Watched that linked video... And it kinda does have a point in that 10(!) year old talk. With the little compromises that stack up over time.
Like with the DRM-thing W3C where the eff got shunned by the Berners-Lee.
The same is happening with survaliance (though even in the US there seems to be resistance against that as with datacenters)
And the same is true for AI and open Source. And reflecting on what even I said like a week ago it kinda fits into the "small compromise".
Like AI/LLM came in, stole all the work, all the code, all the knowledge and become good at coding stuff to exploit this for-profit (while also taking all the stuff closed-source)
Pretty sure there is the question in the room if the collection of the training data and using that to create a model can substantiate a claim for breaching Copyleft licensing.
The Problem is proving the violation. And no big open-source Orga took up the torch to challenge this by a court.
Rather big open-source beacons like the Linux-Kernel jump on the AI bandwagoon. Which was that first "small compromise", other projects would follow also making small compromises along the way (see debian).
So once unacceptable behaviour (using AI in Coding gets more and more normal.
But what CAN we do? The Video suggests building something that does not compromise but how the hell do you do that with something like the linux Kernel? Forking while possible would an big burden with it.
I can't tell you tbh. Or if you compromized too much already...
RE: https://fosstodon.org/@psh/117192705643170458