TLDR: Akka is now source available, Alex is mad. Insofar as I can find a moral argument it’s that this represents a bait and switch. Alex also goes over the practical reasons devs like open source, which most significantly is avoiding bureaucratic processes.
I think this is right - any sustainable alternative to open source has to preserve that property. (My hunch is that a collecting society type model based on turnover is the way to do this). Alex mentions commoditization of ones complements. This points to open source being supported by businesses that specifically create it as a complement to their business; I think that’s fine but can only support a certain slice of software.
Agree. Data storage has had a reasonable path for monetizing open source by making it REALLY EASY to pay money, e.g., a cloud service that maintains my MongoDB cluster for me. That cuts a lot of corners for enterprises. If you want your open source project to churn out cash, make it easy to spend money on. There’s lots of ways to do that. Forceful license changes ignore what the customer wants, which is a recipe for a failing business.
I think what they’re aiming towards is something like what people like to do with physical things:
Buy them once
change, sell, share or modify as you want
distribute blueprints of their reverse engineering to maybe replace parts of it, or build something compatible
Problem is that this is software, which has no physical, single-quantity and changes a lot, so you would have to re-buy it (or pay monthly fees), to always have an “up to date purchase”. And the physical-object industry also tries to restrict such use cases (personal use only, no modifications, subscription based add-ons..)
Ultimately this is the GraalVM debate all over again: Do I want to settle my product runtime and core on something that may just go away at any point? Everyone that had to interact with the company-wide legacy ERP system, which they never managed to replace, knows the fear.
What I mean is that multiple open source projects would band together to basically commonly issue commercial licenses, and if a company buys in a licence the revenue gets allocated between projects. The more projects get under a given umbrella, the easier it is for companies to adopt them.
One project could be with multiple societies.
In terms of pricing I was thinking at the most basic level something linked to turn over and then something less convenient for companies that want to try to save money.
No, it’s very different. For a start tidelift doesn’t provide a better license. Tidelift is more like a consulting and support shop that kicks some money back to the free software projects.
Thanks for the summary! It’s exciting that Akka is doing this. With luck it’ll kill the whole misbegotten mess. The actor model on the JVM never made any sense.
My employer (Couchbase) did this too, about a year ago. Very sad. But I can’t refute the business justification they and other companies like Mongo give for it: loss of customers due to big cloud providers, mostly Amazon, forking the software and providing their own rebranded version.
I’m curious whether this is working for them. I’m opposed to these types of licenses but I wonder, in practice, how they’re working out. My hunch is that it’s not the license that’s the problem, it’s other aspects of the business model that are the problem. But, that’s not data - so I’d be interested in the outcomes.
Part of the problem IMO is that single-vendor / single-product FOSS is a bad idea to start. That is - when you have a company like Red Hat* that brings together a bunch of different FOSS and contributes to a bunch of FOSS, it adds enough value that they can charge without closing the source. If your product is a single project that’s FOSS it’s hard to provide enough value that running the open source version isn’t more attractive. So, they start tinkering with the license to force a relationship. That’s bad. It’s anti-FOSS, even when the license is FOSS.
IMO the ideal states for everybody’s benefit is vendor-neutral upstreams where the vendors bring real value above and beyond the code and bits in the products they ship and/or services they offer.
If Vendor X’s entire offering is around one codebase, that’s less interesting to me. The benefits of open source don’t really come into play because it’s always an uneven playing field. And it’s almost inevitable that as the project matures people will decide it’s easier to run the FOSS version and not pay, thereby driving the vendor to experiment with fauxpen source licenses - especially when VCs start wondering where the money is.
Former Red Hat employee here, but I felt this way before I worked there and feel that way after working there, so.
The things is that there is nothing wrong with cloud providers doing that. The licenses allow it. Even the GPL allows this.
For foundation-operated open source projects - like the apache software foundation this seems to work great. For single vendor OSS projects, not so much.
There is truth in that. What is also true is that the authors chose the original license. They picked it and should have known what it allowed and forbade.
People seem to still think “oh, they will all be nice people and send me money, because I spent so much time on it.” Nobody does that. Nobody cares about a software that works and they got for free. It is not on peoples minds.
Lightbend did choose a FOSS license. It didn’t work for them so they chose another license. They put forth a reasonable argument for it even though they didn’t have to.
OP, however, took offence to that: “how dare you put on my mind thinking about sending money to you?!”.
You make it sound so innocent. It just is not on peoples minds. Well, maybe it should be?
You’re not saying it but it seem like you’re implying that picking a non-FOSS license would’ve been better. And that might be true. This is only the most resent example of switching from a FOSS license calling forth undue toxicity form some people. Maybe it is prudent to start with a non-FOSS license to better manage expectations?
It seems like Source Available license with a generous free usage would signal “I believe it’s valuable I just don’t want to deal with payments at this time” at the beginning of the project. Later it has a clear path to sustainable business around that value. And if it didn’t pan out it can switch to a FOSS license to signal “AS IS. Don’t bother me”. It might be a solution because the other way around ruffles too many feathers, apparently.
OP, however, took offence to that: “how dare you put on my mind thinking about sending money to you?!”.
I am not commenting on OP, just on the licenses.
You make it sound so innocent. It just is not on peoples minds. Well, maybe it should be?
It really is innocent I think. Annoying, yet understandable. People are busy esp. at work. I think the majority here uses tons of tools and libraries free of charge w/o ever even thinking about it. Have you ever given all the projects you rely upon the money the “deserve”? I sometimes donate to projects, but not really the amount of value I extract from their work.
You’re not saying it but it seem like you’re implying that picking a non-FOSS license would’ve been better.
I am not really saying that. I once worked for a company that made an OSS framework. We had tons of succesful users, but since it mostly just worked, nobody had it on their mind to give us a slice of the pie. Now with a different license, that may have been possible, but that license would have hindered adoption in the first place. Permissive licences give you adoption/users, but they do not convert to paying customers. It’s complicated.
Lightbend did choose a FOSS license. It didn’t work for them so they chose another license.
Yes, after years of community- and reputation-building on the basis of being an open source software provider. People trust the reputation of open source software because they know they can get take the source code and pay someone else to maintain it for them if the original vendor doesn’t do it to their liking. It destroys the monopoly of vendors over software and fosters competition.
Perhaps Akka should have chosen a strong copyleft license from the very beginning since Apache ended up so unsuited for them?
People trust the reputation of open source software because they know they can get take the source code and pay someone else to maintain it for them if the original vendor doesn’t do it to their liking. It destroys the monopoly of vendors over software and fosters competition.
This is a hypothetical. Forks do happen but “pay” is rarely involved. What happens more often is that payments go to service providers who explicitly do not maintain the source code. This is the core of the issue in this case and in many similar.
Perhaps Akka should have chosen a strong copyleft license from the very beginning since Apache ended up so unsuited for them?
Hindsight 20/20, eh? Things change. Perhaps earlier Apache or whatever FOSS license they used was perfectly aligned with their goals. Now it doesn’t.
In the same vein we can say that perhaps users should have supported Lightbend financially more from the very beginning since they don’t like proprietary licenses so much?
But they didn’t. And that’s OK. What is not OK is being cross about it now and being angry at Lightbend but not themselves. This is not the first time this happens. The cause is clear. The problem is well known. There’s an obvious solution, too: pay the maintainers.
