Citation
Pugatsov, A., Ileri, C. U., Decouchant, J. “The Blockchain Execution Dilemma: Optimizing Revenue XOR Fair Ordering.” arXiv:2604.23266v1 [cs.DC] (25 Apr 2026). Delft University of Technology / IOTA Foundation.
Core Question
As DAG consensus pushes the bottleneck to the execution layer, transaction sequencing (not just parallel execution) becomes decisive. The paper studies the tension between validator revenue and order fairness as a dynamic optimization problem in a continuous execution model where block executions can overlap.
Method
- Blockchain-independent sequencing-layer model (the layer between consensus and execution that schedules txs across workers and manages congestion via deferral), building on Sui’s defer-on-congestion design.
- An anytime genetic algorithm that sequences using gas prices, object (read/write) sets, and predicted execution times — integrating execution-time + object-conflict data into the initial ordering, not just the deferral phase.
- Optional fair-ordering constraint: non-independent txs are always executed in consensus-determined causal order; independent txs may be reordered.
- Evaluated on real Sui data and synthetic Ethereum data, with perturbed execution-time predictions.
Headline Results
- Genetic sequencer raises validator profit ~15% vs the gas-price greedy baseline and clears congestion up to 58% faster (50–60% faster congestion relief).
- Enforcing strict fair ordering reduces validator revenue 50–60% during high congestion.
- First evidence that strict fair ordering effectively nullifies the advantages of advanced sequencing — the two paradigms sit at opposite ends of a spectrum (discretionary, opaque validator control vs rigid arrival-order fairness).
Why It Matters
- Quantifies the price of Order Fairness (Batch-OF) Protocols: rigid fair ordering is expensive enough that revenue-maximizing chains have a strong incentive to avoid it — context for why Ethereum pursues FOCIL: Fork-Choice Enforced Inclusion Lists (EIP-7805) (inclusion lists) and Encrypted Mempools rather than arrival-order fairness.
- The object-conflict-aware genetic scheduler is complementary to Block-Level Access Lists (BALs) and Parallel Execution (EIP-7928) (BALs make conflict sets statically available, which is exactly the input this scheduler needs).
See Also
- Order Fairness (Batch-OF) Protocols — concept page; this paper supplies the revenue-cost number
- Paper: Herring — Parallel Batch-Order-Fairness on DAG-based Blockchain Consensus — sibling Decouchant paper attacking the throughput cost of fair ordering
- Block-Level Access Lists (BALs) and Parallel Execution (EIP-7928) — object/conflict data that makes such scheduling tractable on Ethereum
- MEV Auction Design: Open vs. Sealed, Timeboost, and Kairos — the revenue side of the ordering-rights question