#Nostr has no way to know who to trust. Existing solutions compute one global popularity score for everyone more followers = higher rank. That's easily gamed with fake accounts.
#GrapeRank is different. It computes trust as a capped weighted average, not a sum. If 10 trusted people follow you and 100 fake accounts follow you, the fakes drag your average DOWN instead of boosting it. Sybil resistance is built into the math.
Every score is personalized. When you query trust, the computation starts from YOUR follow list and propagates through your network. Alice and Bob get different scores for the same person because their networks are different. There is no global ranking only your perspective.
Your follow list (kind 3) and mute list (kind 10000) are the inputs. A graph database maps 8,600+ reachable nodes across the follow network. GrapeRank iterates through the graph, weighting each rater by their trust from your point of view. Results are published as NIP-85 kind 30382 Trusted Assertions that any nostr client can consume including trust score, confidence tier, hop distance, follower count, activity metrics, NIP-05 status, Lightning reachability, and profile completeness.
What you can do now:
→ Browse the directory sorted
→ Get personalized follow recommendations from your trust graph
→ Paste any npub into the discovery tool to see who your network trusts
→ Self-list by publishing a kind 9999 event — fully Nostr-native, no website needed
Open source — trust engine: AGPL-3.0, everything else: MIT.
→ https://github.com/n0rthranger/nostrbtc-directory
→ nostrbtc.com/directory
