Summary

On high-throughput, low-fee blockchains (Base, Optimism, Arbitrum, Solana), a qualitatively different form of MEV has emerged: searchers submit large volumes of speculative transactions whose profitability is resolved only at execution time, rather than pre-computing an opportunity off-chain and sending a single targeted transaction. This “spam MEV” consumed >50% of block gas on Base and Optimism in Q1 2025, with only 6–12% of probes resulting in a trade.

What Makes Spam MEV Different

On Ethereum L1 (Traditional MEV):

  • Searcher pre-computes profitable opportunity off-chain
  • Submits single targeted transaction with high confidence
  • Failed bids are filtered before execution (builder-level)
  • Competition mediated by block builder auctions

On High-Throughput L2s (Spam MEV):

  • Both opportunity detection and execution reside in on-chain smart contract logic
  • Transaction probes whether an opportunity exists at execution time
  • If no opportunity: transaction reverts or consumes gas without a trade
  • High volume, low cost per probe → profitable even at low success rates

Example: Cyclic arbitrage bot on Ethereum L1 precomputes a path and executes atomically. On Base, the same bot speculatively submits hundreds of transactions probing liquidity pools, accepting a high failure rate in exchange for marginal profits.

Empirical Scale

  • Base and Optimism Q1 2025: spam MEV >50% of block gas
  • Trade rate: only 6–12% of speculative probes result in actual trades
  • Fee share: spam MEV transactions pay <25% of total fees despite consuming >50% of gas (low priority fees)
  • Cause of growth: sharp increase after block capacity was scaled up; fell when minimum gas prices were introduced

Why Spam MEV Is Viable on L2s

  1. Low transaction fees: failed speculative transactions are cheap enough that repeated probing is profitable even at low success rates
  2. Fast block times (<1 second on rollups): insufficient time for searchers to observe state changes and submit targeted transactions before the next block
  3. No public mempool: centralized sequencers + Solana-style direct-to-producer forwarding limits observable mempool information
  4. Encrypted mempools (if deployed): further reduce information available for targeted extraction, increasing spam

Theoretical Framework (Wang et al. 2026)

Equilibrium model: spam transactions compete for on-chain opportunities in a competitive equilibrium that drives their profits to zero.

Equilibrium spam volume = f(block capacity, minimum gas price, fee mechanism)

Three main results:

  1. Spam is always costly:
    • When blockspace is scarce: spam displaces users and drives up gas prices
    • As blockspace grows: spam increasingly consumes execution resources (externalities for nodes)
    • Spam takes an increasing share of each additional unit of block capacity
  2. Priority fee ordering reduces spam: spammers must pay more to reach early block positions
  3. Spam’s share plateaus: as user demand grows and blockspace scales, spam share stabilizes rather than growing indefinitely

Key policy insight: capping block capacity before all users are included creates a favorable trade-off — forgoing a small amount of user welfare eliminates disproportionate spam and externality.

Probabilistic Backrunning (Mazorra, Schlegel, Mamageishvili 2026)

A related phenomenon on FCFS (First-Come-First-Served) chains: searchers compete to place orders immediately after an opportunity (oracle update, arbitrage trigger) that arrives at an unpredictable time.

Equilibrium structure:

  • Opportunity arrives uniformly at random during a time interval
  • Observed with a delay; exact arrival time unknown during the window
  • Searchers randomize submission timing — not pure speed competition but strategic positioning in time
  • At high cost (c ≥ 1/e): at most one transaction per searcher in equilibrium
  • At low cost (c < 1/e): multiple transactions (spam) in equilibrium
  • “Spam” is the natural equilibrium outcome of probabilistic backrunning under low costs

Worst-case inefficiency: spam volume grows with competition; the equilibrium is socially inefficient even though it’s individually rational.

Spam MEV and Encrypted Mempools

Paradoxically, encrypted mempools (LUCID, GhostPool) may increase spam MEV by removing the ability to observe and target specific opportunities. When searchers cannot see pending state changes, they must probe blindly — exactly the spam pattern.

This is a key design tension: encryption protects users from frontrunning but may push MEV into the spam modality, which is harder to mitigate.

Mitigation Approaches

ApproachHow it helpsDownside
Minimum gas price floorRaises cost per spam probeDiscourages legitimate low-value txs
Priority fee orderingSpam must bid higher to be earlyComplex on FCFS chains
Block capacity cappingLimits spam volume directlyReduces throughput for users
Targeted builder-level filteringLike L1: filter failed probes before executionRequires centralized builder
Private mempool with access rightsReduces blind probingCentralizes info access

Relationship to L2 Design Choices

The prevalence of spam MEV is a direct consequence of L2 design choices:

  • Arbitrum’s FCFS sequencer → probabilistic backrunning spam
  • Optimism/Base’s low fees → probe-and-hope strategy
  • Ethereum L1’s builder auctions → traditional targeted MEV (less spam)

ePBS and BALs help L1 maintain targeted-MEV dominance. L2s migrating toward auction-based or builder-based ordering (Timeboost, Kairos) can reduce spam but face their own tradeoffs.

Key Sources

  • Blockspace Under Pressure: Spam MEV on High-Throughput Blockchains (Wang, Saraf, Heimbach, Babel, Zhang, 2026) — Framework; empirical evidence from Base/Arbitrum; policy analysis
  • Timing Games: Probabilistic backrunning and spam (Mazorra, Schlegel, Mamageishvili, Flashbots/Offchain Labs, 2026) — Game-theoretic equilibrium of probabilistic backrunning; spam as equilibrium outcome