Well there are things like mastodon, for example, they serve the first 2 purposes (content and networking) without any real tampering with recommendations. I us

Well there are things like mastodon, for example, they serve the first 2 purposes (content and networking) without any real tampering with recommendations. I use it, it is very popular. UI wise it is almost identical, UX wise it has some unique behaviors that come with very powerful advantages IMO. It also doesn't have any design features built with a goal of maximum engagement. For the most part, I'd say people have found recommendation engines to be nuisances at best, tools of control to many. The idea of the network effect seems to suggest that the largest draw with these sites is the content. Of course, the recommendation engine has it's effect on the perceived content, and it's possibly addictive nature has an effect on the available content, so the interaction there is still fuzzy. As another comment here pointed out, people see the algorithm the same way a fish sees the water it is in. They only see content, in their mind the site is the content, they don't see an engine. My idea of a "maximum wellbeing" algo for recommendations is a chronological timeline. It is up to me to engage with what I want to engage with. Maybe that isn't ideal for applications similar to YouTube. I've seen a pretty simple ranking algorithm used in a project called Lemmy (FOSS federated community oriented link aggregator) that I think is phenomenal.