james on Nostr: I have an idea. What if to be consider to be running a "full node" one participates ...

I have an idea. What if to be consider to be running a "full node" one participates in a database sharding proof of storage system, so it's ensured that the full chain will persist, but not necessarily on any single node? I can imagine a RAID striping type system across nodes. If a node keeps X of every Y blocks, and somehow proves it to its neighbors, and there are Y groups who all keep track of neighbors, the storage needed per node will be X/Y of the total size of the chain. Let's say 5 of 256, so adjacent groups have an 80% overlap, while reducing load to about 2%. Reconstruction would be quite rapid. If a group goes empty, the data could be reconstructed from four other groups, and a node could automatically manage to which group it is assigned. This would allow meaningful participation while greatly reducing load. Whole-chain full nodes would still be a thing for those who would want that, obviously.

To maintain the full node aspect of verifying chain integrity, each node proves to itself the change to the database, and the group publishes the agreed upon proof to the chain with a many-multi-sig (allowing for a latency), sorta like a parity, so the shard system will occupy exactly one transaction per block. This would become the basis, then, for a layer 2 of virtual addresses, wherein an on-chain address is non-custodially occupied by many sub addresses, and the per-shard database proof also contains the transactions of virtual addresses. Cross group transactions might have to be on-chain, but they could at least be rolled up after a signed cross-group acknowledgement (making the other group acknowledges the transactions into its virtual addresses). Cross-group lightning network could also become a thing.

This is starting to feel a bit like drive chains, NGL, but still a bit different.