Love the idea! What if instead of stakeholders voting on everything you implemented a “steering committee” style model. Stakeholders meet/organize at some cadence to make larger decisions and decide the direction to “steer” the instance and the smaller decisions made in service of the direction decided by committee are left to the admins (decided/maintained by committee). The committee would have veto power over those decisions.

Just thinking of communication overhead and how the more is decided by committee upfront the less agile you can be.

They work just fine with real-debrid.

Yeah, it’s trivial work for a capable AI. It isn’t farfetched to envision an AI tracking down your alt accounts by analyzing writing style, post/comment topics, and various other bits of commonality.

My knee-jerk reaction is that I’m generally against it. I’m all for AI in a variety of applications, but I don’t participate in discussion in online places to give free training days to corporate LLM’s. If somehow it could be guaranteed that it was only used in open models I suppose I would feel a little better, but the second issue in my mind is that even careful people leave a trail of identifying breadcrumbs sprinkled across their posting history. A human having to sift through thousands of posts and comments will have a much harder time putting pieces together than an AI will. So I see it as a privacy concern mostly.