If we can create our own custom algorithm then personally i will create: 1. Recommendation feed based on my topic interest. This can be started with analysis of users preferences (topics that they likes/zaps/reposts/comments) assuming we have topic classification (manual label or automated label) data on each notes/posts. 2. "What i have missed". Thankfully one example is already exist which is Pablo's DVM. It analyzes user activity time and check when they were inactive (have no posts/interactions) and will suggest notes that were posted by other users during their inactivity. 3. "Feeling lucky". This is similar to point 1 but instead of directly give recommendation based on user interest, it will give recommendation topic a bit outside of user preferences based on other people's preferences they have interacted with. Example: Alice likes Sport and Business. Bob (Friend of Alice) likes Business and Science. Charlie (Friend of Alice) likes Business and Travel. The algorithm will give Alice Science and Travel topic based on intersection of common interest (Business). Just a little bit idea @npub1zafcm... , maybe Damus with Nostrdb can achieve that 😅