactly the same reasons you state here.
Those discussions were pointless then, and this is pointless now.
Free Software does not mean “Free as in freedom from competition”. This relicensing is clearly being done, like all the others in this recent wave of relicensings, to make it more difficult for others to compete with a specific SaaS business. The tool used to accomplish it in this case is AGPL + a CLA, which ensures that the original SaaS business can continue to monetize the work of those who contributed back, but nobody else can.
You at first seemed to think that that last bit was a bad thing. Have you changed your mind in the past few hours? Does it change your mind now if you remember that Grafana itself started out as a fork of another project?
There is no high and lofty principle of freedom being fought for here. It’s about using copyright to squash the ability of others to compete, which is the antithesis of Free Software. And it’s using the AGPL to do this, not as some abstract hypothetical FUD you can reflexively dismiss, but as the actual thing that’s actually happening in the actual case at hand. This is why people have concerns about the AGPL, because of how often in the real world we really see it really being used for the real purpose of protecting a business’ monopoly over a software project.
This seems to help me, as a user, get contributions from other users upstream.
If I use grafana on Amazon’s hosted product, I’m paying them to change or improve it, but then lose those changes to Amazon. I would like to pay for a product, like grafana, but not lose improvements. This seems like it fixes that situation while still allowing me to choose my preferred grafana vendor.
The AGPL does not protect you against the original author, or against the original author prioritizing their business model over your needs.
All of these recent relicensings have in common that the original author(s) (really, their VC backers) are prioritizing their business model over your needs.
This criticism doesn’t make sense since nothing stops the authors from changing the license at any point as is. The GPL and the AGPL were always about dealing with the free-rider problem where people want to extend off existing software but not contribute the changes back. It that breaks your business model, too bad, you aren’t entitled to the work of the authors.
It sucks if you were a contributor and you expected the software to stay on the Apache license. If you were in that boat, I’d hope you could comply with the new license without too much trouble.
You don’t have to contribute the changes back, you have to make the source code of the software available if you distribute it later. I was more stating motivation than a license clause.
It that breaks your business model, too bad, you aren’t entitled to the work of the authors.
But the authors are claiming to be entitled to the work of others. People who are genuinely concerned about a free-rider problem don’t talk about the need to protect their business and go to anticompetitive licensing with CLA for “participants” in the project.
They are not entitled. Those people freely contributed it. And if they did so under an expectation that the more permissive license would stay, the option always exists to fork at the commit before the license changes over.
As a reminder, freedom from competition is not one of the core freedoms of any FSF/GNU manifesto. Nor is suppression of competition a pragmatic goal of the Open Source movement.
So what?
“Open source” and “free” are neither precisely defined nor inherently good. Their value isn’t self evident, it’s a function of the positive effects they have on real human beings. And not abstract or theoretical effect in some model of reality, but real and observable effect in reality as it exists. Does freedom from competition in this specific circumstance have a net positive effect for human beings? I think you can at least argue that the answer is yes. If you don’t agree, let’s have that debate, but let’s have it on the merits, the FSF/GNU definitional stuff is just semantic noise.
I’m surprised by that. I think very few people are aware of, much less mean to convey, the specificity in the OSI’s and FSF’s definitions of “open source” and “free” when they use those terms.
If you don’t agree, let’s have that debate, but let’s have it on the merits, the FSF/GNU definitional stuff is just semantic noise.
When a relicensing whose goal is to use copyright to maintain a monopoly or monopoly-like position on commercial exploitation of a piece of software wraps itself in a cloak of high idealist language about “freedom” and “openness”, arguing about the “semantic noise” of “the… definitional stuff” is a perfectly valid response.
When a relicensing whose goal is to use copyright to maintain a monopoly or monopoly-like position on commercial exploitation of a piece of software…
…for which they are they principal authors and sole maintainers…
wraps itself in a cloak of high idealist language about “freedom” and “openness”, arguing about the “semantic noise” of “the… definitional stuff” is a perfectly valid response.
OK. They claim
AGPLv3 is an OSI-approved license that meets all criteria for Free and Open Source Software.
for which they are they principal authors and sole maintainers
Let me state the argument, as clearly as I can, in order not to get dragged off into derailments like this.
Yes, copyright law allows this. Yes, that is a fact.
However, the blog post uses the language of the Free Software and Open Source movements, which tend to look beyond the mere minimal test of “does copyright law allow this”. The Free Software movement, for example, is founded on the idea that using one’s copyright in software to control what others may do with the software, for the purpose of creating and maintaining a monopoly on commercial exploitation of the software, is morally evil. Yet that is the effect of this relicensing: attempting to create a de facto commercial monopoly over the software, by controlling what others may and may not do with the software, via copyright.
Yes, the AGPL is considered a Free Software license by the FSF. However, I do not see a way to reconcile the practical effects of this use of the AGPL with the ideals the FSF alleges Free Software should involve.