The correct way to react to this is admit you were wrong. “Well, it was great while it lasted. We had all the Freedom in the world and we chose to exploit the maintainers beyond the point they could bear it. This is not the first time it happens so I should probably learn the lesson and start paying to keep the Freedom.” Unfortunately, it doesn’t seem to go that way. Instead we get “My Freedoms are getting violated again! I can’t exploit any more so I will make a lot of fuss about it and make it look like I don’t use the software in protest. Anyway, who’s next?”.
What is not OK is being cross about it now and being angry at Lightbend but not themselves…The correct way to react to this is admit you were wrong.
By ‘you’ do you mean me, the guy who suggested using a strong copyleft license from the beginning so that Lightbend wouldn’t have found themselves in the situation in the first place of not being able to extract value out of their work? To me a perfectly reasonable path would have been to start with the GNU GPL v3 and then later switch to AGPL once they started offering value-added services. It’s the switching to a non-Free license that I take issue with.
By “you” I meant OP, or whoever might feel angry they can’t use Akka now because they have to pay for it now.
As for AGPL, I don’t think it’s a solution. It doesn’t prevent cloud providers from hosting a product or charge for it. It only prevents them from having a competitive advantage in a form of code changes. The best we can get out of it is those cloud providers contributing their changes. I don’t believe many of them actually have them.
AGPL doesn’t force those cloud providers to pay for the software. So they still can provide their services without spending on the development. And the developers still incur all the expenses.
At the same time the narrative goes that the developers still can charge for their services of hosting the very same software they develop. Except, they have to branch out into services, ops, reliability, support, and many other things that are not the actual development.
It’s been a long time since it became obvious that FOSS is not compatible with our reality. Be it capitalist economy, human nature, moral failing, or whatever. Every time it takes impressive mental gymnastics to show that it doesn’t exploit developers. People keep pointing at Linux and Red Hat to try and convince that it’s possible to not starve working on FOSS. At the same time if OpenSSL vanished without a trace more than one country would fall in economic collapse. No one seem to notice that until just a few years ago it was maintained by like two dudes making $30k/y. Heartbleed changed trajectory for OpenSSL but not for FOSS in general. A project has to be very popular or become “critical infrastructure” to attract any money in FOSS.
Look, it’s clear you don’t understand what Akka is (or you’re being deliberately obtuse to win an argument), otherwise you would never claim that cloud providers could ‘host’ Akka and charge for it. I think it’s time for you to call it a day.
If only we had a license that required Big Companies to contribute their changes back if they fork a project and provide a hosted, rebranded version of it (even if they do not distribute the fork itself)…
The problem is not generally big companies failing to contribute their changes back. That implies that big companies are making changes (which costs money) but you can make more money if you don’t write any code. For a lot of SaaS-style things, you can take an F/OSS project and provide people with a one-click deployment service and charge them for hosting. You can undercut the company that’s writing the software, because they have to pay software engineers.
The root problem is that we still don’t have a good business model for monetising zero marginal cost goods. It costs money to produce a piece of software but it costs nothing (or, as close to nothing as makes no difference) to produce a copy of that software. All of our current models evolved from goods with high marginal costs: you may have needed to spend some R&D money to design a widget, but building each widget also cost a lot and could amortise the cost, setting up a competing widget-production operation was a large investment. With software (or music, or films), it’s trivial for someone who has a copy of the artefact to sell (or give away) millions of copies of it.
As a result, we’ve ended up in a situation where we do the difficult thing (writing software) for free and try to make money from the easy thing (copying or hosting the software). I like to use the following analogy to explain this: Imagine if Henry Ford had decided, rather than providing only black cars, to give cars away for free but charge people 150% of the cost of building a car for painting them. Then had lobbied the government to pass laws that prevented unpainted cars from being allowed on the road. Then, when other companies popped up offering to paint cars for a fraction of the cost (or when DIY enthusiasts started painting their own cars) to make it illegal to use third-party paint to paint a car. After 50 years, it’s so normalised that the idea that you’d pay for a car is almost impossible for people to comprehend and if you suggest reducing the restrictions on who can paint a car then you’re told that you’d destroy the economy and put car companies out of business overnight. At some point, you need to step back and say ‘this is an unsustainable model for making money out of building cars’ and stop propping that business model up with legal hacks.
The root problem is that we still don’t have a good business model for monetising zero marginal cost goods.
Does proprietary licensing not work well for that? It worked well enough that Bill Gates could become the richest person in the world for a while.
IMO the root problem is that we software developers have gotten absolutely spoiled by open source, and now we expect too much for free, even when we’re using it to develop commercial software. There will always be good reasons for releasing some software under permissive open-source licenses, but we should accept both strong copyleft (e.g. AGPL) and non-OSD-compliant source-available licenses as permissible options for developers that need to make money from the software itself (and deserve to, because good work should be paid for).
Does proprietary licensing not work well for that? It worked well enough that Bill Gates could become the richest person in the world for a while.
It works well during the growth phase, not during the maintenance phase. It was easy to sell the first versions of Windows. It’s increasingly hard to sell new versions of Windows / Office, but it’s also hard to fund the maintenance of a project that big with only the sales to new customers. Maintenance contracts work for big enterprise software but getting individuals to buy a maintenance contract is hard.
The core problem that proprietary software encounters is that it ends up competing with pirates, who copy the software but don’t develop it and so have far lower costs. You can add DRM, but that just makes the experience worse for your customers than for the pirates’ customers. You can pass draconian laws that enforce harsh penalties on the pirates, but now you have a lot of cost for enforcement and you need to convince every jurisdiction to enforce them, even when it’s against their economic interests to do so.
IMO the root problem is that we software developers have gotten absolutely spoiled by open source, and now we expect too much for free, even when we’re using it to develop commercial software
In theory, F/OSS respects the economic reality: a lot of people are paid to write F/OSS but not paid for the copies of it that are made. It is hard to scale that up to a company supporting F/OSS developers without a different revenue stream though.
I’m not sure this is the riposte you think it is, because the AGPL has become somewhat popular for this use case, but not due to copyleft.
The root issue here is a project taking significant monetary investment and promising a profitable return by building some kind of business around the software. Sooner or later they have to try to prevent others from competing with them, because nobody else has the overhead of also having to be the primary maintainer of the software, and many other people may well be better at running a software business.
So they switch to a “source available” or to AGPL or to some other license that lets them make mouth noises about openness and freedom while using it to enforce a de jure (“source available”) or de facto (AGPL) monopoly on commercial exploitation of the software in hopes of returning a profit to their investors. Which is ironically at odds with the ideals of Free Software.
enforce a de jure (“source available”) or de facto (AGPL) monopoly on commercial exploitation of the software
So you’re saying AGPL would be a perfect fit, because they’d remain FLOSS, and still de facto the only entity to commercially exploit the software?
Which is ironically at odds with the ideals of Free Software.
FYI, there have been a fair number of examples where a fork of a GPL licensed project became more popular than the original, and even cases where it literally supplanted upstream. GCC & EGCS, for one. There’s also MariaDB, which - from what I can tell - pretty much took over from MySQL, and unlike what the Lightbend post claims, it is still GPL + LGPL, with BSL for additional stuff (ie, it’s open core, unlike Akka). There’s Joomla (a fork of Mambo), where the fork supplanted the original within ~3 years. X.org from XFree86, the latter dead since 2011. Just to name a few cases.
