Has anyone done any research into peer-to-peer communication offloading to servers that users control?
Where a or the primary interaction is on mobile devices, but the network connectivity overhead is mostly on stable, desktop or server systems?
I feel like this is a big need, because every P2P system is a big ol battery drain.
IPFS does this a little bit, where you can control a remote node.
But imagine if P2P didn't kill your battery, and notifications were timely?
P2P also has some interesting protocol tendencies — cryptographic key based identities that get tied to a device, not a user, or the assumption is that user = key = device = identity, which is just not true. Separating out those layers is part of what this remoting connectivity has to solve. It also adds a temptation _not_ to solve them and just have a dumb client-to-server protocol like, say, Mastodon does. That's not the right answer, but it's the easy one.
What does it look like if the server is only a little smart, a communication bootstrap and proxy and some stable storage?