Therefore I criticize the post and the relicensing on grounds that they appropriate language and ideals to which they do not live up and which, in fact, are antithetical to at least one such set of appropriated ideals.
By your own acknowledgement, they’re abiding the letter of the law, so to speak; your criticism is that they’re violating the spirit of it, by appropriating language inappropriately? I mean, I guess my counterargument would be that this is an expression of your opinion, and therefore unfalsifiable, so 🤷
If you really want to show you’re doing the right thing, ditch the CLA.
Exactly. I’m all for AGPL and have no problem with this license. But the CLA gives them propriety of the entire code base, so I’m asking: what prevents them from pulling an ElasticSearch move and say “we’re going ‘source available’ now, it’s kinda like FLOSS, but not really”
If the code was own by multiple people and covered by AGPL, this would prevent such a move.
The most likely use for the CLA is to sell/use exceptions to the AGPL terms. Purely as one data point, the FSF finds selling exceptions to be an acceptable practice: https://www.fsf.org/blogs/rms/selling-exceptions
I suppose they could do what sqlite does (“Open-Source, not Open-Contribution”), and just not accept patches?
I typically don’t object to seeing a CLA, but a CLA sure does make me far more hesitant to contribute upstream. I think if the project maintainers are fine with getting fewer code contributions, then that is fine too. If they take their project proprietary at some point, then folks can still fork the most recent AGPL version.
EDIT: A CLA could also let them easily relicense to a better license down the road, if one happens to appear.
I do agree with you in spirit though, as I find the sqlite approach to be more virtuous in my mind.
As a reminder, freedom from competition is not one of the core freedoms of any FSF/GNU manifesto. Nor is suppression of competition a pragmatic goal of the Open Source movement.
Yet all of these re-licensings have in common that their main and obvious goal – more or less openly admitted in this post – is to ensure that a core person or team get to have freedom from competition, by imposing burdens on potential competitors that put them at too much of an effective disadvantage.
No amount of talking about “freedoms for our community” or “open source will always be at the core of who we are” can get around this. Nor can it change the fact that when these relicensings happen, people who previously were participants in an open project effectively become unpaid interns working on improving the “open source” company’s product.
A reminder that GNU is not open source but free software, that open source was a word used commonly at the time and redefined to mean something new and that the point of the GNU license has been and always will be freedom of the end user, not the person who owns the machine or source code.
The amount of revisionism from people who want to make money off the backs of software others made would be disappointing if not completely expected. In todays world the AGPL is the minimal license that any software which pretends to care about users freedom should be.
… where “end user” is narrowly defined. I’m not opposed to the FSF’s project, but it is very rooted in a specific context.
Maybe the issue is that none of the participants really care about the end user. To me it sounds more like tool vendors who want to make money off selling support for software tooling to companies. Nothing wrong with that, but the incentives are slightly different.
One (unintended?) consequence of removing monetary considerations from FLOSS software is that a signal path is removed. If you make software for pay, and you’re losing money, you can check what your paying customers are unhappy about and adjust from that.
A user’s preference for Free Software as opposed to Open Source software is much less pronounced. In fact, I’d argue that the vast majority of users don’t even know the difference (as evinced by the comment I’m replying to). This leaves the next level - producing FLOSS software but charging for consulting and support (as suggested by the FSF). And then the signal is there - companies using FLOSS software and who might be prepared to pay for these services avoid GPL licenses and prefer permissive ones.
A reminder that the very first sentence of my comment began by pointing out:
Meanwhile:
A reminder that the essentially ideal software business, as advocated by the Free Software Foundation, Richard Stallman, and the Free Software movement, consists of selling the effort of software, which includes not just new greenfield development projects, but also maintenance and improvement of existing software which, yes, often will have been written by others (since in the ideal Free Software world, a recipient of software always has the freedom to perform or contract out such maintenance and improvement).
And a reminder that this wave of recent relicensings is entirely about ensuring that the VC-backed SaaS startup of the original authors of (insert project name) get to monetize the work of everyone who contributed under the prior more-Free license, while nobody else does.
A further reminder that if you check my history, I do not personally consider the AGPL to be a Free Software license, and point out at every opportunity that it only is considered one by the FSF due to a combination of fiat (rather than adherence to principle) and an explicit AGPL-specific exception clause added to the GPLv3. The plain GPL does not allow passing on less freedom to downstream recipients than you yourself received from upstream. The AGPL requires that you pass on less freedom in certain circumstances. The moment that principle of conservation of freedom is compromised, the entirety of the Free Software moral position is not merely compromised, but utterly collapses beyond repair.
Any further discussion is moot, and anything you’ve said has been pedantry at best and pointless at worst.
I am reminded of the discussions in the 90s as to why the GPLv2 was not free software and why the only real free software license was the MIT license for exactly the same reasons you state here.
