I have chosen to not host source code for my own projects on Github at the time being, until such as time as:
- There are solid free software licenses available that forbids use of the licensed source code for statistical models used for anything but editing the source code itself
- And Github recognises those licenses
Information in the public domain could be freely available as training data to neural networks. However, just because something is publically available does not mean that it is in the public domain. Open Source software is an example of public information which is not in the public domain: open source software is still under copyright, for use under a license. See also:
- Software FreeeDom Conservancy's reasons for giving up GitHub
- A class-action lawsuit in US Federal Court against Github
This issue is just not about ownership. For one thing, I do absolutely intend to publish ugly source code that I have written to get good performance in/on specific cases/platforms/compilers, that which I absolutely not hold as being good examples for others to learn from. See also: CRAPL.
Apparently, Github doesn't have any intention to mend their ways. There are now also multiple actors doing the same thing. And that technology has also allowed for automated clean-room reimplementation from available source code, again without attribution.
The corporation that owns Github has for a long time wanted to destroy Open Source. In so doing, they have chosen also maximum insult: adopted the brand of the exploitation tool everywhere in its product lines, even so far as demanding affiliated computer manufacturers to put it on a special key on their keyboards.
Open source advocates often talk about the Four Freedoms. Some key people in defining the concept of "Open Source" back in the day are well known to be libertarians, and the concept of Open Source is definitely a reflection of their ideology. I think that like many other isms, libertarianism is a pretty idea that does not work in practice to serve the common man, because it does not take the dark side of human nature into account. For a society to function, there need to be laws, rules and regulations and active vigilance to keep the playing field fair, and people free from being exploited by bad actors.
We need to find a way to ensure a fifth software freedom: Freedom from exploitation.
The European Union has an "Artificial Intelligence Act" that requires web scrapers to respect copyright owners right to opt out of AI training. That is a step in the right direction, but it does not make any difference in practice because it is simple to ignore.
I am going to forthwith look for how to build a closed web, for people who have moral fibre, who believe in personal freedom, solidarity and community.
I think that perhaps software will have to be available in a web-of-trust, in which only software from verified authors (human authors) will be allowed, and from which it source code is not allowed to leak. Any license within this framework will have to require users to respect secrecy, and share source only with individuals trusted not to share the source with bots on the open web.
I had rather not wanted to do this. I want to create interesting, useful software and hardware for everyone to use and contribute. But I'd have to do this to protect it, to protect myself and our future. If you who read this feel like me, have a project that is aligned with this way of thinking, or know of some other project that does, then please reach out. (I have the same username on a couple of email services)