Vitalik Pushes EIP-8288 Over New Precompiles

Vitalik Buterin made the case Sunday night for Ethereum to quit baking new cryptography into the EVM through precompiles and lean on EIP-8288 instead, writing that precompiles and the EVM are “the wrong place to put most cryptography support into Ethereum.”

— (@)

Enjoying this article?

Subscribe to Bankless or sign in

• The case against: Precompiles are built-in contracts that make specific cryptographic operations cheap, but every new one requires a hard fork plus an implementation in every client, and the data they check must still be posted onchain as calldata. Vitalik estimated that data budget tops out around 8 MB/s, PeerDAS included.

• The alternative: EIP-8288, which Vitalik co-authored with Thomas Coratger and which we covered in September, would have users prove any signature or proof scheme locally inside a STARK, with mempool nodes recursively folding those proofs together until each block carries just one. Under that model, he wrote, “Ethereum can then never add a new precompile again.”

• The lattice debate: His posts capped a weekend argument that started Friday, when a16z crypto released Aerie, which aggregated 32,768 lattice-based Falcon-512 signatures on a single laptop in 0.65 seconds. a16z argued the recursive proving that Ethereum’s hash-based plans rely on is “intricate and bug-prone,” while StarkWare’s Eli Ben-Sasson countered that lattices are “ripe for AI to attack.” Under EIP-8288, lattice fans could still use their schemes by wrapping them in a STARK client-side.

• Not settled yet: EIP-8288 remains a draft, and the spec still marks its leanSTARK route as “reserved.” The EF’s post-quantum roadmap meanwhile keeps “PQ sig precompiles” on the docket for a later fork it calls J*, so Vitalik’s no-new-precompiles stance isn’t yet reflected in the Foundation’s own plans.