If I get something from this post and the "Are we still open source?" slides linked in the sibling comment [0], is that Open Source is a dangerous route for software that wants or expects to eventually be commercialized. I've read countless threads on HN and that seems to always be the final lesson. There are lower risk but less "idealistic" licenses such as Apache, and higher risk but also ideal licenses such as GPL.
Especially relevant to GPL, it ends up being counterproductive, as the license is a hardcore protection of the users' rights, at the expense of the project owner's rights. See the debate caused by Octotree [1], [2].
It's a pity because the way I see it, GPL is the ideal way to go. But if you need to pay the bills with software you write, it is also one of the worst open source choices out there.
(A)GPL is then relegated to either 100% vocational non-commercial code that one writes as a side project, or to big entities that for some reason write GPL code that is not part of their main business.
GPL is only becoming irrelevant because nobody has the guts to use it.
Congratulations to all the teams getting co-opted by Amazon and becoming just another Docker Enterprise.
Everyone should go GPL. The time to do it was yesterday. The next best time is now.
Make money with licenses so you can actually have a business. Unless you just like funding billion dollar companies for free. (They don't even have to contribute code back!)
Open source without fangs is free, stolen labor that gets turned against us.
GPL software can be monetized and locked away behind cloud services. If it's never actually distributed to users, then those users don't have rights to the source code under the terms of the GPL.
While adding the problem that the AGPL is worded so poorly that no one knows where its virality stops (or even if copyright possibly can be bent so far enough that that virality is enforceable) and everyone is rightfully afraid of it until such time a court of law actually manages to make heads or tails of it and set a precedent. No company I'm aware of wants to be the guinea pig subject to those court costs, much less trying to explain to a judge technical implications of the license. (That includes companies using the AGPL; I've yet to hear of a compny willing to stake their companies on answering that second question of whether or not the virality of the license is legally enforceable in the confines of current understandings of copyright law.)
It's hard for me to dismiss what is the closest to a "Mutually Assured Destruction Pact" in software licenses as merely "FUD".
Absolutely the commercial license business model is driven by that threat of "nuclear" retaliation. But that doesn't necessarily make that business model correct. (In some ways it seems rather skeevy, negotiating with someone that has an untested "nuclear device" and is threatening to use it.)
Also, just because it is untested doesn't make it any less of a nuclear device. Einstein and Bohr and a few other mathematicians had a very good idea of what a nuclear bomb could do before one was ever built or tested. We can make, based on the (poor) wording of the AGPL, some pretty good assumptions about how badly it could be used in a court of law should someone actually try to call in their threatened strike. (Everyone is too afraid to let that happen, but should it happen, it may be devastating, and worse no one knows in which direction: if the AGPL will prove to be so virulent as to be toxic or if the AGPL will open questions about enforce-ability of copyleft clauses in what are ostensibly presented as copyright licenses that risk even the GPLv2 falling to a dangerous counter-precedent. Either way, a source for great damage to the software industry should the AGPL ever go to court.)
Depends whether you value money more than code contributions.
I don’t think it would hurt to have AWS release some of their software, like the Nitro hypervisor. The world would probably benefit from that more than any license they’d have to pay.
I haven't looked at the AGPL in a long time; it is more viral than the normal GPL? My recollection was that the key difference is that "distribution" that triggers the requirement to share changes is expanded to include running it in a SaaS environment, but that's not really the part of the GPL that's "viral." Is there another difference, or what?
This gets into the long history of the GPL versus the LGPL and lots of legal battles in the text of those licenses and in the interpretation of the text in those licenses about the differences between types of "distribution" ("mere aggregation" versus "linking" versus "bundling" versus "inclusion"). In all but the "mere aggregation" case, the GPL is considered to be viral, and the LGPL relaxes that in some more "linking"/"bundling" scenarios.
The AGPL to do its work has to close the "mere aggregation" "hole" in at least some cases, which alone makes it "more viral" on the spectrum of GPL licenses. That's before you get into problems with how the APGL defines "some cases", in that it doesn't do a great job so no one knows exactly how viral it is because it will take court cases to sort out its bad wording on the subject.