So being *GPL licensed does not necessarily mean one will remain the de facto monopoly over it, see the Mambo->Joomla case, or MySQL->MariaDB, or EGCS & GCC.
So you’re saying AGPL would be a perfect fit, because they’d remain FLOSS, and still de facto the only entity to commercially exploit the software?
I’m saying that when people do pick AGPL for this, it’s not to encourage freedom or sharing – it’s so that they become the only entity that can make and monetize proprietary add-ons or derivatives, because that’s basically always the underlying business model. And that this is, to put it charitably, not a good outcome for Free Software ideals.
And to pick on your supposed counterexamples, consider MySQL: the original business model for MySQL had some traditional elements of support contracts and training/setup, but really was a pay-for-proprietary scheme where you could buy copies of MySQL under non-GPL license terms. Which… again, is not really what I think of as the ideals of Free Software, since they were literally hoping to profit from increasing the amount of proprietary software in the world.
I’m saying that when people do pick AGPL for this, it’s not to encourage freedom or sharing – it’s so that they become the only entity that can make and monetize proprietary add-ons or derivatives, because that’s basically always the underlying business model
Uhh. No. The AGPL doesn’t do that. Dual licensing and copyright assignment lets you do that. The AGPL in and of itself does not. People pick the AGPL to discourage companies from keeping their fork private, and only sell the hosted service - and thus, the AGPL encourages sharing. Nothing to do with being a monopoly. Case in point: the Bitwarden server code is under the AGPL. Yet, they’re not the only entity making a Bitwarden server, nor the only one being financially supported. Mastodon is AGPL too, has a fair number of forks, and some of the features from the forks even trickled back upstream, while the same forks continue to receive various levels of funding (though, likely not as much as Mastodon itself, but none of them are in the business of supplanting upstream anyway - they’re about sharing, shocking).
My comment was about Lightbend’s opinion that the risk of other players not contributing back was a significant factor of them choosing the BSL. That is something the AGPL helps with, while still being a FLOSS license. But AGPL is a license Big Tech is allergic to, so using that would pose a bigger financial risk than being open source in the first place. That’s pretty telling, about how they view open source and their community.
People pick the AGPL to discourage companies from keeping their fork private, and only sell the hosted service - and thus, the AGPL encourages sharing. Nothing to do with being a monopoly.
Everything to do with being a monopoly, and if you can’t see that I don’t think we can have a productive discussion. The whole reason why these “open-source-as-a-startup” projects suddenly switch licenses is to maintain their own monopoly power as the only entity that can successfully run a SaaS based on the software, because all their competitors are forced to give away their in-house modifications while the original folks’ company doesn’t have to. It doesn’t matter what people hoped for when drafting the AGPL – that is its de facto effect, and like the dual-license schemes which came before it with GPL, the net effect is to increase the amount of commercially-exploited proprietary software in the world.
The whole reason why these “open-source-as-a-startup” projects suddenly switch licenses is to maintain their own monopoly power
Again, nothing to do with the license. You can switch from any license to any other as long as you’re the sole copyright holder, which is what these open-source-as-a-startup companies do.
What has to do with the license, is that the reason they list for going from APL-2 to BSL is partly due to the risk of competitors not sharing. Which would be a far smaller risk with AGPL. This I find disingenuous. We know it’s not the reason, because then they’d consider the AGPL. But that’s not what they communicate.
What they communicated in this case is “we love open source, but there’s too much of a risk of a competitor not contributing back, so we’re going proprietary”, rather than “we want more money”.
I don’t think we’re ever going to see eye-to-eye on this, so I’ll just state it plainly:
I do not think the reason for relicensing by these open-source-startup companies/projects is a fear that competitors won’t contribute back. I think the reason for relicensing is “competitors exist and are out-competing us, putting at risk the profits our investors expect”. And I think the track record of such relicensings makes this plain.
I do not think the reason for relicensing by these open-source-startup companies/projects is a fear that competitors won’t contribute back.
I agree on that. My jab wasn’t about that. My jab was at the disingenious way the reasons were presented. They presented that fear as a contributing reason, while it clearly wasn’t.
Maybe I’m picking nits, but my understanding of what the AGPL does is a little different, and that concerns me because I’m 99% sure you know more about this than I do. As I understand it, the AGPL logically takes the GPL and extends the meaning of “distribution” to include exposing the software as a network service. So it says that exposing the software as a service triggers the obligation to distribute the source to the software.
I don’t think it allows anyone to make proprietary add-ons or derivatives of a community project, unless that project also requires copyright assignment or something similarly obnoxious, which the OP also is railing against. The copyright holder is, naturally, free to also offer their software under a different (even proprietary) license in tandem with offering it under the AGPL.
But offering it under a dual AGPL/proprietary license seems like a better outcome for Free Software than offering it under a dual Less Proprietary (BSL)/proprietary license, to my eye.
I hope you’re interested enough to correct where I’m misunderstanding, because I’m pretty sure you understand this better than I do.
Start a project, release it under a permissive license, require some form of assignment or at least CLA with permission to relicense as a condition of others contributing to it, so that you will always be able to relicense in the future.
Take a bunch of venture capital money and start a company that offers the software as a hosted service or whatever.
When someone else inevitably starts out-competing you, exercise your right to relicense to AGPL, and now your competitors have to either fork you and maintain their fork, or go out of business, or adopt the AGPL’d version and release all their in-house stuff to the world for free, while you don’t have to do the same.
That’s why the AGPL is a popular choice for this. The original VC-backed startup/project can continue to build a proprietary software business on top of the original software. Nobody else can. Thus: monopoly on commercial exploitation, which is what the original project’s investors want.
require some form of assignment or at least CLA with permission to relicense as a condition of others contributing to it, so that you will always be able to relicense in the future.
OK. So I wasn’t missing anything. The bad behavior I was thinking of requires an assignment or the right kind of CLA to do this with community-authored code.
now your competitors have to […], or adopt the AGPL’d version and release all their in-house stuff to the world for free, while you don’t have to do the same.
Forcing other people who offer your software as a service to also publish their source code is the exact point of the AGPL. And while initially, you don’t have to do the same, you do if these competitors are making and publishing worthwhile changes without signing your CLA/assignment, and you choose use their stuff.
If they’re not making worthwhile changes that you’d want to use, it’s hard to argue that they’re outcompeting you.
I think even given your recipe, AGPL is more of a free software win than BSL or similar. I now feel more confident than before that it’s certainly the case if contributors refuse assignments and bad CLAs.
If they’re not making worthwhile changes that you’d want to use, it’s hard to argue that they’re outcompeting you.
If their service is more reliable, or easier to use, or cheaper, or any of a multitude of other factors, it will easily outcompete the original. Most of these projects turn out not to be very good at running a SaaS, and get outcompeted by people who got good at running SaaS first and then went looking for more software they could offer as a service.
And the whole idea of going AGPL is to try to make it unpalatable for such entities to compete, thus preserving the commercial monopoly. It’s actually very hard to offer a modern platform-as-a-service setup involving AGPL’d software to which you don’t own the full rights, and that’s the point.
And the whole idea of going AGPL is to try to make it unpalatable for such entities to compete,
Correction–to compete using proprietary modifications to the upstream software. They are entirely free to compete on the open source playing field, just like many other companies do. And which (surprise!) turns out to be better for users who don’t need to get locked into vendors of proprietary software.
