Weird licenses and the new open source
26 points by ema-pe
26 points by ema-pe
People who haven’t actually done license compliance seem to think that the JSON license story is a fun story.
But it involved one big corporation (IBM) getting a waiver.
If you have terms that make corporate lawyers go “nope”, what have you accomplished if you make an exception for one big corporation?
In case other people, like me, haven't heard about this and are wondering what the "JSON license story" is, it's told by Douglas Crockford here: https://www.youtube.com/watch?v=-C-JoyNuQJs#t=2480s
It's probably more fun to watch the video, but the TL;DW is: The JSON license is very much like the MIT license, but it includes the clause: "The Software should be used for Good, not Evil." This was an issue for IBM's lawyers, and they had to get a waiver for that.
I think we're pretty much in a post-license era when it comes to software. Open Source licenses haven't really done anything except provide a lot of free infrastructure to world-devouring corporations. Free Software has turned out to not actually have any features that distinguish it from Open Source software - it too is more or less explicitly produced for the benefit of world-devouring corporations. LLMs can be used to license-wash any free or open-source software for proprietary purposes. Proprietary licenses essentially amount to 'might makes right' at this point - whether they must be complied with depends on who you are. There was maybe the possibility that big proprietary software would face the same license-washing as free software, but the big AI companies have stepped in to protect them by putting guardrails around reverse engineering.
What I wonder now is how are the stainless steel rats going to share software behind the wainscoting of the technology industry without the people that matter getting ahold of it.
While it's true that the big bads have benefitted from open source, it's not true that no one else has. I think overall the benefits to the public outweigh the benefits to the enemy
I was willing to believe that up until 2016? 2020?
Do you think the benefit to the rest of society has decreased?
Well, Facebook, Xitter and Tiktok all exist, all built on a foundation of FOSS, and at this point may well be a net negative for society.
I think the real benefit was to engineers, who gained market power because the skills they were learning on the job would be transferrable between employers.
The employers got a similar benefit though, in that they started to see their employees as more replaceable.
Because the system hangs in a balance, who open source is helping more kind of depends on how the balance of power already sits. In an employer's market (like now), OSS seems to help employers. In an employees market, it helps employees. While this is a tenuous balance, I don't think employers can ever permanently swing the balance of power in their direction.
More than employers or employees though it helps end users: breaking lock in and providing options etc
I don't think any of those would be any less existent or any less evil if FOSS did not exist.
Free Software has turned out to not actually have any features that distinguish it from Open Source software - it too is more or less explicitly produced for the benefit of world-devouring corporations.
My takeaway over the last years here is that this isn't really a problem with Free or Open Source software, and rather with the "world-devouring" entities, which will, as you say, devour and exploit literally anything to its benefit, be it volunteer software, public infrastructure, the legal system, natural resources and even the populace.
I feel like human society has lost the plot. Not being able to do anything about that, I've resigned myself that avoiding engaging in things that are supposed to help fellow people, just to avoid the risk of your efforts being misused by the system, is probably counterproductive overall.
I kinda agree. With REA (reverse engineer anything) we're already seeing so-called "clean room" reimplementations of closed source software. I look at stuff like the FSL (Functional Source License from Sentry) which is used to provide a 2 year delay on MIT/Apache licensing to "source available no compete" terms and I wonder if that's the same level of defense that it was when it was created. I also think about how AGPL has been thought of as a defense against closed source but with agents it may be possible to completely isolate or clean room implement AGPL systems in a manner that is legally defensible or at least difficult for offended parties to seek damages from a multi-billion-dollar company (not that it's very easy right now).
Trying to make a free/open license that cannot be used for harm, cannot be used over X employees, cannot be used on full moon, is like trying to make a programming language that cannot be used to make spam bots, or making a pen that cannot spell hate speech. The IP holder would need to remain control over the program execution, because the arbitrary rules need their human judgment per every case. It is a hard pill to swallow, but software is hard to make while being easy to copy, maybe the acceptance is more healthy than reaching for cancerous growth of big software.
I'm pretty against the thesis of this article, and also against these new licenses that are not really open source or free software. Still the article might be useful to someone. In these times I've found that sticking to the most standard and widely used licenses is the strongest approach (copyleft licenses are my preference, but permissive ones like MIT or Apache are fine too). Even so, I'm not sure what the future holds or whether this is the right path.
