Modern P2P and federated protocols in 2026: nostr, ATProto, ActivityPub, Willow, Iroh, Farcaster

The last four years produced more genuinely new protocols for user-owned internet than the previous fifteen. Each has partisans, none has an honest cross-protocol survey, and the trade-offs between them (moderation model, discoverability, portability, monetization surface) are the interesting part; the vendor blog posts pretend the trade-offs do not exist. This page is a working map for a developer or technical PM deciding what to build on, or a smart lay reader who wants to understand what these words actually mean. I have deliberately not included blockchain-first protocols; the design constraints of consensus mechanisms are different enough that they deserve their own list.

The protocols themselves, canonical starting points

Read the reference docs before the takes. Each of these is the specification or nearest-equivalent source of truth.

The design debates that actually matter

Where the partisans disagree, and where the trade-offs are real. Read these to understand why the protocols look so different from each other.

Local-first and P2P as adjacent movements

Where user-owned social overlaps with user-owned data. The Ink and Switch canon plus the practical libraries.

Building on these protocols, practical entry points

For developers past the specs and ready to build. These are the SDKs and starter projects with actual momentum.

Thinkers, historians, and further reading

The people who have been writing about federated and P2P systems for longer than the current wave. The essential context.