MEV, Pre-confirmations & Block Building

MEV is evolving from a simple front-running problem into a fundamental scaling constraint. The 2026 frontier: L2 spam caused by MEV bots consuming majority of throughput, and pre-confirmations as an application-layer solution to UX without waiting for block inclusion. (→ [[ethdenver]])

MEV on L2s: The Spam Crisis (Daniel Marzec / Flashbots)

The problem: 50%+ of gas on L2s is consumed by MEV bot spam — bots attempting ~350 failed transactions for every successful arbitrage. A $0.12 arbitrage costs 130M gas to execute — a 650× efficiency gap.

Root causes:

  • Transaction expressivity: no native conditional logic → bots must broadcast many speculative transactions
  • Private mempools: bots can’t see each other’s activity → no signal, pure brute force
  • Cheap fees: at 1¢/attempt, spam is economically rational
  • No efficient auctions: no mechanism to bid for MEV rights → blind competition on every block

Impact: L2 throughput capacity is being consumed by economically wasteful competition. This caps the effective TPS available for real user transactions.

Solutions:

  1. Programmable privacy via TEEs: searchers operate inside trusted execution environments with full transaction visibility, extract value more efficiently, drastically fewer attempts
  2. Explicit auctions: bid for MEV rights upfront rather than spamming; auction clears at competitive price; total attempts collapse; winners execute once

Early TEE-based results: searchers with full TX data extract same value with far fewer transactions. MEV auctions are not a luxury — they’re a scaling requirement.

Pre-confirmations (Murat Akdeniz / Primev)

What they are: Credible, cryptographically enforceable commitments from block builders that a specific transaction will be included in a specific block.

Economics: Builders post 5+ ETH stake (slashable); commitments are economically enforced even before ZK/cryptographic enforcement. If a builder commits and then doesn’t include, they’re slashed.

Current scale (Jan 2026):

  • 70%+ of block building power participates
  • 90%+ relay participation
  • 6,000+ validators on network
  • 120ms effective transaction confirmation time
  • 8% of blocks carry pre-confirmations

2026 vision: Pre-confirmations become a first-class Ethereum protocol primitive (like blobs and gas). Enables:

  • Instant wallet UX (confirmation before block close)
  • Synchronous composability between L1 and L2
  • The “instant” button in wallet UX: “hit the button and it’s done by the time you lift your finger”

Pre-confirmations are not a replacement for L2s — they enable asynchronous composability across the L1/L2 stack when combined with rollups.

EIP-7732 (ePBS): Protocol-Level Block Structure

EIP-7732 separates the execution payload from the consensus block. This is the protocol-level foundation for all advanced block building. (Terence Tsao / Ethereum Foundation)

Current bottleneck: Execution payload dominates block size (90–95%). Consensus is only 5–10%. But both are processed on the same tight timeline (2 seconds for the proposer to act).

The pipeline ePBS enables:

Slot timeEvent
0–4sProposer commits to validator set
5–6sPayload builder reveals execution payload
7–9sSeparate attestation committee (512 validators) validates payload
9–12sFinal finalization

Benefits:

  • Attesters now have 9–12 seconds (vs. 2 seconds) to download and verify
  • ZK provers get more time for proof generation within each slot
  • Enables slot auctions, MEV burning, and advanced validator market structures
  • Foundation for much shorter slot times (4 seconds target)

Proprietary AMMs (Prop AMMs) on EVM (Fahim Ahmed / Flashbots)

Context: Proprietary AMMs captured 73% of Solana volume by giving market makers control over their own price curves. Now landing on EVM (Base) despite historically higher gas.

Key design: Pricing contracts + oracle fallback; spreads widen for risky traders and imbalanced pools; negative spreads prevented by design. Wintermute and AlphaLQD are early deployments.

Why possible on EVM now: Batch updates (one transaction updating multiple price curves) are cheaper than per-order CLOB updates.

Best fit: Tokenized stocks and RWAs requiring cheap, frequent price updates without the latency trade-offs of on-chain CLOBs.

Continuous Clearing Auction (Dingyue Liu / Uniswap Labs)

A new token distribution mechanism that eliminates the gaming of traditional airdrops and fixed-price sales:

  • Problem with existing mechanisms: Airdrops get immediately dumped (66%), fixed-price auctions create gas wars on timing, insider allocations built-in
  • CCA mechanism: Demand curves aggregated across all participants; algorithm finds one uniform clearing price per block; repeated continuously across the auction period
  • Results (Aztec CCA): 99% of users contributed under 100K, 17,000+ unique bidders, $60M raised with no insider advantage

Weiroll: DeFi Execution Primitive (Colin Platt / Makina)

Lightweight on-chain VM for sequential DeFi execution:

  • Chains outputs across arbitrary protocols: Uniswap swap → Pendle deposit → Morpho borrow in one atomic transaction
  • Single revert protects the entire chain; no race conditions; no multi-tx gas bloat
  • Merkle-tree whitelisting: approve specific protocol/asset combinations for vault strategies
  • Can execute cross-chain (e.g., Hyperliquid perps via EVM precompiles without bridging)
  • Used by Machina for multi-chain institutional vaults

Weiroll and Brevity: Interpreted EVM Scripting Language serve similar use cases from different directions — Weiroll is opinionated about DeFi composition; Brevity is more general-purpose.

Connections

Open Questions

  • At what MEV auction participation threshold does L2 spam materially decline — or does it require mandatory auctions?
  • Will pre-confirmations remain an application-layer feature or get enshrined in protocol (as 2026 roadmap suggests)?
  • Do prop AMMs fragment liquidity from public AMMs, or do they draw in new flow that doesn’t currently exist on-chain?
  • How does ePBS interact with real-time mini-blocks (ETHGas) — can both run on the same block structure?