I remain of the opinion that licenses are the wrong tool to try to solve problems related to human psychology, economic incentives, and software-design methodology. But they do keep being the tool the movement keeps trying, in spite of very little evidence that they actually achieve their desired results.
in spite of very little evidence that they actually achieve their desired results
I think they are quite effective in scaring away big companies from using them directly due to legal risks. This is unfortunately not true for indirect use through LLMs, but those don't respect copyright of anything else either…
I keep hearing that claimed but there is very little evidence behind it. Companies don’t avoid things because of the risk, they avoid things because the benefits don’t outweigh the risks. But big companies are far less averse to legal risks than small ones. In my experience (having worked with a large number of small companies and for Microsoft), copyleft licenses are much more likely to put off a small company that wants to do the right thing but can’t afford to deal with a malicious lawsuit if a copyright troll ends up buying a project’s IP than a big company than they are to put off big companies that actively want to circumvent the license.
I used to work at a bigcorp that would yell at you if you even had AGPL software installed on your machine (not even checked into source control, just installed like a normal program) and would do license audits of third party dependencies; any "weird" license would get rejected. I've also worked at small companies that cared, and small companies that didn't. (I've only worked at the one bigcorp so I don't have any other samples to compare to.) So I think it varies.
The Microsoft policy was rewritten while I was there (the original one with WSL was painful because apt install could violate policy). The new one allowed installing and using anything under an OSI-approved license. The only time that you needed to get approval was if something was shared with customers / partners or incorporated into a product. All of those discussions were cost-benefit analyses. Most of the GPLv3 things simply failed on the benefit side: they were small projects and it was easy to rewrite them and release a permissively licensed version. The exceptions were things like GCC, where the cost of making a Linux distro work with clang was not something the company wanted to pick up, so they shipped GCC in a bunch of things. Apple made a different analysis, in part because they also wanted to do a load of things that tightly coupled the compiler to other tools (Microsoft has their own compiler for this, so GCC is purely an 'other people want to use this' thing).
Possibly the same bigcorp, but where I worked the ability to install cowsay on workstations was restricted because the Debian maintainer (or someone else maybe, I forget) used an odd license like this. Fortunately common sense prevailed and they at least allowed things like cowsay but still defaulted to "weird license => don't allow implicitly".
I'm pretty against the thesis of this article, and also against these new licenses that are not really open source or free software.
I often hear the term post-FOSS license used for such licenses. Yes, they are not free software or open source, but that's not the goal anymore.
the best thing that licenses do is serve to do is act as a waiver of the creator's ability to deploy IP law against you. you're not going to get much else done with them.
My favorite weird license remains the Baba Yaga:
https://smallandnearlysilent.com/baba-yaga/LICENSE.txt
That said, GPLv3 is pretty weird already and seems fairly successful in scaring off big corporations.
I'd love a weird-license repositories in Linux distributions with lovely, quirky, useful, useless, handcrafted software!
Many options here: https://github.com/ErikMcClure/bad-licenses/
There is also the term Open Source, which is more or less Free Software but without the sociopolitical aspect of it, which is like a pizza without olives, or art without context: something devoid of meaning, a business-oriented product instead of a creation.
Unrelated to the topic of the article, but olives don't belong on pizza, they are a modification catering to consumer taste.
Consumer taste? I sure hope it does!
(That's the page for Neapolitan pizza, a specific type of pizza. The freedom to study and modify our pizza toppings is quite safe.)
Your link doesn't support your argument. At best, it says there are only two traditional kinds of "Neapolitan pizza", but that is just a subset of pizzas (even in Naples). Unless you consider it the only "real" kind of pizza, but that's not mentioned there :)
idk the Italians don't seem to have an issue with olives on pizza given pizza capricciosa and pizza quattro stagioni are a thing.
or do you consider only napolitan pizza to be pizza?
Do we care? Well, Linux distributions usually have strict policies (debian, fedora, …) when it comes to what software they package
Thankfully we don't have to care about distros anymore, at least for anything close to an "end-user app", because Flatpak exists.
If copyleft isn't the answer, maybe it's time to find better licenses? In my opinion, the freedom 0, being able to run the program for whatever purpose, might be the crux of the problem.
There is a reason why that particular freedom is first on the list. Preventing people from using your ostensibly-free software is like giving a gift, then taking it back. It also betrays totalitarian thinking and delusions of control over the behavior of fellow human beings.