Ethereum’s Proof of Stake (PoS) with Proposer-Builder Separation (PBS) mitigates Maximal Extractable Value (MEV) concentration but relies on centralized builders and relays, with 80% of blocks currently proposed by just two entities. This compromises decentralization and fairness. This research proposes a decentralized system where all Ethereum clients propose blocks using a shared random algorithm. By leveraging Byzantine Fault Tolerance (BFT), it eliminates block-level MEV, fully democratizes block proposing, and may accelerate propagation, while supporting Danksharding’s rollup future. Compared to PBS, it prioritizes trustlessness over optimization, offering a transformative shift for Ethereum.
1. Introduction
Ethereum’s shift to PoS and adoption of PBS separates block proposing (validators) from building (specialized builders) to address MEV disparities. Yet, centralization persists: as of February 2025, approximately 80% of Ethereum blocks are proposed by just two entities—large builder-relay coalitions like Flashbots and their peers—concentrating power and profits. This undermines Ethereum’s decentralized ethos. MEV, the value extracted by reordering or censoring transactions, remains a challenge, with sophisticated actors dominating via PBS’s builder ecosystem.
This proposal introduces a decentralized random block proposal system. All Ethereum clients—not a handful of builders—construct blocks using a cryptographically random algorithm. Validators execute these blocks, achieving consensus via N ≥ 3T + 1 BFT. This eliminates block-level MEV, fully democratizes block proposing across Ethereum’s node network, and aligns with its trustless roots, while remaining compatible with Danksharding’s blob requirements.
2. Proposed System
2.1 Core Mechanism
-
Block Proposal: Every Ethereum client (e.g., Geth, Nethermind) runs an identical random algorithm, seeded by RANDAO and a Verifiable Delay Function (VDF) from the validator set. It selects transactions and rollup blobs from the client’s local mempool, ensuring uniform blocks across honest nodes with aligned mempools.
-
Broadcast: Clients relay their proposed block to validators simultaneously.
-
Execution and Consensus: Validators execute the block, prune invalid transactions (e.g., double-spends), and compute a hash of the valid remainder. With N ≥ 3T + 1 (N = total validators, T = faulty), the majority hash wins if 2T + 1 agree, tolerating up to T dissenters (e.g., 33%).
-
Rewards: Validators with the “correct” hash form a subset; a random subgroup is rewarded, keeping it fair and simple.
-
Blob Integration: Clients randomly select up to the maximum blob capacity (e.g., 16 blobs at 125 KB each) alongside transactions, supporting Danksharding.
2.2 Design Goals
-
Eliminate Block-Level MEV: Random selection prevents profit-driven manipulation.
-
Fully Democratize Proposing: Shifts block-building from two dominant entities to all clients.
-
Enhance Speed: Parallel proposals may reduce slot times (e.g., 6-8 seconds vs. 12).
-
Support Scalability: Handles rollup blobs for Ethereum’s future.
3. Analysis
3.1 MEV Suppression
Random selection ensures no entity can predict or control transaction order, eliminating block-level MEV (e.g., arbitrage, front-running). Unlike PBS, where builders extract and redistribute MEV, this system leaves mempool-level MEV as the only remnant—a smaller, less controllable slice.
3.2 Decentralization
Currently, 80% of Ethereum blocks stem from two builder-relay coalitions, centralizing proposing power. This system flips that: block-building spreads to thousands of clients globally, fully democratizing the process. No single entity dominates—unlike PBS’s builder pool or a centralized mixer—and BFT mitigates mempool variance, ensuring robustness.
3.3 Propagation Speed
Parallel client proposals and validator execution could shrink Ethereum’s 12-second slot to 6-8 seconds, outpacing PBS’s builder → relay → proposer → network flow. Mempool sync is key, but N ≥ 3T + 1 tolerates drift.
3.4 Resilience
Randomness blunts DDoS—spam dilutes the mempool but doesn’t bias selection. Client redundancy eliminates central choke points, surpassing PBS’s relay risks. N ≥ 3T + 1 handles up to 33% validator faults (e.g., T = 100,000 of N = 300,000).
3.5 Validator Simplicity
Validators execute, prune, and hash—lighter than PBS’s proposing and attesting. Random rewards within the correct-hash subset keep it straightforward.
3.6 Scalability and Danksharding
Clients pack blobs randomly up to capacity (e.g., 1-2 MB per block), supporting rollups. Randomness doesn’t prioritize specific blob inclusion, potentially delaying rollup sequencing versus PBS’s optimization. Mempool sync scales with blob broadcasts—feasible with Ethereum’s P2P—though drift risks BFT limits if T grows.
4. Comparison to PBS
| Metric | Decentralized Mixer | PBS |
|---|---|---|
| MEV | Eliminated at block level | Redistributed via builders |
| Decentralization | High (all clients propose) | Moderate (80% by 2 entities) |
| Speed | Potentially faster (6-8s slots) | Slower (12s slots) |
| Resilience | Strong (client redundancy) | Good (builder failover, relay risk) |
| Simplicity | High (execution-only validators) | Moderate (propose + attest) |
| Scalability | Good (blob-compatible, less optimized) | Excellent (roadmap-optimized) |
5. Trade-Offs
-
Advantages: Eliminates MEV for unmatched fairness; fully democratizes proposing beyond PBS’s 80%-two-entity bottleneck; speeds up blocks; leverages client redundancy; keeps validators light.
-
Drawbacks: Sacrifices gas-tip and rollup-specific blob optimization, potentially reducing L2 efficiency (this can be redesigned to favor L2 rollup-specific direction). Mempool sync must scale with blobs, risking consensus if drift exceeds T. Mempools may differ due to latency, so the initial block value V proposed by clients might not align. In this case, we select the most popular value V (via majority hash) or reselect at random from the proposed blocks, ensuring consensus persists.
6. Conclusion
With 80% of Ethereum blocks currently controlled by two entities under PBS, this decentralized random block proposal system offers a radical fix. It eliminates block-level MEV, shifts proposing to all clients, and aligns with Ethereum’s trustless core while supporting Danksharding. PBS optimizes for rollup scalability, but this system prioritizes fairness and full democratization—crucial as centralization creeps in. It’s a trade-off: trustlessness over L2 precision. For an Ethereum valuing equity over efficiency, this wins; for one chasing scalability, PBS holds. Redesigning for rollup-specific optimization could bridge the gap, making it a versatile contender.
7. Future Directions
-
Simulate mempool sync with blob-heavy loads to test drift tolerance.
-
Develop rollup-specific randomization to enhance L2 efficiency.
-
Test consensus fallback (most popular V or random reselection) under high T scenarios.