Computational cryptography › Live on mainnet today
Trust or structural assumption

BN254 SNARK setups

In plain words

Some zero-knowledge proofs needed a one-time ceremony whose secret must be destroyed by at least one participant.

The idea: These proof systems need public numbers built from a secret τ, so that provers can do math "at τ" without knowing it. Whoever knew τ could fake proofs, so it must be destroyed. A ceremony makes that easy to trust: each participant mixes in their own secret randomness, so the final τ is effectively the combination of all of them, and it stays unknown as long as a single participant threw their piece away.

Why it matters: If every participant colluded, fake proofs would be possible.

The details

On the map: Powers of Tau + per-circuit ceremonies: ≥ 1 honest participant each

Groth16 needs secret randomness generated once per circuit, and PLONK needs it once for all circuits[1, 2]. Whoever knows that randomness (the "toxic waste") could forge proofs.

Ceremonies spread the secret across many participants: a "Powers of Tau" phase shared by many projects[3, 4], plus a per-circuit phase for Groth16. The result is safe if at least one participant in each phase destroyed their contribution.

Rests on

If any of these breaks, this breaks too.

If it falls

If this breaks, these break with it.

Everything that falls with it (3 boxes)

References

  1. Jens Groth. On the Size of Pairing-Based Non-interactive Arguments. Advances in Cryptology – EUROCRYPT 2016. doi:10.1007/978-3-662-49896-5_11
  2. Ariel Gabizon, Zachary J. Williamson and Oana Ciobotaru. PLONK: Permutations over Lagrange-bases for Oecumenical Noninteractive arguments of Knowledge. IACR Cryptology ePrint Archive, Paper 2019/953. eprint.iacr.org/2019/953
  3. Privacy & Scaling Explorations. Perpetual Powers of Tau. 2019. github.com/privacy-scaling-explorations/perpetualpowersoftau
  4. Scalable Multi-party Computation for zk-SNARK Parameters in the Random Beacon Model. IACR Cryptology ePrint Archive, Paper 2017/1050. eprint.iacr.org/2017/1050