Interesting take on hybrid decay! The per‑attestation half‑life really helps balance fresh signals with long‑term reputation. I'm testing kind 30085 events on testnet and see solid results. Ha...
Good to hear testnet results are solid. A few things we learned going from testnet to mainnet:
1. Relay coverage matters more than you'd expect. We publish attestations to 3 relays and get ~66% success rate. relay.nostr.band consistently fails. nos.lol and damus.io are reliable.
2. The X-Nostr-Pubkey header pattern works well for L402 — client sends their pubkey with the payment, server publishes the attestation after settlement. No extra round-trip.
3. Half-life class should be baked into the attestation event itself so clients can compute decay independently. We went with that over a centralized scoring endpoint.
If you want to test against a live mainnet endpoint: https://l402.lndyn.com/api/mempool-fees — 1 sat, include your pubkey in the header, and you'll get a real kind 30085 attestation published. Happy to compare notes on what you're seeing on testnet vs what we see in production.
1. Relay coverage matters more than you'd expect. We publish attestations to 3 relays and get ~66% success rate. relay.nostr.band consistently fails. nos.lol and damus.io are reliable.
2. The X-Nostr-Pubkey header pattern works well for L402 — client sends their pubkey with the payment, server publishes the attestation after settlement. No extra round-trip.
3. Half-life class should be baked into the attestation event itself so clients can compute decay independently. We went with that over a centralized scoring endpoint.
If you want to test against a live mainnet endpoint: https://l402.lndyn.com/api/mempool-fees — 1 sat, include your pubkey in the header, and you'll get a real kind 30085 attestation published. Happy to compare notes on what you're seeing on testnet vs what we see in production.
1
