alt Hacker NewsIn the long run, GPL code becomes irrelevant (2015)
Comments
Just anecdotally, but this aligns with my observations on the trend/growth of successful useful open source projects that go with permissive BSD-like license. ~20 years ago there were way less of those than now.
And as a SW developer doing client side/apps as well, using GPL/LGPL is a total pain and basically cost prohibitive, unless I work on my personal small project where I don’t care about having to/risking to open source the rest of the code and getting sued/cloned…
Real life example from ~2010 - we ended up including an LGPL library in our mobile app code, and published/upstreamed all the modifications we did to that code (mostly ARM optimizations). Once the app became popular, our competitors came to us demanding the source code of our app - just because iOS didn’t support dynamic libraries (so we had to statically link it), and giving them the object code to relink it wasn’t enough for them (which would satisfy the spirit of LGPL), because they really wanted to see how we hacked around iOS camera input APIs…
In the very long run I'd hope we get sane intellectual property law: Software and logic circuitry is mathematics and should be unpatentable and trade secret is about the only protection for software/firmware/hardware; copyright terms shortened to the actual window of profitability (5-10 years) and only applying to "business logic". Nothing else makes much sense in the modern world; standards and interoperability benefit everyone to such an extent and change happens so rapidly that the majority of existing IP protection duration only harms historians.
I can think of at least a few counterexamples. The logic behind the hypothesis is decent and tells a plausible story, but I'd like to see a more robust analysis; has anyone done one since the article was first posted 10 years ago?
Noteable counterxamples (excluding e.g. Linux):
- Git (GPLv2)
- Blender (GPLv2 & 3, from the looks of it)
- Krita (GPLv3)
- MySQL (GPLv2) -- still seems very popular in 2025
- QGIS (GPLv2)
Licenses that allow more collaboration end up winning, and permissive licenses allow more collaboration.
That said, the source-availability requirement of the GPL did help kickstart things.
In the 80s every unix vendor had their own closed source version of everything. The GPL - through the requirement of source availability - helped flip the industry over to collaborative building. It would’ve never gone that way if those working in the open did it under permissive licensing - that would have given closed-source vendors free labor and discouraged those working on free software.
Nowadays it’s about velocity. In domains where collaborative building is producing more better software faster, it’s more expensive to work behind closed doors than to take others’ work for free. So you may as well use the open thing. And then when you want to add features or fix bugs, it’s a hassle to maintain your own fork while integrating new work from upstream. So you may as well upstream your work too!
It may be true, but the GPL is uniquely positioned to create a floor that ratchets upwards, whereas the other licenses are susceptible to being consumed and extended.
This whole argument hinges on the statement that the GPL "does nothing for users" but is "annoying for developers". I dispute both of these claims.
GPL is all about doing something for users. It is users who are able to request source code. My entire network setup is based on the fact that I can customize my router which I can only do because users were able to request the source for the router in order to customize it.
When it comes to "annoying for developers" we need to be clear. The GPL is annoying for developers of software that is not open source. It annoys them because it says they must either take the open source deal or else rewrite it themselves. Apple has famously used a lot of time and money to rewrite GPLd thengs. This is the goal.
OTOH open source developers need not be annoyed by any GPL dependency since they can always use it without any trouble to themselves.
>But think of the multi-trillion dollar companies!!
If companies want to use manpower to reimplement GPL code, then fine. It’s always funny how accommodating people are for these companies.
> With enough time, any sufficiently large company can implement their own version of any software that anyone else has written. They usually won't write their own version if a high quality opensource version exists with a permissible license.
Is this a true assertion? If you define sufficiently large company to be Google, Microsoft, Amazon, etc. then sure, of course they can. That's an extremely high bar though and I would bet that even then, these companies would have to pick and choose their battles.
Yeah, the Apache/BSD licensed code will always come out on top, when some big company eats it, and effectively dictates its development with no other fork gaining any traction and with the project effectively becoming enterprise-ified.
This happens to a lesser or greater extent, but it happens.
