Summary

Multi-Party Block Construction (MPBC) is a redesign of Ethereum’s transaction pipeline in which a single block is assembled from the contributions of multiple builders rather than one monolithic winning builder. Introduced May 2026 by the Blockspace Forum (Michael M., Kubi Mensah, Alex T., Drew V.; reviewed by Barnabé Monnot, Justin Drake, Thomas Thiery, et al.), it generalizes Relay Block Merging: a highest-bid “base block” is extended by operators who append non-conflicting transactions from other builders’ blocks, widening blockspace allocation from one builder’s view to the shared view of many. It is strictly additive, requires no protocol changes, and is designed for a world where ePBS: Enshrined Proposer-Builder Separation (EIP-7732) has absorbed the relay’s escrow role.

Key Points

  • Builds on the structural-gap analysis from the Buenos Aires and Cannes (April 2026, 32 teams, >95% of out-of-protocol blocks) Blockspace Forum workshops.
  • The base block is the highest-bid single-party block; operators append “missing” transactions (those absent from the base) that don’t change its execution guarantees, then add distribution txs for the surplus.
  • A multi-party block is only delivered if val(B_MP) ≥ max(val(B_SP), val(B̂_SP)) at delivery time — strictly improves on the PBS bid floor (same anti-gaming logic as relay merging).
  • Always extends the highest-bid base block (not the combination that would maximize total value) to prevent builders from spamming subset-blocks.
  • Initial scope: non-contentious state only (simple transfers, self-contained bundles) — fast append-only merging. Future scope: oracle updates, active preconf enforcement, contentious state.
  • New actor: the Operator — receives blocks/txs from builders, merges eligible contributions, submits the highest-paying block to the proposer. Operators compete; compensation scales with value generated. Not required (only used when they add value), unlike relays today.
  • Relays’ escrow role is replaced by ePBS (in-protocol trusted handoff for single-party blocks); operators are an out-of-protocol value-add layer on top.

Mechanism / How It Works

Each slot: (1) originators submit to builders + mempool; (2) builders construct single-party blocks, flag which are extendable and which txs are mergeable; (3) operators extend the highest-bid base block with non-conflicting txs from competing builders; (4) operators compare the multi-party block to the latest highest-bid single-party block at delivery and submit the higher-paying one; (5) proposer signs the highest-paying header across operators.

Two stages: merging (identify missing constraints C_missing, merge into B_SP, filter reverts to form B_MP, append distribution txs) and delivery (ΔV = val(B_MP) − max(val(B_SP), val(B̂_SP)); deliver B_MP iff ΔV ≥ 0).

When several builders contribute the same tx, it is attributed to the builder with the higher PBS auction bid (incentivizes competitive bidding). Exact value-distribution rule is an open research question.

Operator coordination (incentive-compatible info sharing): constraints (proposer commitments), geolocation (run the auction near the proposer), payloads (combine geographic coverage for propagation), and demotions (consistent view of excluded builders → less duplicated collateral).

Tradeoffs & Controversies

  • Robustness ↑: multiple inclusion channels, more diverse builder/operator set, transactions filtered/unknown to one builder still land — reduces defensive overbidding.
  • Operator trust: operators are a new semi-trusted role; can deploy TEEs (cf BuilderNet and Decentralized Block Building) to carry forward privacy guarantees and reconcile TEE-built with non-TEE blocks.
  • Largely subsumes Relay Block Merging but framed for the post-ePBS world and generalized to many contributing builders + competing operators.

Open questions: How is the MPBC surplus distributed so that originators, base builder, contributing builders, operators, and proposer all gain? Can operators be made trustless via economic/consensus mechanisms or TEEs/FHE without reintroducing relay-style centralization?

Timeline

  • 2026-04 — Cannes Blockspace Forum workshop explores solutions to the structural gaps.
  • 2026-05-27 — “Building towards Multi-Party Block Construction” introduces the initial MPBC design.

See Also