💻 computer science

Slow and Steady: Preventing MEV with Verifiable Delays

This paper proposes a defense mechanism against Maximal Extractable Value (MEV) in distributed ledgers by enforcing verifiable delays on transaction generation, demonstrating its effectiveness in both Byzantine and game-theoretic models while outlining its theoretical limitations and practical feasibility based on historical data.

Original authors: Zeta Avarikioti, Dimitris Karakostas, Karl Kreder, Shreekara Shastry

Published 2026-08-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Zeta Avarikioti, Dimitris Karakostas, Karl Kreder, Shreekara Shastry

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where digital money isn't just numbers in a bank account, but a giant, public ledger that everyone can see and update. This is the world of blockchain, the technology behind cryptocurrencies like Bitcoin and Ethereum. Think of it as a massive, shared notebook where every transaction is written down in a new "block" of pages. To keep this notebook honest, a special group of people (or computers) called "block creators" take turns writing these new pages. They are the referees of the game.

However, in this high-speed digital race, a sneaky problem has emerged called MEV, or Maximal Extractable Value. Imagine a race where the person running the track (the block creator) can peek at the runners' plans before they start. If they see a runner about to buy a rare item, the referee can jump in, buy it first, and sell it to the runner for a higher price, pocketing the difference. In the blockchain world, this means the person writing the block can rearrange transactions to make a quick profit, often at the expense of regular users. It's like a referee rigging the game to win a bet on the side. The big question for scientists and developers is: How do we stop the referee from rigging the game without slowing the whole game down to a crawl?

This paper, titled "Slow and Steady: Preventing MEV with Verifiable Delays," proposes a clever solution inspired by the old fable of the Tortoise and the Hare. The authors, Zeta Avarikioti and her team, suggest that sometimes, slowing things down is exactly what you need to win. They introduce a mechanism that forces a "verifiable delay" on every transaction. Think of it like a magical hourglass that a user must flip before their transaction can be accepted. This hourglass takes a set amount of time to run out, and crucially, no one can speed it up, even if they have a supercomputer.

Here is how the magic works: When a user wants to send a transaction, they don't just shout it out to the network. Instead, they first put the transaction into a "locked box" (a cryptographic commitment) and then run it through a special puzzle called a Verifiable Delay Function (VDF). This puzzle is designed so that it takes a specific amount of time to solve—say, 30 minutes—and there is no way to solve it faster, no matter how many computers you throw at it. Only after the puzzle is solved can the transaction be revealed and added to the blockchain.

The paper finds that this delay creates a powerful shield against MEV. Here is the logic: If a block creator sees a transaction that looks like a money-making opportunity (like a chance to "sandwich" a user's trade), they would normally try to jump in and rearrange the order to steal the profit. But with this new system, by the time the block creator figures out the opportunity and tries to build a new block to exploit it, the original transaction has already been locked in and finalized because of the mandatory delay. The block creator is too slow to react. The paper proves mathematically that if the delay is set longer than the time it takes for a transaction to normally get confirmed, the "rigger" simply cannot win.

The authors also looked at the "game theory" side of things, asking if rational people would still try to rig the game. They found that if everyone follows the rules, no one has an incentive to break them because the delay makes it impossible to profit from rigging. However, they also identified a "negative" scenario: if everyone decides to collude and ignore the rules together, they could still cause problems, but this is a general issue with any system, not just this one.

Crucially, the paper doesn't just offer theory; it checks the real world. The authors analyzed historical data from Ethereum and other networks. They found that while some MEV opportunities are huge (millions of dollars), the vast majority—over 99%—are actually quite small, often just a few dollars or even less than a dollar. This suggests that a delay of about 30 minutes (which is roughly twice the normal safety time for Ethereum) would be enough to stop almost all of these small, annoying scams, specifically those that are non-predictable.

The paper concludes that this "Slow and Steady" approach is a realistic defense. It suggests that we can make blockchains more fair and user-friendly by accepting a tiny bit of slowness. However, the authors are careful to note that this isn't a magic wand for everything. It works great for things like registering an NFT or sending money, where waiting 30 minutes is fine. But for high-speed trading on a decentralized exchange, where prices change in seconds, this delay might be too long and could actually hurt the users it's trying to protect. The paper suggests that this tool should be used wisely, perhaps only for specific types of applications where speed isn't the most important thing.

In short, the paper argues that by forcing a "wait time" that cannot be rigged, we can level the playing field. It turns the race from a sprint where the fastest (and sometimes the sneakiest) wins, into a steady march where everyone arrives at the same time, and the referee can't rig the finish line.

Technical Summary: Slow and Steady – Preventing MEV with Verifiable Delays

Problem Statement

The paper addresses the threat of Maximal Extractable Value (MEV) in distributed ledgers. MEV refers to the ability of block creators (miners or validators) to profit by injecting, reordering, or reorganizing transactions within a block. Common MEV strategies include front-running, back-running, and "sandwich" attacks, where a malicious actor exploits a user's transaction to generate profit at the user's or application developer's expense.

