In a Monday post on X, Buterin said his view of leanPCS had shifted after comparing three ways to process signatures and proofs. He cited a potential jump in signature size from 65 bytes with ECDSA to 1–6 kilobytes for quantum-resistant alternatives, while STARK proofs could reach 128 kilobytes. “This is unsustainable to put on chain, and 8288 fixes this,” he wrote.
The June draft of EIP-8288, co-written by Buterin and Thomas Coratger, proposes aggregating signatures and STARK proofs in Ethereum’s mempool before transactions enter blocks. Nodes would check evidence and produce proofs of claims, which could then be combined into a recursive proof for a block. The chain would keep short descriptions of verified claims, while the underlying data stayed outside ordinary block calldata. Buterin estimated about 64–96 bytes of information passed between nodes per claim, in addition to the aggregate proof.
The approaches differ in scope. LeanSPHINCS is designed for hash-based signatures and could support quantum-resistant wallets, multisignature accounts and transfers. LeanSTARK could verify a wider range of computations, including privacy applications and account rules that conceal sensitive relationships. But it would require developers to agree on an instruction-set architecture, a decision Buterin said could take time.
LeanPCS, or a polynomial commitment scheme, could offer similar flexibility without requiring that choice immediately. Buterin said its verification feature might allow complex STARK proofs to be checked through existing Ethereum execution for around 20,000–40,000 gas. He also cautioned that leanPCS could require several on-chain claims for one computation and be less compact than leanSTARK, while potentially being faster to implement.
EIP-8288 builds on proposed transaction format EIP-8141, whose frame design has been scheduled for Hegotá, though its specification is still a draft and activation dates are unset. The current EIP-8288 draft does not include leanPCS: its active dependency rules cover leanSPHINCS, while leanSTARK verification remains reserved pending further specification. The document sets verification charges of 3,000 gas for leanSPHINCS and 30,000 for leanSTARK, which is not enabled. It also caps transactions at 16 distinct signature dependencies and one STARK dependency, with one-second aggregation intervals for mempool nodes. EIP-8288 remains incomplete, and no mainnet activation date or hard-fork inclusion has been confirmed.

Comments (0)
No comments yet. Be the first!