The AGPL is a win-win situation in almost every scenario and offers exactly the right incentives to the right market participants–more rights and freedoms to the original developers of the software for their risk and investment, level playing field of competition to every other software vendor that eliminates the fear of proprietary lockin by other vendors and thus barriers to entry, and finally merit-based competition that benefits consumers.
Correction–to compete using proprietary modifications to the upstream software.
Which is typically what the original “upstream” VC-backed startup has as their business model – they reserve to themselves the exclusive right to do that, enforced via the mechanism of copyright.
This is not a “win-win”. Incentivizing people to build proprietary-software VC-backed SaaS companies is not, I would hope, what the Free Software community considers a “win”.
With a hat tip to @algernon’s sentiment, which closely aligns with my own first reaction…
Have they (or others who have undertaken similar license changes) articulated why they prefer this route over releasing under the AGPL? I’ve started to prefer the AGPL for my web things and feel like it addresses most if not all of the concerns the BSL does, while being more comfortable (IMO) for potential contributors.
I would figure a couple things, primarily and secondarily etc: risk of getting other unrelated tools accidentally GPL-ified like it was some kind of contagious sludge; a percieved inability to sell privately modifiable implementations to customers; communism; confusing language around patents in v3; etc.
AGPL requires release of modification for hosted software but it does not impose any restrictions on who’s getting paid. So the main issue of cloud providers hosting AGPL software, capturing fee and not paying to the developer/maintainer remain unresolved.
I don’t know what Akka is. I don’t think it matters for the discussion of licensing.
Overall this is a very bad take. It has no moral core, nor legal basis, not even a good drama.
It restricts free usage to non-commercial purposes only. […] This violates Open Source’s rule 6 […] Or Free Software’s freedom zero
No, it does not. It uses different license. They’re not trying to present it as a FOSS license.
Akka, like other products that pulled a bait-and-switch, is popular because it was marketed as being Open Source / Free Software
I can see how some people can see it that way. We’ll get back to this shortly.
as developers such as myself would never touch proprietary libraries with a ten-foot pole. Such libraries for me simply don’t exist
It’s a great shame, dear OP, you feel that way. There’s a lot of great proprietary software out there. But this is probably irrelevant to the discussion.
I need control of whatever runs in my program, I need the ability to fix it myself, or to have other people fix it for me, people that may not be affiliated with the maker of those tools;
Lightbend actually address this:
Can Akka community members continue to contribute to the project?
Yes. This is a source available license that allows and encourages community involvement.
See? This is not an issue in any practical sense. You have a bug fix? Nice! You can contribute it back and let Lightbend maintain your code.
Software licenses are expensive, add up, and even in big companies that can afford it, going through the endless bureaucracy of having such expenses approved is freaking painful, which is why FOSS may be even more popular in corporations than it is in startups;
I don’t want to assume, but it starts looking like OP is being obtuse on purpose here. Of corse, corporations would love free stuff they don’t have to deal with. Of course, licenses can be expensive. Of course, approving expenses can be painful. But you know what? A corporation with “US $25m per annum” revenue probably can afford both the bureaucracy and expense for the business-critical software.
The new terms and pricing is one the more approachable in the industry.
Selling support or extra tooling in FOSS sometimes works, because it’s complementary — employees can introduce a FOSS library or tool, without any kind of expense approval from upper management, and then the contract for extra stuff can come later, after it has proven its value.
Well, this is exactly the proposition from Lightbend. You start completely free, then buy license with support and extra stuff.
I assume, there was some sort of paid support available before but for some reason there wasn’t much demand.
It’s morally wrong to make the product popular, by advertising it as Open Source / Free Software, and then doing a reversal later. Don’t get me wrong, I am sympathetic to the issue that Open Source contributors aren’t getting paid. But in the Java community nobody wants to pay licenses for libraries. If that model ever worked, it was in other ecosystems, such as that of .NET, and that model has been dying there as well. Turns out, trying to monetize software libraries (or other FOSS products) is a losing proposition.
Oh, here it goes. It’s whole lot to unpack.
Is it morally wrong to use someone’s work without paying them? Is it morally wrong to expect a proper level of support in a big community from a relatively small company without paying for it? Let’s not go there.
Let’s instead focus on “Turns out, trying to monetize software libraries (or other FOSS products) is a losing proposition.” Why is that? How can one wave around Freedoms and appeal to morality and completely miss how they’re rallied up about someone wanting to collect benefits of their work? How OP can be angry about not being able to have this very good thing someone else built for free forever? If this thing is so good why wouldn’t OP want to pay for it?
I don’t have true answers as I’m not an OP but I will hazard a guess that it’s just a casual case of hypocrisy.
Akka is a great project, and has been one of the major reasons for why people chose Scala, or even Java. We’ve been using it at work, with great success.
OP completely agrees that Akka is good and valuable and everyone is happy it exists.
However, going forward I can no longer recommend any kind of investment in it.
But the moment the author made any effort to capture the value they created OP noped out right that moment.
I understand and have known the struggle that FOSS developers and companies go through. FOSS is just not a good business model.
I just can’t understand how OP can expect software to be built? Who’s going to pay for it? FOSS is already made way under market rate. Does OP expect authors to starve? Do authors have to work a day job to earn money and then spend extra hours on FOSS? What would OP suggest?
Consider that before this license change I have recommended Akka, and I may have contributed to Akka in my free time, if I ever found the need for it. But after this change, I can no longer do so without getting paid 🤷♂️
At this point it ventures into satire Inception-land. OP never paid a cent to the authors, never even contributed to the project, yet is very upset that some big corporations might start paying for software that makes them money.
I do want to thank Lightbend, from the bottom of my heart ❤️, for all they have contributed. I always loved their work.
Just not enough to start paying for it.
I said it before, I’ll say it again. FOSS did its thing. It’s time to move on to more sustainable models.
It still has its place. A weekend project is perfectly fine to be released under MIT or GPL. But any sort of moderately (and up) popular project requires just so much effort that it can only be sustained by corporate sponsorship.
And even then it might be a corporation capturing free labor from community rather than a proper community effort of classic FOSS.
A weekend project is perfectly fine to be released under MIT or GPL. But any sort of moderately (and up) popular project requires just so much effort that it can only be sustained by corporate sponsorship.
Agreed on the need for sponsorship - but, why is that incompatible with MIT or GPL?
It is compatible but it is not a business model. It’s hard to forecast sales. It is much harder to forecast sponsorships.
Sponsorships do not scale well, either. It’s quite straight-forwards to sponsor an individual or a few. But assumption is that they’re doing everything a project needs: development, technical writing, support, community management, infrastructure, ops, security. In principle it’s possible to sponsor individuals that will cover broad spectrum of project needs but I have not seen it in practice often (ever?).
Sponsorships are hard from the incentives point of view, too. Where do sponsored people loyalties lay? Is it community? Is it their sponsor? What to do when there’s a conflict of interest (e.g. tension between the community desires and sponsor’s request)?
I don’t have much to say about your whole post, but “FOSS did its thing. It’s time to move on to more sustainable models” is a strange thing to say. The GPL has been sustained for over thirty years. Only profiteers are looking to move away from it to licenses that restrict the user. If you think software is primarily about profit, then maybe the GPL isn’t sustainable (though companies like WordPress and OpenSIPS make it work). If you think software is about fun, or tools to make our lives easier or better, or even just fairness - there is always incentive to release your software as FOSS.
