# Blockspace Under Pressure: An Analysis of Spam MEV on High-Throughput Blockchains

Crypto.world

> On fast, cheap blockchains, searchers flood the network with speculative transactions that only discover at execution time whether a profitable trade exists.

**arXiv**: 2604.00234

**Published**: 2026-03-31

**Authors**: Wenhao Wang, Aditya Saraf, Lioba Heimbach, Kushal Babel, Fan Zhang

## AI summary

## Overview

**Research area:** Blockchain economics and mechanism design — specifically maximal extractable value (MEV), transaction fee mechanisms, and blockspace allocation on high-throughput chains. The paper is filed under Computer Science and Game Theory (cs.GT) and carries CCS tags for "Distributed systems security" and "Economics of security and privacy."

**Technical level:** Advanced. It combines a formal equilibrium model (demand curves, zero-profit entry conditions, closed-form solutions) with empirical chain data. Readers need comfort with marginal analysis and basic game theory.

**Scope:** The paper builds a framework that predicts how many spam MEV transactions a chain will attract as a function of block capacity, minimum gas price, and transaction ordering rules, then tests those predictions against data from Base and Arbitrum.

## What This Paper Is About

On fast, cheap blockchains, searchers flood the network with speculative transactions that only discover at execution time whether a profitable trade exists. These probes can eat more than half of block gas while rarely producing a trade. The paper asks a design question rather than a descriptive one: how do the knobs a chain operator controls — block size limit, a minimum gas price, and the fee/ordering mechanism — determine how much of this spam appears, and what does that spam cost users, validators, and the network itself?

## Key Contributions

1. **Closed-form equilibrium spam volume and gas prices under random ordering.** The authors derive the equilibrium spam count S* as a function of block capacity B_max, floor price g_min, and demand, showing three distinct regimes: no spam entry, a "slack" equilibrium pinned at g_min, and a congested equilibrium where spam pushes the clearing price above the floor.
2. **A welfare, revenue, and externality accounting of spam.** They compare a world with spam to a counterfactual world without spam at identical parameters, measuring user welfare, validator revenue (R = g*·B), and network externality (E(B_max) = c1·B_max + c2·B, where c1 is per-unit capacity cost and c2 per-unit computation cost).
3. **Practical parameter-setting rules, including a "minimum marginal user share" (MMUS) criterion.** The baseline rule scales block size to B_plat, the point where user welfare plateaus. The refined rule stops earlier by requiring that at least an η fraction of the next unit of capacity go to genuine users, since spam captures an increasing share of marginal capacity.
4. **Extension to priority fee ordering (PFO) and to demand scaling.** The paper analyzes an approximate PFO mechanism (showing it raises the cost of early block positions and therefore reduces spam), and studies whether spam absorbs capacity indefinitely as demand grows, concluding its share plateaus rather than growing without bound.
5. **Empirical case studies of Base and Arbitrum** that the authors say support the model's predictions. Note: the supplied text is truncated partway through Section 3.3, so the detailed quantitative results of Sections 4 through 7 are only available in summary form via the contribution list, not in the excerpted body.

## Main Findings

- **Spam is always costly, but the cost changes form.** When capacity is scarce, spam displaces genuine users and raises gas prices; as capacity grows, spam increasingly consumes execution resources and raises network externality (the cost of provisioning and processing blocks).
- **Spam takes an increasing share of each additional unit of block capacity.** The marginal user share m_user is strictly decreasing in B_max under the entry-and-congested regime (Proposition 3.2), for any twice continuously differentiable, strictly decreasing demand curve satisfying a stated curvature condition — which holds for both linear and exponential demand. Consequently, capping capacity before all users are included is a favorable trade: a small loss of user welfare eliminates a disproportionate amount of spam and externality.
- **Equilibrium spam peaks at the plateau capacity B_plat.** Below B_plat the block is effectively scarce, so added capacity lowers the clearing price and increases spam. Once B_max reaches B_plat = D(g_min) + (r0·D(g_min)/(D0·g_min) − s)₊, the price is pinned at g_min and spam volume stops growing with further capacity.
- **Priority fee ordering reduces spam.** Because spammers must pay more to reach early block positions, approximate PFO raises their cost relative to random ordering. The paper reports that on Base, spam "appears later in the block as predicted by the PFO analysis."
- **Spam's share of block capacity plateaus as demand scales.** Using empirical data, the authors estimate that MEV opportunity size grows approximately linearly with non-spam activity; under that assumption, spam remains a nontrivial share of included gas as demand and capacity grow, but its share levels off rather than diverging.
- **The model's slack-regime solution matches an independent result in the literature.** In the slack case, S* = r/(s·g_min) − 1 = 1/c − 1 where c is a cost-to-reward ratio; this coincides with the large-n limit of the lower bound [1/c − 1, 1/c] derived by Mazorra et al. from an n-player timing game, despite a completely different modeling approach.
- **Empirical magnitudes reported in the paper.** On Base and Optimism, spam MEV consumed over 50% of block gas in Q1 2025, yet only 6–12% of speculative probes resulted in an actual trade, and spam paid less than a quarter of total fees. On Base, spam gas grew 122× from Dencun (March 13, 2024) to its April 2025 peak while non-spam gas grew 11.2×. After the gas target was cut from 70M to 50M on June 18, 2025, spam gas fell 34% and non-spam gas 24% over the following 30 days, with spam's share of total gas dropping from 36% to 32%. From October 2025 to February 2026, average daily spam gas fell from 450 Bgas/day to 302 Bgas/day (−33%).

