Glossary
Zero-knowledge proof (zk-SNARK, zk-STARK)
A cryptographic proof that a statement is true without revealing why, used in crypto to verify transactions privately and to scale blockchains.
On this page
A zero-knowledge proof lets one party, the prover, convince another, the verifier, that a statement is true without revealing anything else. You can prove you know a password without sending it, or that a batch of transactions follows the rules without the verifier redoing every one. The idea was first described in a 1985 paper on interactive proof systems by Shafi Goldwasser, Silvio Micali and Charles Rackoff.
A valid scheme has three properties:
- Completeness. A true statement, honestly proved, is accepted.
- Soundness. A false statement is rejected, except with negligible probability.
- Zero knowledge. The verifier learns only that the statement is true.
SNARKs and STARKs
Two families dominate in crypto:
- zk-SNARK (succinct non-interactive argument of knowledge). Proofs are very small and fast to check, which keeps on-chain costs low. Many SNARKs need a one-time trusted setup: if the secret randomness from that ceremony leaked, someone could forge proofs. Ceremonies with many participants reduce the risk, since one honest participant is enough.
- zk-STARK (scalable transparent argument of knowledge). No trusted setup; the randomness is public. Proofs are much larger, which costs more to post, but they scale well for very large computations and rest on hash functions that are expected to hold up against quantum computers.
Uses and limits
The main uses are zero-knowledge rollups, a kind of layer 2, and privacy coins or tools that hide amounts and parties. Generating proofs takes heavy computing, so many systems rely on a few specialized provers. And a proof only shows that the code's rules were followed: a bug in the circuit or the smart contract that checks it can still let bad transactions through.