Only profiteers are looking to move away from it to licenses that restrict the user.
I don’t believe that’s true. I also don’t think you’re convinced it’s true either. There are many projects out there with very small teams and disproportionally big user bases. Those user bases demand support. Not only in the sense that they have some problems with the projects they want to be resolved but even things like contributions require attention of the teams. The asymmetry completely clogs the teams bandwidth. A solution to that is to dedicate more time to the project. That time might come either from the teams “free” time: curving time from their sleep, their time with family or friends, their recreation, etc. Or their work time: cutting into their profits. Either situation is unsustainable. An obvious solution is to pay the team so that they could spend their work time on the project. But this happens very rarely in FOSS for whatever reason.
If you think software is about fun, or tools to make our lives easier or better, or even just fairness - there is always incentive to release your software as FOSS.
Well, I wholeheartedly agree. The thing is, supporting a FOSS project is less fun than it seems. Writing docs is less fun than writing code. Dealing with inconsiderate users at times very not fun. An accidental breaking change in a new release resulting in a torrent of complaints is not what brightens an average maintainer’s mood.
You see, everyone is happy to take the “free” part of FOSS but also most act as if “AS IS” part is not there. Fun, altruism, and ideology (fairness being in this category) are all good but they are very small parts of a project maintenance. What’s worse these parts, being internal, at best remain constant, and usually diminish with time. The other parts though, more or less scale with the user base.
This is the issue. That point where fun becomes not enough people start looking for a solution. Some do not take any action and burn out. Some, distance from the project. Others are looking for a way to make it sustainable. For example, Lighbend were providing paid support. Turns out their product was so good that the profit was not at the sustenance level.
We (users) are burning out brilliant people who come up with good ideas and are willing to spend their time implementing them without hesitation while becoming all indignant at any hint of supporting them. But no one will step in to do their work.
There a calls to fork Akka, for example. Maybe there will be a few forks. Though, I’m 99% confident, not a single one of them will last longer than 6 months.
FOSS has it’s role but it also has a glaring problem. People know about it. And yet people rally against any effort to fix it. They don’t just point out issues in proposed solutions, or suggest alternative solutions. They say “you can’t fix it, do not even try”.
TLDR: Akka is now source available, Alex is mad. Insofar as I can find a moral argument it’s that this represents a bait and switch. Alex also goes over the practical reasons devs like open source, which most significantly is avoiding bureaucratic processes.
I think this is right - any sustainable alternative to open source has to preserve that property. (My hunch is that a collecting society type model based on turnover is the way to do this). Alex mentions commoditization of ones complements. This points to open source being supported by businesses that specifically create it as a complement to their business; I think that’s fine but can only support a certain slice of software.
Agree. Data storage has had a reasonable path for monetizing open source by making it REALLY EASY to pay money, e.g., a cloud service that maintains my MongoDB cluster for me. That cuts a lot of corners for enterprises. If you want your open source project to churn out cash, make it easy to spend money on. There’s lots of ways to do that. Forceful license changes ignore what the customer wants, which is a recipe for a failing business.
I think what they’re aiming towards is something like what people like to do with physical things:
Problem is that this is software, which has no physical, single-quantity and changes a lot, so you would have to re-buy it (or pay monthly fees), to always have an “up to date purchase”. And the physical-object industry also tries to restrict such use cases (personal use only, no modifications, subscription based add-ons..)
Ultimately this is the GraalVM debate all over again: Do I want to settle my product runtime and core on something that may just go away at any point? Everyone that had to interact with the company-wide legacy ERP system, which they never managed to replace, knows the fear.
Can you elaborate on this? I’m pretty much guessing what this might mean.
What I mean is that multiple open source projects would band together to basically commonly issue commercial licenses, and if a company buys in a licence the revenue gets allocated between projects. The more projects get under a given umbrella, the easier it is for companies to adopt them.
One project could be with multiple societies.
In terms of pricing I was thinking at the most basic level something linked to turn over and then something less convenient for companies that want to try to save money.
Thanks. In theory, something like this approach could also be useful for the governance and supply chain issues that we’re worried about today.
How so?
That sounds like https://tidelift.com/
No, it’s very different. For a start tidelift doesn’t provide a better license. Tidelift is more like a consulting and support shop that kicks some money back to the free software projects.
Thanks for the summary! It’s exciting that Akka is doing this. With luck it’ll kill the whole misbegotten mess. The actor model on the JVM never made any sense.
My employer (Couchbase) did this too, about a year ago. Very sad. But I can’t refute the business justification they and other companies like Mongo give for it: loss of customers due to big cloud providers, mostly Amazon, forking the software and providing their own rebranded version.
I’m curious whether this is working for them. I’m opposed to these types of licenses but I wonder, in practice, how they’re working out. My hunch is that it’s not the license that’s the problem, it’s other aspects of the business model that are the problem. But, that’s not data - so I’d be interested in the outcomes.
Part of the problem IMO is that single-vendor / single-product FOSS is a bad idea to start. That is - when you have a company like Red Hat* that brings together a bunch of different FOSS and contributes to a bunch of FOSS, it adds enough value that they can charge without closing the source. If your product is a single project that’s FOSS it’s hard to provide enough value that running the open source version isn’t more attractive. So, they start tinkering with the license to force a relationship. That’s bad. It’s anti-FOSS, even when the license is FOSS.
IMO the ideal states for everybody’s benefit is vendor-neutral upstreams where the vendors bring real value above and beyond the code and bits in the products they ship and/or services they offer.
If Vendor X’s entire offering is around one codebase, that’s less interesting to me. The benefits of open source don’t really come into play because it’s always an uneven playing field. And it’s almost inevitable that as the project matures people will decide it’s easier to run the FOSS version and not pay, thereby driving the vendor to experiment with fauxpen source licenses - especially when VCs start wondering where the money is.
The things is that there is nothing wrong with cloud providers doing that. The licenses allow it. Even the GPL allows this.
For foundation-operated open source projects - like the apache software foundation this seems to work great. For single vendor OSS projects, not so much.
I’m so tired of this bitching. Why should everybody get rich with software except the people writing it?
There is truth in that. What is also true is that the authors chose the original license. They picked it and should have known what it allowed and forbade.
People seem to still think “oh, they will all be nice people and send me money, because I spent so much time on it.” Nobody does that. Nobody cares about a software that works and they got for free. It is not on peoples minds.
Smells a little like victim blaming.
Lightbend did choose a FOSS license. It didn’t work for them so they chose another license. They put forth a reasonable argument for it even though they didn’t have to.
OP, however, took offence to that: “how dare you put on my mind thinking about sending money to you?!”.
You make it sound so innocent. It just is not on peoples minds. Well, maybe it should be?
You’re not saying it but it seem like you’re implying that picking a non-FOSS license would’ve been better. And that might be true. This is only the most resent example of switching from a FOSS license calling forth undue toxicity form some people. Maybe it is prudent to start with a non-FOSS license to better manage expectations?
It seems like Source Available license with a generous free usage would signal “I believe it’s valuable I just don’t want to deal with payments at this time” at the beginning of the project. Later it has a clear path to sustainable business around that value. And if it didn’t pan out it can switch to a FOSS license to signal “AS IS. Don’t bother me”. It might be a solution because the other way around ruffles too many feathers, apparently.
