It should be designed as: 
1- the servers should inter-communicate using a p2p protocol 
2- we should query a server using many available protocols (including current websocket, but also webrtc and UDP) offering to them a utxo so they can charge us for their “query mesh service”.
3- complex query language só the server can be able to query against their entire database and measure the amount of data he can “sell”
4- if the server hasn’t enough data as requested (xxxKb) they should automatically request his other server peers for more information to buy, excluding the information he already has.

5- as at each request we are exchanging UTXO we can include blockchain info as (blocks, transactions) inside nostr protocol so we don’t need to rely on #lightning and can replace it completely by a buildin tech.

That’s the design I’ve thinking about. Something about mixing nostr and lightning into something I like to call reNostr p2p server.

#bitcoin #nostr #lightning #kademlia #reNostr 

 how about dedicated relays that only store links to notes / relationships between notes ( which one is a reply to which ) while all the actual data ( text, images, video ) is stored by users similar to how bit-torrent works.

this takes the load off relays making the system faster and more financially viable while the users shoulder the cost of hosting large files.

ultimately the users have to pay for the service one way or the other - let them pay with their storage and bandwidth.  

depending on how much storage and bandwidth you make available to the system will also determine what kind of storage and bandwidth the system will make available to you.  so some users will be able to watch 8K video and others will only be able to read plain text.

ultimate fairness.

@fiatjaf