Scionic Merkle DAGs are actually much better than IPFS Merkle DAGs.
Plus, IPFS focuses on a P2P model that forces everyone to become a server — whereas Nostr is a user-server model with paid servers.
Specifically, Scionic Merkle DAGs have much smaller branches with far fewer useless hashes needed to create the same proof. It scales logarithmically, see stats below: https://github.com/HORNET-Storage/scionic-merkletree 
Scionic Merkle DAG branches are much smaller than IPFS Merkle DAG branches. The gains grow as the number of files in a folder grow, as the stats above demonstrate. nostr:npub1qny3tkh0acurzla8x3zy4nhrjz5zd8l9sy9jys09umwng00manysew95gx
I do believe that NOSTR needs a p2p small scale 2nd way to retrieve data from clients and servers. I made a post about it, hopefully I can find it. I have some problems with relays and accezsing or viewing my own notes.
Cant find it. All my messages from a time frame a week ago are gone. Cant see them nowhere anymore. This is one of the problems with Nostr that NEEDS to be solved!
a local backup of all profile and personal text data, in order to save yourself hassles with losing configurations or entire posts!! (including the data on posts that others gave (zaps, likes, reposts) in order to proof that you have send them in the past.)
Anyway, my message was like this in short:
People may need to cut on bandwidth and thus need to reduce amount of connections to NOSTR relays. Only dedicated relays for specific services and top 3 "global nostr" relays will be selected then. In other words: a relay needs to be really popular some point in order to stay in the top3 and thus the top 3 are a centralizing point of failure.
What do you think about this?
