I would say the future is in pooling resources.

Like it happens with torrents. As one p2p protocol very successful.

Self-hosting not applications, but storage and uniform services. Let different user applications use the same pooled storage and services.

All services are ultimately storage, computation, relays, search&indexing and trackers. So if there’s a way to contribute storage, computing resources, search and relay nodes by announcing them via trackers (suppose), then one can make any global networked application using that.

But I’m still thinking how can that even work. What I’m dreaming of is just year 2000 Internet (with FTP, e-mail, IRC, search engines), except simplified and made for machines, with the end result being represented to user by a local application. There should be some way to pay for resources in a uniform way, and reputation of resources (not too good if someone can make a storage service, collect payment, get a “store” request and then just take it offline), or it won’t work.

And global cryptographic identities.

Not like Fediverse in the end, more like NOSTR.