Staying online without servers

The most common gotcha in P2P: you publish something, close your laptop, and it
is gone. Data lives on peers; if no peer holding it is online, nobody can fetch
it. We call this the laptop-lid problem.

The honest answers

1. More readers = more replicas. Anyone who read your post holds a copy and

can serve it. Popular content stays alive on its own. Unpopular content —
the long tail — does not.

1. Always-on peers. Some machine has to be awake. The trick is keeping it

a dumb peer instead of letting it become a server.

Relays: dumb storage, not authority

HiveRelay is our answer: a
fleet of always-on nodes that will pin (replicate and re-serve) any
Hypercore you ask them to. The crucial design line:

• A relay cannot forge anything — logs are author-signed, so a relay is

just a mirror.

• A relay does not index, validate or moderate — it stores bytes. All

meaning lives at the edges.

• Losing every relay degrades availability, never integrity.

p2pbuilders seeds your log (and every author you track) to the relay fleet on
boot — the relays command in the TUI shows who is pinning you. Posts
survive with 5+ relays across 3 continents even when every author sleeps.

Design guidance

• Treat relays as a cache tier, not a dependency. The app must work

peer-to-peer with zero relays reachable.

• Seed everything the data needs to be reconstructed — a lesson we learned

the hard way with Hyperdrives; see
Seed the blobs core.

• Re-announce periodically. Pins age out; a live peer re-requesting keeps them

fresh.