Not Fediverse, but an example of what works:
Quora used to have a fantastic crowd-edited question topic ontology/taxonomy for precise feed shaping with thousands of disambiguated topics interlinked in a graph/tree structure, but they enshittified the system away 2023, and now it's all LLM misunderstandings of what the question topics should be. The bot thinks a question is about apples when the question is a mathematics word puzzle; "How many apples...". The excuse for AI tagging was "tags can be abused" when in reality the crowd corrected the abuses quickly and reported the offenders. With the new AI tagger, my feed turned into viral trash from Quora-celebrities, mostly about current news.
They never made the obviously needed features of:
- being able to topic tag answers, not just questions (for when the question is general and the answer is specific, like "What do you think everyone needs to know in 2025?").
- the ability to follow a topic from a user, because most people write about many topics, so a follow-a-user brings uninteresting topics to my feed, but not all people write well about that topic I want to follow, so I'd need to follow that topic from good writers only.
We're making the same mistake with AI as we did with cars; not planning human future.
Cars were designed to atrophy muscles, and polluted urban planning and the air.
AI is being designed to atrophy brains, and pollutes the air, the internet, public discourse, and more to come.
We should change course towards AI that makes people smarter, not dumber: AI-aided collaborative thinking.
https://www.quora.com/Why-is-it-better-to-work-on-intelligence-augmentation-rather-than-artificial-intelligence/answer/Harri-K-Hiltunen
https://www.quora.com/Who-invented-the-modern-computer-look-and-feel/answer/Harri-K-Hiltunen