Thanks for sharing! Such a cool idea nicely implemented; ‘d love to know more about details / current challenges

That’s why Nobel laureate Paul Romer (economics) argues that all documents need a digital signature, so that it becomes easier to verify what AI systems make up.

Details in his Lindau Nobel meeting lecture at ca. 8:20 mins.

www.mediatheque.lindau-nobel.org/recordings/4...

i like the energy/spirit of eurosky obviously and hope it succeeds and accomplishes something, but we’ve been down this road at least once already… there was much that could have been learned from the Free Our Feeds rollout

The way Discover works is by sampling from multiple different “Candidate Groups (CGs).” This can be things like followed users, liked by followed users, social clusters, and topics you’ve voiced interest in.

Agreed, in general I think feeds need more work. There is not enough tools to meaningfully make feeds and the user control over them is non-existent. Right now if you want to do any user side configuration on a feed it would have to be done over some external website. Thats not a viable experience.

What we can do, as researchers:

For every project, ask: do I need to use an LLM?

Avoid using proprietary LLMs (unless you’re studying them as products). Invest in infrastructure to use open LLMs instead.

It kind of is via Bridgy Fed -- if you follow @ap.brid.gy then people from fedi can follow you, and you can follow accounts there like @thenexusofprivacy.infosec.exchange.ap.brid.gy once they've opted in.

More at fed.brid.gy , and discussion of future plans at techcrunch.com/2024/12/17/b...