Those discussions were pointless then, and this is pointless now.
Free Software does not mean “Free as in freedom from competition”. This relicensing is clearly being done, like all the others in this recent wave of relicensings, to make it more difficult for others to compete with a specific SaaS business. The tool used to accomplish it in this case is AGPL + a CLA, which ensures that the original SaaS business can continue to monetize the work of those who contributed back, but nobody else can.
You at first seemed to think that that last bit was a bad thing. Have you changed your mind in the past few hours? Does it change your mind now if you remember that Grafana itself started out as a fork of another project?
There is no high and lofty principle of freedom being fought for here. It’s about using copyright to squash the ability of others to compete, which is the antithesis of Free Software. And it’s using the AGPL to do this, not as some abstract hypothetical FUD you can reflexively dismiss, but as the actual thing that’s actually happening in the actual case at hand. This is why people have concerns about the AGPL, because of how often in the real world we really see it really being used for the real purpose of protecting a business’ monopoly over a software project.
This seems to help me, as a user, get contributions from other users upstream.
If I use grafana on Amazon’s hosted product, I’m paying them to change or improve it, but then lose those changes to Amazon. I would like to pay for a product, like grafana, but not lose improvements. This seems like it fixes that situation while still allowing me to choose my preferred grafana vendor.
The AGPL does not protect you against the original author, or against the original author prioritizing their business model over your needs.
All of these recent relicensings have in common that the original author(s) (really, their VC backers) are prioritizing their business model over your needs.
This criticism doesn’t make sense since nothing stops the authors from changing the license at any point as is. The GPL and the AGPL were always about dealing with the free-rider problem where people want to extend off existing software but not contribute the changes back. It that breaks your business model, too bad, you aren’t entitled to the work of the authors.
It sucks if you were a contributor and you expected the software to stay on the Apache license. If you were in that boat, I’d hope you could comply with the new license without too much trouble.
Could you please state were in the GPL it states you have to contribute changes back? I don’t think that’s actually in the license.
You don’t have to contribute the changes back, you have to make the source code of the software available if you distribute it later. I was more stating motivation than a license clause.
But the authors are claiming to be entitled to the work of others. People who are genuinely concerned about a free-rider problem don’t talk about the need to protect their business and go to anticompetitive licensing with CLA for “participants” in the project.
They are not entitled. Those people freely contributed it. And if they did so under an expectation that the more permissive license would stay, the option always exists to fork at the commit before the license changes over.
So what?
“Open source” and “free” are neither precisely defined nor inherently good. Their value isn’t self evident, it’s a function of the positive effects they have on real human beings. And not abstract or theoretical effect in some model of reality, but real and observable effect in reality as it exists. Does freedom from competition in this specific circumstance have a net positive effect for human beings? I think you can at least argue that the answer is yes. If you don’t agree, let’s have that debate, but let’s have it on the merits, the FSF/GNU definitional stuff is just semantic noise.
I think “open source” and “free” are well-defined by the OSI and the FSF, respectively.
Do you believe those definitions are what most people mean when they use the terms?
…yes?
I’m surprised by that. I think very few people are aware of, much less mean to convey, the specificity in the OSI’s and FSF’s definitions of “open source” and “free” when they use those terms.
When a relicensing whose goal is to use copyright to maintain a monopoly or monopoly-like position on commercial exploitation of a piece of software wraps itself in a cloak of high idealist language about “freedom” and “openness”, arguing about the “semantic noise” of “the… definitional stuff” is a perfectly valid response.
…for which they are they principal authors and sole maintainers…
OK. They claim
Is that not true?
Let me state the argument, as clearly as I can, in order not to get dragged off into derailments like this.
Yes, copyright law allows this. Yes, that is a fact.
However, the blog post uses the language of the Free Software and Open Source movements, which tend to look beyond the mere minimal test of “does copyright law allow this”. The Free Software movement, for example, is founded on the idea that using one’s copyright in software to control what others may do with the software, for the purpose of creating and maintaining a monopoly on commercial exploitation of the software, is morally evil. Yet that is the effect of this relicensing: attempting to create a de facto commercial monopoly over the software, by controlling what others may and may not do with the software, via copyright.
Yes, the AGPL is considered a Free Software license by the FSF. However, I do not see a way to reconcile the practical effects of this use of the AGPL with the ideals the FSF alleges Free Software should involve.
Therefore I criticize the post and the relicensing on grounds that they appropriate language and ideals to which they do not live up and which, in fact, are antithetical to at least one such set of appropriated ideals.
Do you have a counterargument to this criticism?
By your own acknowledgement, they’re abiding the letter of the law, so to speak; your criticism is that they’re violating the spirit of it, by appropriating language inappropriately? I mean, I guess my counterargument would be that this is an expression of your opinion, and therefore unfalsifiable, so 🤷
So your position is that anything which is legally permitted is morally good and should not be criticized?