In the discussion on algorithmic content delivery there seems to be a different side of it missed. Algorithms don't have to be used to smartly give you more content, they can also smartly choose what to display if you want less of it.
If you follow people who post a lot but only a part of their posts interests you as much as others', a pure chronological timeline is very sub-optimal. An algorithm can make you a timeline that prioritises the kinds of posts you like and de-prioritises ones you don't, so you get maximum value per watched posts, decreasing as the important new posts run out and the less important ones get presented. This can reduce FOMO of "I'm sitting here for a long time but haven't caught up on all the posts of my favourite people". This is something I remember Twitter doing pretty well, except of course with also inserting random crap it thinks you'd like in-between.
Furthermore, it doesn't have to be a learning algorithm that requires tracking your data, it can be a "dumb" one that simply has priorities set based on some qualifiers (author, keywords, the like) and creates a mix of mostly higher priority posts with some lower priority ones. It can all be configurable.
The fediverse with its open nature allowing one to choose their servers, clients and if they appear algorithms has a big potential for better UX that we're refusing by the allergic reaction to any mention of algorithms.