(For instances: 1) The AGPL is intended to expand to include "running it in a SaaS environment" yet the way it (doesn't) define "SaaS environment" is so poor it could mean running it anywhere. 2) The AGPL was written by a PHP company and so a lot of the definitions, distinctions, and clauses from the GPL regarding the differences between binary and source distributions were struck out. Because most of the known legal definitions on "mere aggregation" with regards to the GPL hinge on those distinctions we have fewer and fewer ideas of what "mere aggregation" means with respect to the AGPL, and it again possibly implies that the AGPL's virality is potentially unbounded, that their "some cases" is intentionally or accidentally written as "all cases".)
Most of the complaints about AWS are not about license compliance and source code release, but about money and competition.
AGPL doesn't solve the AWS problem, which is that Amazon will take the project your open source company wrote and soak up all the money to be had offering that project as a service and they will do a better job of it, while still complying with the license.
A related problem is that Amazon have enough engineers they can just look at your proprietary or source available project and just clone your programming interfaces on top of a new project (and if you are unlucky, open source that, killing your market entirely).
This kind of comment does little to advance the discussion. All you're doing is steamrollering over legitimate opinions with an ad hominem attack.
For example, while I don't mind the GPL, I prefer to use more permissive licenses for my own code, because that's what's more in line with my personal ethics. I'm just not a "strings attached" kind of person. There is a courage criticism to make of how I do things, but it works in the opposite of the direction you're trying to take things: I sometimes wish I had the guts to follow SQLite's lead on software licensing.
> All you're doing is steamrollering over legitimate opinions with an ad hominem attack.
Where?
> This kind of comment does little to advance the discussion.
What are you talking about? Wider GPL adoption is central to the discussion!
Companies are taking permissively licensed work, making billions, not contributing back, and building massive platforms we'll all be forced to build upon.
AWS and the like are a step forward in terms of not needing to image and upgrade machines, but it's ten steps back in terms of computing freedom.
The future of computing is closed and they're using open source to corral us in.
> Wider GPL adoption is central to the discussion!
Sure. And so the question is, did the GPLv3 improve or degrade adoption of copyleft licensing?
To quote from Star Wars, "The more you tighten your grip, Tarkin, the more star systems will slip through your fingers."
Like it or not, companies fund the vast majority of open source software development, not hobbists, and when the FSF tried to "tighten its grip" by making a license which moved the balance of benefits more towards free software users, and away from the companies which funded the work, things didn't go well for them. And so what the FSF has found is that the the goals of the free software manifesto has been, more and more, slipping through its fingers.
Unless you can make non-free software illegal across all countries, at the end of the day, you need to find a balance where all stakeholders are comfortable with the result so that users, developers, and company freely and happily choose to use free software. This is a negotiation, and if there is more value that can be gained by signing onto the free software ecosystem, then perhaps the balance can be shifted a bit more towards the free software goals. But until you can, given that most open source developers do like food with their meals, a successful open source strategist has to be grounded in the economic and business realities. Just raising your right hand in a fist and shouting, "power to the people" and "freedom" is generally not going to get you very far.
I don't think your reference shows that. The GPLv3 was released in 2007. It obviously depends a lot on the language, since each language tends to have its own ecosystem, but the massive shift to permissive licenses really didn't start until 2008-2010, which is after the GPLv3 was promulgated. C didn't tip over to being majority permissive until 2011. So I'd argue that the analysis in the graph tends to support my contention that the GPLv3 hurt rather than helped the copyleft cause.
Post 2005 is when cloud computing & the web as an app platform began to gain traction, and a lot of open source software from ‘big tech’ is either cloud or web related.
The adoption of more permissive licenses is just because the big players favour those licences.
>For example, while I don't mind the GPL, I prefer to use more permissive licenses for my own code
This argument does little to advance the discussion either.
Some projects take a weekend to build. Others are the result of years of blood, sweat and tears.
The latter is probably where you're going to want to have a little more control over how the fruits of your labor are used and maybe get yourself paid a bit more than $0.
Not so much an attack, but my reaction was something like the post you respond to: "GPL3 doesn't work because people are abandoning GPL 3" seems like a self-reinforcing prophecy.
I am OK with people using whatever license they want for software they write. But if you think GPL 3 is ideal, by using something else you are just killing your ideal.
