This is an interesting idea. Replace follow count with eigenvector centrality score 🤔
Follower count is a lie anyway.
Maybe nostr:npub1ye5ptcxfyyxl5vjvdjar2ua3f0hynkjzpx552mu5snj3qmx5pzjscpknpr can build this into the #WoT stuff he is currently working on? Quality of connection could be determined by number of interactions, at least in part. Also age of connection, realness of peer, etc.
Any additional thoughts from the #SovEng crew? #WebOfTrust
Discussion
Not only that, but rank your own connections using eigencentrality (or similar). I want to glance at my follows and see the most relevant first, where relevant might be a mix of recent, real, frequent, etc.
please leave an option to drive stick. some of us want to make those decisions manually and don’t require ever increasing levels of convenience or feature creep.
Fair. Client diversity and forks of clients will hopefully offer all that. In the end it's filters and should be treated as such.
Lots of existing interfaces got this right imho, from app launchers to (old) reddit's comment filtering (new/top/best/controversial). 
Yup, filter UI's are the best benchmarks for this. #progressivedisclosure
Ideally, they wouldn't even be hardcoded and can be discovered (like DVM's).