## Methodology in Plain English

The authors treat spam as a rent-seeking contest. A block contains some number of on-chain opportunities (arbitrage paths, liquidatable positions) whose total value grows with how much genuine user activity the block contains. Spam bots submit fixed-size transactions; each transaction can only capture opportunities that appear *before* it in the execution order and that no earlier spam transaction already took. Under random ordering, with S spam transactions in the block there are S+1 equally likely positions for an opportunity, and exactly one of those positions leaves it uncaptured — so the probability an opportunity is claimed is S/(S+1), and each of the S symmetric bots earns r/(S+1) in expectation.

Bots enter freely as long as expected profit is positive, so in equilibrium each included spam transaction earns exactly zero net profit: expected MEV revenue equals the fee, which is its gas limit times the clearing price. That single zero-profit equation, combined with a linear demand curve for genuine users and a protocol price floor, yields a closed-form spam count. The clearing price itself rises with spam volume because spam occupies space, and it can never fall below the floor.

With the equilibrium in hand, the authors vary block capacity and the floor to trace out user welfare (the area under the demand curve above the clearing price), validator revenue, and network externality, comparing each against a spam-free counterfactual at the same parameters. They then repeat the ordering logic under priority fee ordering, where the block is split into sub-blocks with different clearing prices. Finally, they check the model's qualitative predictions against public data from Base and Arbitrum: spam rising with gas-target increases, falling when floors are raised, capturing a disproportionate share of marginal capacity, and appearing later in blocks.

## Why This Matters

**Impact on research.** Prior MEV work focused overwhelmingly on Ethereum Layer 1, where searchers precompute targeted, high-confidence transactions and builder auctions filter out failures before execution. This paper formalizes the qualitatively different regime that appears on cheap, fast chains, and connects it to existing results (the Mazorra et al. timing-game bound) while giving the community a tractable model for design rather than post-hoc description. It also reframes the Flashbots thesis that MEV limits scaling into a testable claim about marginal capacity allocation.

**Real-world applications:**

- **Setting block gas limits.** The B_plat and MMUS rules give operators a way to decide when scaling further mostly buys spam and externality rather than user throughput.
- **Setting minimum gas prices.** The model predicts a floor is the key lever in the slack-capacity regime, where the market-clearing price would otherwise approach zero while the network still pays to process every transaction.
- **Fee mechanism design.** Charging for gas limit rather than gas used (as Monad does) and priority-based ordering both raise spammer costs in the model.
- **Capacity planning for infrastructure operators.** The externality term quantifies why high transaction volume stresses indexers and full nodes even when fees are low — the dynamic behind Etherscan discontinuing free API access for Base.

**Industry relevance.** The paper documents that responses so far have been reactive and chain-specific: Base cut its gas target from 70M to 50M and later raised its minimum transaction fee citing spam reduction; Arbitrum set a floor well above 1 wei and recently doubled it as part of a fee mechanism overhaul; Aptos raised its minimum gas fee citing spam; Monad launched with a non-trivial floor and gas-limit-based charging. The framework turns these ad-hoc moves into a coherent design logic.

## Future Directions

- **Extending the analysis beyond linear demand and gas-limit charging.** The paper notes that the methods extend to exponential demand (Appendix C) and to mechanisms charging only for gas used (Appendix D); the full implications of those variants for the parameter-setting rules are open for elaboration.
- **Refining the priority fee ordering model.** The paper treats PFO as an approximate mechanism; a fuller treatment of ordering, sub-block structure, and interaction with builder behavior is a natural next step.
- **Empirically estimating the opportunity-size parameter r0 and its growth.** The demand-scaling conclusion depends on the finding that MEV opportunity size grows approximately linearly with non-spam activity; validating or falsifying that relationship across more chains would sharpen the plateau prediction.
- **Comparing mitigations.** Section 7 discusses mitigations, and the paper notes that encrypted mempools — by reducing the information available for targeted MEV — are likely to *increase* spam MEV rather than reduce it, an uncomfortable trade-off worth systematic study.

## Target Audience

Blockchain protocol designers and mechanism designers deciding block limits and fee floors; researchers in blockchain economics and game theory working on MEV and transaction fee mechanisms; Layer 2 and high-throughput chain operators evaluating fee policy; and infrastructure and indexing teams who bear the externality costs the model quantifies. Readers should be comfortable with equilibrium reasoning and marginal analysis; the empirical sections are accessible to data-oriented practitioners, but the core derivations are not beginner material.

## Authors’ abstract

On high-throughput, low-fee blockchains, a qualitatively new form of maximal extractable value (MEV) has emerged: searchers submit large volumes of speculative transactions, whose profitability is resolved only at execution time. We refer to this as spam MEV. On major rollups, it can at times consume more than half of block gas, even though only a small fraction of probes ultimately results in a trade. Despite growing awareness of this phenomenon, there is no principled…

[The full abstract on arXiv](https://arxiv.org/abs/2604.00234)

[Read the original paper](https://arxiv.org/abs/2604.00234)

---

Canonical: https://crypto.world/paper/blockspace-under-pressure-an-analysis-of-spam-mev-on-high-throughput-2604.00234
