Summary: Human Network is best understood as a lower-layer cryptographic service beneath identity and proof products rather than as another credential app. Its core mechanism is a threshold scalar-multiplication network that performs shared-key elliptic-curve operations on masked user inputs, letting applications derive secure keys from low-entropy secrets or enforce programmable decryption without exposing the underlying inputs to any one operator. That makes Human Network a useful comparison class for Human ID, Human Passport, Self Protocol, Sismo, and wallet-recovery systems because the main control surfaces are relay ingress, operator admission, epoch resharing, optional semi-trusted add-on nodes, and the split between key-derivation infrastructure versus downstream credential or proof products.
What it does:
Operates a threshold network for blind elliptic-curve point-by-scalar multiplication over user-supplied masked points
Supports OPRFSecp256k1 for deriving secure keys or nullifiers from low-entropy inputs such as passwords, biometrics, and security questions
Supports JWTPRFSecp256k1 for deriving keys from web-account credentials through a permissionless-style node network rather than a fixed issuer-side service
Supports DecryptBabyJubJub for provable encryption and configurable decentralized decryption rights or reveal conditions
Markets use cases like private biometrics, wallet recovery, ZK identity, compliance-triggered reveal, and undercollateralized lending
Runs inside the broader human.tech stack and is described as an AVS secured by EigenLayer and Symbiotic, with resharing-based and permissioned deployment modes documented separately
Key claims:
The clearest way to classify Human Network is as threshold-cryptography infrastructure, not as a standalone identity registry. The official docs define Human Keys as keys derived from human attributes and explain that the network adds entropy to low-entropy inputs without seeing the sensitive underlying data.
The official methods page makes the product split unusually legible: one method for low-entropy OPRF key derivation, one for web-account-derived keys, and one for provable encryption plus decentralized decryption. That is analytically stronger than flattening it into a generic identity or wallet recovery label.
The GitHub README is explicit that the core primitive is threshold scalar multiplication over elliptic-curve points with DKG, DLEQ proofs, and Lagrange interpolation. That means the reusable insight is the cryptographic service layer itself, not merely the human-facing app surface built on top of it.
The documentation positions Human Network beneath several downstream products and use cases — secure keys, nullifiers, private biometrics, ZK identity, and conditional-reveal flows. This makes it a useful lower-layer comparison point for Human ID, Human Passport, Self Protocol, and related proof systems.
The main control surfaces are visible in the docs. Requests flow through a relay node; operator participation depends on EigenLayer or Symbiotic plus Human Network registration; epoch resharing determines how node churn and permissionlessness work; and Robustnet remains explicitly permissioned because resharing is disabled there.
The README also introduces optional semi-trusted nodes as additive-share companions for users worried about threshold-network collusion. That is analytically important because it shows the system does not treat decentralization as binary; it exposes an extra trust-composition layer above the network itself.
Human Network belongs in the active corpus because it separates a lower cryptographic control plane — key derivation, nullifier generation, programmable decryption, relay ingress, and operator economics — from the higher-layer credential, attestation, and proof-container products that usually sit on top of it.
Whitepaper: No standalone Human Network whitepaper was located during this pass. The main primary-source packet is saved at ../whitepapers/human-network-primary-sources-2026-05-11.md.