One of the interesting things about the experience of LLVM and Clang isn't that it's killing off gcc--a decade after this was written, gcc is still the compiler of choice for most Linux distrubtions--but that it's killing off EDG, the C++ frontend that all proprietary C++ compilers used to use. Killing off proprietary software because the open source stuff is so much better is the vision of Stallman, and it's telling that a permissive license has been much more effective than a copyleft license in doing so.
using BSD/MIT licences is like betting against black swan event
sure, "contributing is cheaper than maintaining a fork" is true most of the time - but the moment new Microsoft comes in with "embrace, extend, extinguish" (or just copy and change), you're doomed
and heck, we had that exact thing happen last autumn, iirc - making big news on this website
Great writeup! I couldn't agree more.
I've been thinking about this a lot recently, because GPL was meant to ensure that vendors couldn't take OSS, turn it into closed source, and use it to extinguish the OSS.
As JS-writing eng I live in an MIT-native offshoot of the OSS world and for us the ratchet that ensures we always get more and more free software is basically the fact that when your product is a script run in a scripting engine you can't ever truly hide anything.
Since we have an alternate ratchet that has proven that it works to increase the amount and quality of OSS (over a 20-year time period), the GPL does seem as you say: a relic of times when we it seemed like software might only be a hobby.
I'm writing a VCS kernel, basically, and its cost me the last 5 years of my life. My code is MIT. Do I have to think about the dangers of embrace-extend-extinguish? Yeah, but having the best product is a very strong defense, and building the widest coalition of supporters you can is how you get there.
I think this is largely an illusion. Most open-source software isn't that successful, so it might seem like the choice of license didn't really matter for their success. But you'll notice that all of the big, successful open-source projects are either GPL, or can't be GPL because of the GPLs murky legality around linking (the article mostly hinges on such a case - LLVM).
The author talks about GPL projects feeding back on themselves to create technological dominance. But it's much more than that. GPL encourages organizational dominance, too. It starts with watch dogs looking out for GPL violations. And it ends with a big nonprofit foundation providing training and paying developer's salaries. Why did Linux blast past BSD? The popular story is that some company was trying to claim ownership over BSD. But the same thing happened to Linux a decade or so later with SCO. I think the license created a no-win situation for anyone who wanted to create their own Unix-based OS. If it wasn't Linux-compatible, it wasn't valuable. And nobody could keep up with Linux's rapid pace of development. So everyone gave up and started contributing to Linux, causing the pace of development to increase even more. Now, Linux has what, a million commits per year? Something insane like that. Try achieving that with a BSD license.
Ignores the MPL as usual. IMO the MPL is better than BSD/MIT/Apache since it prevents your code getting closed, and better than GPL by not forcing others to open their code.
This article doesn't reckon with cloud providers (etc) eating your project. (2015) indeed.
BSD & Co are open source for developers
GPL & Co are open source for users
To me the larger question is whether or not GPLv2 achieves the goal of forcing big companies to contribute back to a project better than a more permissable license does.
As stated in the article, Linux is the notable success story, but anecdotally it seems that the majority of projects with large company contributions are licensed with BSD/Apache/MIT but with contributor agreement attached.
I wonder if anyone has compiled any data on whether GPLv2 does actually encourage contributions more than the more permissive alternatives.
GPLv3 isn't even worth mentioning in this context, great concept, abject failure to thrive IMO.
I think people who prefer GPL and I hope for the same outcomes but we have a different (irreconcilable?) philosophical approach to getting there.
In my view code (and all knowledge) wants to be free and a commons of knowledge enriches the whole world. I am opposed to most forms of copyright, patents and intellectual 'property'. Aside: My compromise position to this maximalist view is that I'd accept a 5 year copyright term with an exponentially increasing renewal fee.
For me MIT/BSD/Apache is a way to provide code with minimal encumbrances under the current dominant legal system. GPL is an attempt to free knowledge that relies on the legal system and the threat of men with guns coming to force you to comply. However noble the intentions at the end of the day it is reliant on state force and reduces freedom, it is very good at providing fees for lawyers.
Corporations can't embrace-extend-extinguish open source. This is because the source is always available. Sure they can use that knowledge to build a new, more popular, thing, but the existing source never goes away. It represents an un-enclosable commons.
