I wouldn't be surprised if Twitter really did have an issue with aggressive datamining by OpenAI, Google, Microsoft, etc and people getting training data for their own private LLMs
This will happen to instances on the Fediverse too. I dread my timelines being full of whinging about it though
> and there will be a netsplit based on whether your instance allows data to be "public/unauthenticated" (our current default) or "to privacy-preserving instances only"
except anyone can just modify their code to not obey and this mechanism will fail
@feld We will end up with a machine-readable contract for sharing data between fediverse nodes (i.e. server-to-server capabilities negotiation) and there will be a netsplit based on whether your instance allows data to be "public/unauthenticated" (our current default) or "to privacy-preserving instances only"
I think I'd like this to happen, some people will benefit from it.