Unfortunately, I have no clue how this could be solvable. _Perhaps_ part of the root cause is the decentralized structure and different routing to actually access any given profile/note ... So the easy way out would be not only a faster information protocol but also nodes with more locally stored data - aka centralization, which simultaneously steer small node operators away from the network (cost). Maybe better client side caching could solve it more sustainable - e. g. when reading a thread the profiles of the thread participants might already be loaded. With the cost of increased network traffic. Having all clients behave as small nodes would be cool but probably not feasible. If that is, how its already how it works, I'll better shut up :D I am not a technical person really.