> The flaw is that in the long run, it keeps getting easier to write software. And people love reinventing the wheel.
See: Bram's Law
In the long run MIT code developer becomes irrelevant
After the million dollar company forks it and includes it in their product without even saying hello. And the developer is left with a depression
I think each project, and its conditions, can take advantage of diferent licences, there are very successful projects in all of them
This article is a decade old. I wonder if the author still agrees with it given how the software and corporate landscape has changed over the past ten years.
While I have always used maximally permissive licenses on my own open source software, I've been rethinking that stance in the past couple of years. I'm not sure where I stand now, but I don't fully agree with this post.
In particular, perhaps my number one fear about the world at large is that the untethered effects of economies of scale are clearly leading to a net transfer of power into the hands of fewer and fewer corporate leaders.
Permissive licenses are arguably agnostic to that effect: anyone can use the software, corporation or not. But given that large corporations already have significant economies of scale, the emergent effect is that a corporation can extract more value out of a piece of open source software than you or I can. If your goal is to discourage a handful of oligarchs eating the world, a permissive license may be opposed to that.
It's sort of like breeding fish and dropping them in a lake. Sure, anyone can then grab their rod and reel and catch a few, so you aren't privileging the commercial fisheries industry by doing so. But once the trawler shows up, they're going to harvest a hell of a lot more fish than the dude with a bamboo rod.
You may be thinking this analogy doesn't work because software isn't like fish. Copying a piece of software doesn't remove a fish from the lake for others to catch. But think about this at one level of abstraction higher.
Copying software accomplishes nothing. It's just bits sitting on a disk. It's software being used by humans that matters. When a corporation takes a piece of open source software and puts it in front of users, time a user spends using that corporation's code is time not spent doing anything else.
While software itself isn't a consumptive good, human attention is.
Notice how all of the biggest, fastest growing corporations understand this. Attention is the ultimate economic commodity. Any company who can mine it effectively wins and any company that fails loses. This is why in the past decade we've seen seemingly weird business moves like Apple producing movies, NVIDIA doing game streaming, Amazon shipping games, Walmart selling video streaming, etc.
We are shambling towards a post-material world where the most valuable good, the thing that produces the most value, is human attention. And, unfortunately, a few people figured this out sooner than the rest of us and a gobbling up all of that mental real estate and leaving nothing for anyone else.
The fact that very large corporations are willing to spend tens and hundreds of millions of dollars to replicate software is exactly why the GPL is not irrelevant.
What makes the GPL irrelevant is AI. AI in it's current incarnation is basically a magic copyright remover. It can memorize all the patterns, tricks, algorithms, and architecture of open source software and implement it again. In a few years you won't need git because you will just have an AI stamp out a git compatible tool if you need one.
So if you care about Freedom, we need to move the battle from making sure you can get a copy of the source code to git, to making sure you can get a copy of the weights of the AI you are using to create software on demand.
All large FLOSS programs are GPL licensed.
Even with all the Rustaceans rewriting in Rust while abandoning the GPL, their flagship - Servo - is GPL licensed.
There is an aversion to CLAs among those donating their time, and with that an aversion to permissive licenses.
Im sure both will be around for a long time, but in the very long term GPL wins.
This article is already largely irrelevant. The GPL (and the FSF), whether you like it or not, always has been a political movement. The aim of the movement is to expand the pool of free/libre software and to disallow commercial entities from gaining an unfair advantage without contributing back. With the GPLv2 they already have, as it permits them to run the software in the cloud, with their proprietary additions, without contributing back. AGPLv3 closes that loophole that's why it's even less popular.
You can license your software as you wish, but in the long run the GPL has ensured that contributions reach back upstream for the common good, rather than for profit. The GPL gives protections for the people/end consumers, much like labour laws do in your own country. The GPL ensures that your contributions are respected, available to all, and not abused for profit (not always true, but tribunals have enforced the license terms before). The GPL has the effect of doing this globally while allowing contributions back from a global audience. It's genius and the companies absolutely hate it.