Swapping BLS for leanXMSS leaves Ethereum's P2P layer classical, research post says

Victoria Shinder28 October 2026, 09:191 min read

A post on ethresear.ch by adust09 of the Nyx Foundation, published on 27 October, takes stock of a gap in Ethereum's post-quantum planning: the consensus layer's networking has no adopted migration profile. Replacing validator BLS signatures with the hash-based leanXMSS scheme does nothing for the identities and handshakes used by Ethereum Node Records, Discovery v5, libp2p Peer IDs or QUIC/TLS.

The author splits the work into three linked targets. Discovery: ENR signatures, identity proofs and key exchange. Connection key exchange: the TLS 1.3 key agreement inside QUIC. Connection authentication: the libp2p TLS identity binding and the TLS CertificateVerify step. The binding constraints are size limits — an ENR is capped at 300 bytes and a Discovery packet at 1,280 bytes — and the fact that a hybrid key exchange protects confidentiality but does not provide post-quantum authentication.

That is why, the post argues, hash-based signatures may suit consensus while lattice-based schemes are the de facto candidates for networking. It notes that Lighthouse, Prysm, Teku, Nimbus, Lodestar and Grandine all derive both the Discovery node ID and the libp2p Peer ID from one secp256k1 key, although the identifiers are not interchangeable. The analysis is limited to QUIC; the TCP/Noise fallback and the execution layer are out of scope but still need compatible work. The post explicitly does not choose an algorithm, encoding, fork or activation date.

What it means

The useful contribution is the map, not a proposal: it shows that "post-quantum Ethereum" is at least two migrations with different constraints, and that the networking one is boxed in by packet sizes most signature debates ignore. A validator with quantum-safe signatures that still discovers and authenticates peers with secp256k1 has moved the weak point, not removed it.