Vitalik pushes EIP-8288 to new precompilations

Vitalik Buterin made his point Sunday evening for Ethereum to stop integrating new cryptography into EVM via precompilations and instead rely on EIP-8288, writing that precompilations and EVM are “the wrong place to put most crypto support in Ethereum.”

— (@)

Do you like this article?

Subscribe to Bankless Or log in

• The case against: Precompilations are built-in contracts that make specific crypto operations cheap, but each new one requires a hard fork as well as an implementation in each client, and the data they verify must still be published on-chain as call data. Vitalik estimated that the data budget tops out at around 8MB/s, PeerDAS included.

• The alternative: EIP-8288which Vitalik co-wrote with Thomas Coratger and which we covered in September, would ask users to prove any signature or proof scheme locally inside a STARK, with mempool nodes recursively folding these proofs until each block carries only one. Under this model, he wrote: “Ethereum will never be able to add a new precompilation again. »

• The trellis debate: His messages ended a weekend row that began on Friday, when the a16z crypto was released. Areawhich aggregated 32,768 network-based Falcon-512 signatures on a single laptop in 0.65 seconds. a16z argued that the recursive proof behind Ethereum’s hash-based plans is “complex and bug-prone,” while StarkWare’s Eli Ben-Sasson countered that the networks are “ripe for AI to attack.” Under EIP-8288, networking fans could still use their schematics by wrapping them in a STARK client side.

• Not yet fixed: EIP-8288 remains a draft, and the specification still marks its leanSTARK route as “reserved.” The EFs post-quantum roadmap Meanwhile, the “PQ sig precompilations” are saved for a later fork called J*, so Vitalik’s stance on no new precompilations is not yet reflected in the Foundation’s own plans.