To me, it's not a self-fulfilling prophecy so much as a hard lesson in pragmatism. Some people care very deeply about Free-as-in-FSF. Others like the idea in vague terms, but don't assign much moral weight to it. I suspect that there's a tendency for the former group to project their feelings upon members of the latter. And that that, in turn, leads to some bewilderment when someone makes a blithely practical choice such as, "Among this range of licenses, all of which are perfectly acceptable to me, I'm going to just pick the one that I think will result in more people using the thing I made," or, "I'm going to just pick the one that's easiest for me to understand."
If you don't like "strings attached" then it certainly makes no sense to use a so-called "permissive" license, as the core--in fact, only--permission that such licenses offer over the various copyleft licenses is the permission to take all the other permissions away from downstream users... the people we should all be fighting for! "The only string attached to usage of this software is that you are not allowed to attach additional strings to this software."
I just can't bring myself to see it that way. If someone incorporates code I've open sourced into a commercial product, that takes nothing away from me or anyone else. The only permission I ever gave was for others to use some code I wrote for their own purposes. That doesn't change if someone incorporates it into a closed source product. It's not like someone else choosing not to open source their additions causes the original software license on the original code to disappear in a puff of smoke.
Why? These users overall aren't even willing to pay token money to support the creator so why should the creator care about them more than their own goals and desires?
Or is it perhaps because users don't want to use it unless they have no choice (in terms of software alternatives released under other OSS licenses)?
20 years ago GPL was the only thing most users new to OSS knew about. Since then they have learned there are better choices Meanwhile GPLv3 was the final nail in the coffin of the bossy OSS licenses.
One thing I don't get is why the OP proclaims love for GPLv3. It's the worst GPL license out there. I avoid it as both a contributor and user.
Corporate fangs win at the end. It is social pressure for monopoly autocracy and corporate practices. You should have had the freedom to choose whatever license you had in mind without being harassed by company employees or people who misunderstand how to use a GPL licensed product. Dual licensing would be better than giving up the reasons why you chose GPLv3.
Corporations can use GPLv3 in their codebases, they r just not bothered following the lawyer approvals so that code is shared back. If you cared about that, perhaps a better option is dual licensing.
I hope that the project does not become...yet another free labour camp by developers for big tech money bags.
Note the position of the hardest GPL defenders in this Github thread was that:
a) You need to have absolute 100% of the code ownership/copyright in order to change the project license. This is not true and there have been precedents where a valid and successful relicense was done with much less (if I remember correctly I learnt this from a comment somewhere hidden in that same Github issue)
b) The other project contributors would not necessarily agree with the relicense. Which was not the case: the most prominent contributors showed up and explicitly gave the author ownership (or free license) over their small parts of the code.
In our experience, all of the contributors were happy to give permission to re-license the code. They generally believed in the project (that's why they contributed) and wanted TerminusDB to be as successful as possible - if the core team felt that was going to happen by moving to Apache, they weren't going to stand in the way.
I suppose you could have somebody quite ideologically attached to the GNU/GPL who could then hold up a change. If the code was core and you couldn't easily re-implement, could make the price high.
We had the same experience with the Linux kernel. The original license was even more hard-line that the GPL --- it ruled out any way of making money, even selling CD-ROMs with Linux on it. Linus had to consult with all of the contributors to relax the restrictions to GPLv2, and we all agreed that it was in the best interests of the community. So there was no problems in terms of getting consent from all of the developers to make that license change to something more permissive than the original license.
At the time, the failure modes of the more permissive licensing model (for example, of NetApp's innovations for things like WAFL were never contributed back to BSD) were more foremost in our mind, and so for us the GPLv2 was the right compromise between the rights of the community and the companies using, and improving, the code. Of course, the decision that we made in 1992 was in a world very different from 2020, and so hopefully the choice of the Apache license will be the best one for TerminusDB. I have only the best wishes for you and your community!
Especially relevant to GPL, it ends up being counterproductive, as the license is a hardcore protection of the users' rights, at the expense of the project owner's rights. See the debate caused by Octotree [1], [2].
It's a pity because the way I see it, GPL is the ideal way to go. But if you need to pay the bills with software you write, it is also one of the worst open source choices out there.
(A)GPL is then relegated to either 100% vocational non-commercial code that one writes as a side project, or to big entities that for some reason write GPL code that is not part of their main business.
[0]: http://slides.com/adamretter/are-we-still-open-source
[1]: https://news.ycombinator.com/item?id=24953821
[2]: https://github.com/christianbundy/octotree/issues/1