I am not commenting on OP, just on the licenses.
It really is innocent I think. Annoying, yet understandable. People are busy esp. at work. I think the majority here uses tons of tools and libraries free of charge w/o ever even thinking about it. Have you ever given all the projects you rely upon the money the “deserve”? I sometimes donate to projects, but not really the amount of value I extract from their work.
I am not really saying that. I once worked for a company that made an OSS framework. We had tons of succesful users, but since it mostly just worked, nobody had it on their mind to give us a slice of the pie. Now with a different license, that may have been possible, but that license would have hindered adoption in the first place. Permissive licences give you adoption/users, but they do not convert to paying customers. It’s complicated.
Yes, after years of community- and reputation-building on the basis of being an open source software provider. People trust the reputation of open source software because they know they can get take the source code and pay someone else to maintain it for them if the original vendor doesn’t do it to their liking. It destroys the monopoly of vendors over software and fosters competition.
Perhaps Akka should have chosen a strong copyleft license from the very beginning since Apache ended up so unsuited for them?
This is a hypothetical. Forks do happen but “pay” is rarely involved. What happens more often is that payments go to service providers who explicitly do not maintain the source code. This is the core of the issue in this case and in many similar.
Hindsight 20/20, eh? Things change. Perhaps earlier Apache or whatever FOSS license they used was perfectly aligned with their goals. Now it doesn’t.
In the same vein we can say that perhaps users should have supported Lightbend financially more from the very beginning since they don’t like proprietary licenses so much?
But they didn’t. And that’s OK. What is not OK is being cross about it now and being angry at Lightbend but not themselves. This is not the first time this happens. The cause is clear. The problem is well known. There’s an obvious solution, too: pay the maintainers.
The correct way to react to this is admit you were wrong. “Well, it was great while it lasted. We had all the Freedom in the world and we chose to exploit the maintainers beyond the point they could bear it. This is not the first time it happens so I should probably learn the lesson and start paying to keep the Freedom.” Unfortunately, it doesn’t seem to go that way. Instead we get “My Freedoms are getting violated again! I can’t exploit any more so I will make a lot of fuss about it and make it look like I don’t use the software in protest. Anyway, who’s next?”.
By ‘you’ do you mean me, the guy who suggested using a strong copyleft license from the beginning so that Lightbend wouldn’t have found themselves in the situation in the first place of not being able to extract value out of their work? To me a perfectly reasonable path would have been to start with the GNU GPL v3 and then later switch to AGPL once they started offering value-added services. It’s the switching to a non-Free license that I take issue with.
By “you” I meant OP, or whoever might feel angry they can’t use Akka now because they have to pay for it now.
As for AGPL, I don’t think it’s a solution. It doesn’t prevent cloud providers from hosting a product or charge for it. It only prevents them from having a competitive advantage in a form of code changes. The best we can get out of it is those cloud providers contributing their changes. I don’t believe many of them actually have them.
AGPL doesn’t force those cloud providers to pay for the software. So they still can provide their services without spending on the development. And the developers still incur all the expenses.
At the same time the narrative goes that the developers still can charge for their services of hosting the very same software they develop. Except, they have to branch out into services, ops, reliability, support, and many other things that are not the actual development.
It’s been a long time since it became obvious that FOSS is not compatible with our reality. Be it capitalist economy, human nature, moral failing, or whatever. Every time it takes impressive mental gymnastics to show that it doesn’t exploit developers. People keep pointing at Linux and Red Hat to try and convince that it’s possible to not starve working on FOSS. At the same time if OpenSSL vanished without a trace more than one country would fall in economic collapse. No one seem to notice that until just a few years ago it was maintained by like two dudes making $30k/y. Heartbleed changed trajectory for OpenSSL but not for FOSS in general. A project has to be very popular or become “critical infrastructure” to attract any money in FOSS.
Look, it’s clear you don’t understand what Akka is (or you’re being deliberately obtuse to win an argument), otherwise you would never claim that cloud providers could ‘host’ Akka and charge for it. I think it’s time for you to call it a day.
If only we had a license that required Big Companies to contribute their changes back if they fork a project and provide a hosted, rebranded version of it (even if they do not distribute the fork itself)…
The problem is not generally big companies failing to contribute their changes back. That implies that big companies are making changes (which costs money) but you can make more money if you don’t write any code. For a lot of SaaS-style things, you can take an F/OSS project and provide people with a one-click deployment service and charge them for hosting. You can undercut the company that’s writing the software, because they have to pay software engineers.
The root problem is that we still don’t have a good business model for monetising zero marginal cost goods. It costs money to produce a piece of software but it costs nothing (or, as close to nothing as makes no difference) to produce a copy of that software. All of our current models evolved from goods with high marginal costs: you may have needed to spend some R&D money to design a widget, but building each widget also cost a lot and could amortise the cost, setting up a competing widget-production operation was a large investment. With software (or music, or films), it’s trivial for someone who has a copy of the artefact to sell (or give away) millions of copies of it.
As a result, we’ve ended up in a situation where we do the difficult thing (writing software) for free and try to make money from the easy thing (copying or hosting the software). I like to use the following analogy to explain this: Imagine if Henry Ford had decided, rather than providing only black cars, to give cars away for free but charge people 150% of the cost of building a car for painting them. Then had lobbied the government to pass laws that prevented unpainted cars from being allowed on the road. Then, when other companies popped up offering to paint cars for a fraction of the cost (or when DIY enthusiasts started painting their own cars) to make it illegal to use third-party paint to paint a car. After 50 years, it’s so normalised that the idea that you’d pay for a car is almost impossible for people to comprehend and if you suggest reducing the restrictions on who can paint a car then you’re told that you’d destroy the economy and put car companies out of business overnight. At some point, you need to step back and say ‘this is an unsustainable model for making money out of building cars’ and stop propping that business model up with legal hacks.
Does proprietary licensing not work well for that? It worked well enough that Bill Gates could become the richest person in the world for a while.
IMO the root problem is that we software developers have gotten absolutely spoiled by open source, and now we expect too much for free, even when we’re using it to develop commercial software. There will always be good reasons for releasing some software under permissive open-source licenses, but we should accept both strong copyleft (e.g. AGPL) and non-OSD-compliant source-available licenses as permissible options for developers that need to make money from the software itself (and deserve to, because good work should be paid for).
It works well during the growth phase, not during the maintenance phase. It was easy to sell the first versions of Windows. It’s increasingly hard to sell new versions of Windows / Office, but it’s also hard to fund the maintenance of a project that big with only the sales to new customers. Maintenance contracts work for big enterprise software but getting individuals to buy a maintenance contract is hard.
The core problem that proprietary software encounters is that it ends up competing with pirates, who copy the software but don’t develop it and so have far lower costs. You can add DRM, but that just makes the experience worse for your customers than for the pirates’ customers. You can pass draconian laws that enforce harsh penalties on the pirates, but now you have a lot of cost for enforcement and you need to convince every jurisdiction to enforce them, even when it’s against their economic interests to do so.
In theory, F/OSS respects the economic reality: a lot of people are paid to write F/OSS but not paid for the copies of it that are made. It is hard to scale that up to a company supporting F/OSS developers without a different revenue stream though.
I’m not sure this is the riposte you think it is, because the AGPL has become somewhat popular for this use case, but not due to copyleft.
