A user with a larger following has an easier time getting their content to trend because of their network (reposts). This happens before any algorithms find and amplify based on that trend (likes, zaps, reposts, bookmarks, any other proxy for “signal”). 

And previously existing networks speed up the forming of new connections in new networks. (Verification through posting npubs in other places, nip05)

So there is clearly a self reinforcing mechanism. So across some scale there will always be the “popular”. If you believe like I do, that humans weren’t meant for the kind of social scale they experience on a daily basis. Then “popularity” which we are all accustomed to as adolescents, is even an example of a social feature of humans that has gotten out of hand. Unless there are diminishing returns at some point in the scale, the reinforcement continues forever. 

This is where you are trying to put the trending squelch. Something to generate diminishing returns at some scale so it doesn’t go forever. To make smaller ponds. And maybe to make ideal sized ponds.

points:
1) The acknowledgment that across some scale will always exist a “popular”, “elite”. IMO this is human nature acting in a scale beyond what it was meant for. Biggest fish in the pond no matter how small the pond is above X.

2) The natural diminishing point for humans is physical interaction. Voice and touch. The technology is what enables the scale beyond what the human is innately prepared for. And any scale beyond that is a bit of a problem for the human as an animal. 

3) I think Nostr already has greater diminishing returns than the competitors due to decentralization in relays, WoT specific ponds, lack of centralized AI. But as tech it absolutely increases it beyond the natural limit.

4) Despite all the above, humans seem to choose this flawed mode of interaction. I think we get value from discovering useful and entertaining information so we keep pressing the button.

5) It probably can be improved by making the ponds smaller, increasing the friction of interaction. Increasing the diminishing returns. But how to do that without decreasing discoverability? 

 i think a lot of the solutions come from bypassing the current and ongoing decades of non-routeability between any two points on the network (for which supposedly IPv6 exists) and set up small service providers that provide simply relaying of messages from domain names to ports at the ends of VPN tunnels - and thus enabling personal relays

from personal relays, you have the set of npubs of interest to one of your friends, and from other friends you draw the events that propagate to their different list of friends

this would solve much of the questions you raise , by making each user a repository of their own relevant public events