Crypto glossary
zk-STARK
A zk-STARK is a zero-knowledge proof that needs no trusted setup and relies only on hash functions, producing larger proofs believed to resist quantum attacks.
STARK stands for scalable transparent argument of knowledge. Introduced by Eli Ben-Sasson and colleagues in 2018, STARKs let a prover convince a verifier that a large computation was performed correctly, with verification far cheaper than redoing the work. Transparent means the system has no secret setup parameters: all randomness is public, so there is no toxic waste that could be abused to forge proofs.
Because they rely on collision-resistant hash functions rather than elliptic curve pairings, STARKs are considered plausibly secure against quantum computers. They scale well for very large computations, which is why they power validity rollups such as Starknet and several zkVMs. Proving is fast and parallelises well on modern hardware.
The main drawback is size: STARK proofs are typically tens to hundreds of kilobytes, compared with a few hundred bytes for SNARKs such as Groth16, which makes verifying them on Ethereum more expensive. Many systems therefore wrap a STARK inside a SNARK before posting it on-chain, reintroducing some of the assumptions STARKs avoid.
Example
A rollup proves a batch of 100,000 transfers with a STARK, then compresses that proof into a small SNARK so the Ethereum contract can verify the whole batch for a modest gas cost.