The root issue here is a project taking significant monetary investment and promising a profitable return by building some kind of business around the software. Sooner or later they have to try to prevent others from competing with them, because nobody else has the overhead of also having to be the primary maintainer of the software, and many other people may well be better at running a software business.
So they switch to a “source available” or to AGPL or to some other license that lets them make mouth noises about openness and freedom while using it to enforce a de jure (“source available”) or de facto (AGPL) monopoly on commercial exploitation of the software in hopes of returning a profit to their investors. Which is ironically at odds with the ideals of Free Software.
So you’re saying AGPL would be a perfect fit, because they’d remain FLOSS, and still de facto the only entity to commercially exploit the software?
FYI, there have been a fair number of examples where a fork of a GPL licensed project became more popular than the original, and even cases where it literally supplanted upstream. GCC & EGCS, for one. There’s also MariaDB, which - from what I can tell - pretty much took over from MySQL, and unlike what the Lightbend post claims, it is still GPL + LGPL, with BSL for additional stuff (ie, it’s open core, unlike Akka). There’s Joomla (a fork of Mambo), where the fork supplanted the original within ~3 years. X.org from XFree86, the latter dead since 2011. Just to name a few cases.
So being *GPL licensed does not necessarily mean one will remain the de facto monopoly over it, see the Mambo->Joomla case, or MySQL->MariaDB, or EGCS & GCC.
I’m saying that when people do pick AGPL for this, it’s not to encourage freedom or sharing – it’s so that they become the only entity that can make and monetize proprietary add-ons or derivatives, because that’s basically always the underlying business model. And that this is, to put it charitably, not a good outcome for Free Software ideals.
And to pick on your supposed counterexamples, consider MySQL: the original business model for MySQL had some traditional elements of support contracts and training/setup, but really was a pay-for-proprietary scheme where you could buy copies of MySQL under non-GPL license terms. Which… again, is not really what I think of as the ideals of Free Software, since they were literally hoping to profit from increasing the amount of proprietary software in the world.
Uhh. No. The AGPL doesn’t do that. Dual licensing and copyright assignment lets you do that. The AGPL in and of itself does not. People pick the AGPL to discourage companies from keeping their fork private, and only sell the hosted service - and thus, the AGPL encourages sharing. Nothing to do with being a monopoly. Case in point: the Bitwarden server code is under the AGPL. Yet, they’re not the only entity making a Bitwarden server, nor the only one being financially supported. Mastodon is AGPL too, has a fair number of forks, and some of the features from the forks even trickled back upstream, while the same forks continue to receive various levels of funding (though, likely not as much as Mastodon itself, but none of them are in the business of supplanting upstream anyway - they’re about sharing, shocking).
My comment was about Lightbend’s opinion that the risk of other players not contributing back was a significant factor of them choosing the BSL. That is something the AGPL helps with, while still being a FLOSS license. But AGPL is a license Big Tech is allergic to, so using that would pose a bigger financial risk than being open source in the first place. That’s pretty telling, about how they view open source and their community.
Everything to do with being a monopoly, and if you can’t see that I don’t think we can have a productive discussion. The whole reason why these “open-source-as-a-startup” projects suddenly switch licenses is to maintain their own monopoly power as the only entity that can successfully run a SaaS based on the software, because all their competitors are forced to give away their in-house modifications while the original folks’ company doesn’t have to. It doesn’t matter what people hoped for when drafting the AGPL – that is its de facto effect, and like the dual-license schemes which came before it with GPL, the net effect is to increase the amount of commercially-exploited proprietary software in the world.
Again, nothing to do with the license. You can switch from any license to any other as long as you’re the sole copyright holder, which is what these open-source-as-a-startup companies do.
What has to do with the license, is that the reason they list for going from APL-2 to BSL is partly due to the risk of competitors not sharing. Which would be a far smaller risk with AGPL. This I find disingenuous. We know it’s not the reason, because then they’d consider the AGPL. But that’s not what they communicate.
What they communicated in this case is “we love open source, but there’s too much of a risk of a competitor not contributing back, so we’re going proprietary”, rather than “we want more money”.
I don’t think we’re ever going to see eye-to-eye on this, so I’ll just state it plainly:
I do not think the reason for relicensing by these open-source-startup companies/projects is a fear that competitors won’t contribute back. I think the reason for relicensing is “competitors exist and are out-competing us, putting at risk the profits our investors expect”. And I think the track record of such relicensings makes this plain.
I agree on that. My jab wasn’t about that. My jab was at the disingenious way the reasons were presented. They presented that fear as a contributing reason, while it clearly wasn’t.
Maybe I’m picking nits, but my understanding of what the AGPL does is a little different, and that concerns me because I’m 99% sure you know more about this than I do. As I understand it, the AGPL logically takes the GPL and extends the meaning of “distribution” to include exposing the software as a network service. So it says that exposing the software as a service triggers the obligation to distribute the source to the software.
I don’t think it allows anyone to make proprietary add-ons or derivatives of a community project, unless that project also requires copyright assignment or something similarly obnoxious, which the OP also is railing against. The copyright holder is, naturally, free to also offer their software under a different (even proprietary) license in tandem with offering it under the AGPL.
But offering it under a dual AGPL/proprietary license seems like a better outcome for Free Software than offering it under a dual Less Proprietary (BSL)/proprietary license, to my eye.
I hope you’re interested enough to correct where I’m misunderstanding, because I’m pretty sure you understand this better than I do.
The full recipe is:
That’s why the AGPL is a popular choice for this. The original VC-backed startup/project can continue to build a proprietary software business on top of the original software. Nobody else can. Thus: monopoly on commercial exploitation, which is what the original project’s investors want.
OK. So I wasn’t missing anything. The bad behavior I was thinking of requires an assignment or the right kind of CLA to do this with community-authored code.
Forcing other people who offer your software as a service to also publish their source code is the exact point of the AGPL. And while initially, you don’t have to do the same, you do if these competitors are making and publishing worthwhile changes without signing your CLA/assignment, and you choose use their stuff.
If they’re not making worthwhile changes that you’d want to use, it’s hard to argue that they’re outcompeting you.
I think even given your recipe, AGPL is more of a free software win than BSL or similar. I now feel more confident than before that it’s certainly the case if contributors refuse assignments and bad CLAs.
If their service is more reliable, or easier to use, or cheaper, or any of a multitude of other factors, it will easily outcompete the original. Most of these projects turn out not to be very good at running a SaaS, and get outcompeted by people who got good at running SaaS first and then went looking for more software they could offer as a service.
And the whole idea of going AGPL is to try to make it unpalatable for such entities to compete, thus preserving the commercial monopoly. It’s actually very hard to offer a modern platform-as-a-service setup involving AGPL’d software to which you don’t own the full rights, and that’s the point.
Correction–to compete using proprietary modifications to the upstream software. They are entirely free to compete on the open source playing field, just like many other companies do. And which (surprise!) turns out to be better for users who don’t need to get locked into vendors of proprietary software.
The AGPL is a win-win situation in almost every scenario and offers exactly the right incentives to the right market participants–more rights and freedoms to the original developers of the software for their risk and investment, level playing field of competition to every other software vendor that eliminates the fear of proprietary lockin by other vendors and thus barriers to entry, and finally merit-based competition that benefits consumers.
