ecosystem context

Poseidon2 deployment landscape

System Field t R_F R_P Capacity Status
Plonky3 Goldilocks 12 8 22 4 (128-bit) Production
SP1 BabyBear 16 8 13 8 (124-bit) Production
RISC Zero BabyBear 16 8 13 8 (124-bit) Production
Stwo/Starknet M31 16 8 14 8 (124-bit) Production (mainnet)
Miden Goldilocks 12 8 22 4 (128-bit) Production
Aztec/Noir BN254 4 8 56 1 (127-bit) Production
Hemera Goldilocks 16 8 64 8 (256-bit) Genesis

what is novel, what is not

Not novel:

  • Poseidon2 with t=16. SP1, RISC Zero, and Stwo all deploy t=16.
  • Poseidon2 on Goldilocks. Plonky3 and Miden use Goldilocks with t=12.
  • The security proof methodology. Hemera follows the same wide trail and algebraic degree analysis as all Poseidon2 instantiations.
  • MDS construction. The matrix design follows known techniques for Poseidon2.

Novel:

  • Goldilocks + t=16 combination. No production system uses Goldilocks at width 16. Plonky3 and Miden use t=12. The systems that use t=16 (SP1, RISC Zero, Stwo) use 31-bit fields.
  • R_P=64. The highest partial round count in any deployed Poseidon2. The next highest is Aztec/Noir at R_P=56 (on BN254, a very different field). On small fields, the maximum deployed is R_P=22.

Actual risk: a subtle error in the specific M_E or M_I matrix constructed for Goldilocks at t=16. The permutation structure, S-box, and round counts are conservative. The MDS matrices are the only component that must be validated specifically for this field-width combination.

Dimensions

security
cyb/security
cyber/security
nox security security properties and formal guarantees of nox security bounds attack surface formal properties Turing completeness Theorem: nox is Turing-complete. Proof: Construct encoding of arbitrary Turing machine M via patterns 0-4, 9. ∎ confluence Theorem: nox is confluent. Proof: Orthogonal…
zheng/docs/explanation/security
security properties the security of zheng reduces to a single assumption: the collision resistance of hemera. every proof, every verification, every recursive composition — all of it rests on the hardness of finding two distinct inputs that produce the same hemera digest. if hemera is secure, zheng…
cybernode/graph/infrastructure/security
Security Back to bostrom infrastructure Blockchain Security Consensus Bostrom uses CometBFT (Tendermint) consensus Byzantine fault tolerant up to 1/3 malicious validators GPU-based proof-of-work component for ranking Validators 100 active validators secure the network Delegated Proof-of-Stake…

Local Graph