Which is typically what the original “upstream” VC-backed startup has as their business model – they reserve to themselves the exclusive right to do that, enforced via the mechanism of copyright.
This is not a “win-win”. Incentivizing people to build proprietary-software VC-backed SaaS companies is not, I would hope, what the Free Software community considers a “win”.
With a hat tip to @algernon’s sentiment, which closely aligns with my own first reaction…
Have they (or others who have undertaken similar license changes) articulated why they prefer this route over releasing under the AGPL? I’ve started to prefer the AGPL for my web things and feel like it addresses most if not all of the concerns the BSL does, while being more comfortable (IMO) for potential contributors.
I would figure a couple things, primarily and secondarily etc: risk of getting other unrelated tools accidentally GPL-ified like it was some kind of contagious sludge; a percieved inability to sell privately modifiable implementations to customers; communism; confusing language around patents in v3; etc.
AGPL doesn’t solve the financial part.
AGPL requires release of modification for hosted software but it does not impose any restrictions on who’s getting paid. So the main issue of cloud providers hosting AGPL software, capturing fee and not paying to the developer/maintainer remain unresolved.
BSL specifically addresses that part.
I don’t know what Akka is. I don’t think it matters for the discussion of licensing.
Overall this is a very bad take. It has no moral core, nor legal basis, not even a good drama.
No, it does not. It uses different license. They’re not trying to present it as a FOSS license.
I can see how some people can see it that way. We’ll get back to this shortly.
It’s a great shame, dear OP, you feel that way. There’s a lot of great proprietary software out there. But this is probably irrelevant to the discussion.
Lightbend actually address this:
See? This is not an issue in any practical sense. You have a bug fix? Nice! You can contribute it back and let Lightbend maintain your code.
I don’t want to assume, but it starts looking like OP is being obtuse on purpose here. Of corse, corporations would love free stuff they don’t have to deal with. Of course, licenses can be expensive. Of course, approving expenses can be painful. But you know what? A corporation with “US $25m per annum” revenue probably can afford both the bureaucracy and expense for the business-critical software.
The new terms and pricing is one the more approachable in the industry.
Well, this is exactly the proposition from Lightbend. You start completely free, then buy license with support and extra stuff.
I assume, there was some sort of paid support available before but for some reason there wasn’t much demand.
Oh, here it goes. It’s whole lot to unpack.
Is it morally wrong to use someone’s work without paying them? Is it morally wrong to expect a proper level of support in a big community from a relatively small company without paying for it? Let’s not go there.
Let’s instead focus on “Turns out, trying to monetize software libraries (or other FOSS products) is a losing proposition.” Why is that? How can one wave around Freedoms and appeal to morality and completely miss how they’re rallied up about someone wanting to collect benefits of their work? How OP can be angry about not being able to have this very good thing someone else built for free forever? If this thing is so good why wouldn’t OP want to pay for it?
I don’t have true answers as I’m not an OP but I will hazard a guess that it’s just a casual case of hypocrisy.
OP completely agrees that Akka is good and valuable and everyone is happy it exists.
But the moment the author made any effort to capture the value they created OP noped out right that moment.
I just can’t understand how OP can expect software to be built? Who’s going to pay for it? FOSS is already made way under market rate. Does OP expect authors to starve? Do authors have to work a day job to earn money and then spend extra hours on FOSS? What would OP suggest?
At this point it ventures into satire Inception-land. OP never paid a cent to the authors, never even contributed to the project, yet is very upset that some big corporations might start paying for software that makes them money.
Just not enough to start paying for it.
I said it before, I’ll say it again. FOSS did its thing. It’s time to move on to more sustainable models.
It still has its place. A weekend project is perfectly fine to be released under MIT or GPL. But any sort of moderately (and up) popular project requires just so much effort that it can only be sustained by corporate sponsorship.
And even then it might be a corporation capturing free labor from community rather than a proper community effort of classic FOSS.
Agreed on the need for sponsorship - but, why is that incompatible with MIT or GPL?
It is compatible but it is not a business model. It’s hard to forecast sales. It is much harder to forecast sponsorships.
Sponsorships do not scale well, either. It’s quite straight-forwards to sponsor an individual or a few. But assumption is that they’re doing everything a project needs: development, technical writing, support, community management, infrastructure, ops, security. In principle it’s possible to sponsor individuals that will cover broad spectrum of project needs but I have not seen it in practice often (ever?).
Sponsorships are hard from the incentives point of view, too. Where do sponsored people loyalties lay? Is it community? Is it their sponsor? What to do when there’s a conflict of interest (e.g. tension between the community desires and sponsor’s request)?
I don’t have much to say about your whole post, but “FOSS did its thing. It’s time to move on to more sustainable models” is a strange thing to say. The GPL has been sustained for over thirty years. Only profiteers are looking to move away from it to licenses that restrict the user. If you think software is primarily about profit, then maybe the GPL isn’t sustainable (though companies like WordPress and OpenSIPS make it work). If you think software is about fun, or tools to make our lives easier or better, or even just fairness - there is always incentive to release your software as FOSS.
I concede my wording was unclear.
I don’t believe that’s true. I also don’t think you’re convinced it’s true either. There are many projects out there with very small teams and disproportionally big user bases. Those user bases demand support. Not only in the sense that they have some problems with the projects they want to be resolved but even things like contributions require attention of the teams. The asymmetry completely clogs the teams bandwidth. A solution to that is to dedicate more time to the project. That time might come either from the teams “free” time: curving time from their sleep, their time with family or friends, their recreation, etc. Or their work time: cutting into their profits. Either situation is unsustainable. An obvious solution is to pay the team so that they could spend their work time on the project. But this happens very rarely in FOSS for whatever reason.
Well, I wholeheartedly agree. The thing is, supporting a FOSS project is less fun than it seems. Writing docs is less fun than writing code. Dealing with inconsiderate users at times very not fun. An accidental breaking change in a new release resulting in a torrent of complaints is not what brightens an average maintainer’s mood.
You see, everyone is happy to take the “free” part of FOSS but also most act as if “AS IS” part is not there. Fun, altruism, and ideology (fairness being in this category) are all good but they are very small parts of a project maintenance. What’s worse these parts, being internal, at best remain constant, and usually diminish with time. The other parts though, more or less scale with the user base.
This is the issue. That point where fun becomes not enough people start looking for a solution. Some do not take any action and burn out. Some, distance from the project. Others are looking for a way to make it sustainable. For example, Lighbend were providing paid support. Turns out their product was so good that the profit was not at the sustenance level.
We (users) are burning out brilliant people who come up with good ideas and are willing to spend their time implementing them without hesitation while becoming all indignant at any hint of supporting them. But no one will step in to do their work.
There a calls to fork Akka, for example. Maybe there will be a few forks. Though, I’m 99% confident, not a single one of them will last longer than 6 months.
FOSS has it’s role but it also has a glaring problem. People know about it. And yet people rally against any effort to fix it. They don’t just point out issues in proposed solutions, or suggest alternative solutions. They say “you can’t fix it, do not even try”.
Ah, they’re pulling an ElasticSearch? So what’s gonna be their OpenSearch (Amazon’s still-opensource ES fork)?
Well the bet is presumably that no-one will do this. The difference is that Amazon makes money hosting ES
For everyone who is angry about this and this sort of thing: would you feel different if the companies in question were worker co-operatives?