The authors identify two primary negative impacts of MEV:

Asset Drain: Value flows from users to block creators, degrading system usability.
Security Risks: MEV creates off-chain incentives that are not controlled by the ledger protocol. If these incentives are sufficiently high, profit-driven parties may deviate from the protocol to capture MEV, potentially jeopardizing consensus security and liveness.

Existing countermeasures, such as MEV auctions (e.g., Proposer-Builder Separation), rely on trusted intermediaries. Time-based ordering solutions often struggle to guarantee fairness without complex mechanisms, and content-agnostic approaches (commit-reveal schemes) delay state computation and obscure transaction content, preventing block creators from calculating expected rewards rationally.

Methodology

The paper proposes a defense mechanism that enforces a Verifiable Delay on transaction generation. The core intuition is to slow down the ability of a block creator to react to a newly observed transaction. If the delay is sufficiently long, an honest transaction will be finalized before an adversary can construct a valid, front-running transaction that depends on it.

The Mechanism

The authors propose a transformation of a distributed ledger's transaction validity predicate using three components:

Verifiable Delay Function (VDF): A cryptographic primitive requiring a specific number of sequential steps ( $\tau$ ) to evaluate, which cannot be significantly sped up by parallel processing.
Commitment Scheme: A cryptographic commitment to the transaction payload.
Validity Transformation: A new validity predicate, $\text{Validate}_{VDF, CS}$ $Validate_{V D F, C S}$ , is defined. A transaction $tx = \langle m, (y, \pi) \rangle$ $t x = ⟨ m, (y, π)⟩$ is valid only if:
- The underlying payload $m$ satisfies the original ledger's validity rules.
- The VDF output $y$ and proof $\pi$ are verified correctly.
- The revealed payload matches the commitment used for the VDF evaluation.

Crucially, the VDF is evaluated on the commitment of the transaction payload, not the payload itself. This allows users to outsource the computation to third parties without revealing the transaction content, preventing the service provider from launching a parallel MEV attack.

Theoretical Models

The analysis is conducted under two models:

Byzantine Model: Assumes an adaptive, rushing adversary controlling a subset of parties. The goal is to prove Blockchain Input Causality: an adversary cannot create a valid transaction that meaningfully depends on an honest transaction before that honest transaction is finalized.
Game Theoretic Model: Assumes rational parties maximizing utility (block rewards, fees, and MEV). The analysis utilizes $\epsilon$ -Nash equilibrium and compliance (where parties do not deviate to perform specific infractions like censorship).

Key Contributions and Results

1. Blockchain Input Causality (Theorem 1)

The paper proves that if the VDF delay parameter $\tau$ is greater than the ledger's liveness parameter $u$ (the time it takes for an honest transaction to be finalized), the transformed protocol guarantees input causality.

Logic: An honest transaction broadcast at round $r$ is finalized by round $r+u$ . An adversary attempting to front-run this transaction must first observe it, compute the VDF (taking $\tau$ rounds), and then publish. Since $\tau > u$ , the adversary's transaction cannot be valid before the honest transaction is already finalized.
Result: The transformation preserves the original protocol's safety and liveness guarantees while eliminating the ability to front-run.

2. Equilibrium Preservation (Theorem 2)

The authors demonstrate that if the original protocol $\Pi$ is an $\epsilon$ -Nash equilibrium in the absence of MEV ( $E_{-MEV}$ ), the transformed protocol $\tilde{\Pi}$ remains an $\epsilon'$ -Nash equilibrium even when MEV opportunities exist ( $E_{+MEV}$ ).

Significance: This provides a constructive method for protocol designers to secure rational ledgers against MEV without breaking the equilibrium properties of the underlying system.

3. Compliance Against Censorship (Theorem 4)

The paper shows that if the original protocol is compliant regarding censorship (parties do not censor transactions to gain utility) in a world without MEV, the transformed protocol is compliant regarding blockchain input causality in a world with MEV.

Logic: To claim MEV, a party must censor a transaction for $\tau$ rounds. Since $\tau > u$ , censoring for this duration violates the liveness guarantee of the underlying protocol. Since the adversary cannot unilaterally break liveness, they cannot successfully censor long enough to claim MEV, rendering the deviation unprofitable.

4. Negative Result (Theorem 3)

The paper also identifies a "negative equilibrium" where all parties collude to censor a transaction for $\tau$ rounds to claim its MEV. In this specific scenario, no single party has an incentive to deviate and publish the transaction, as they would be censored by the rest of the network. This highlights a limitation: the mechanism prevents individual front-running but cannot prevent collective censorship attacks if the network consensus shifts to exploit MEV.

Implementation and Practical Analysis

VDF Candidates

The authors evaluate two primary VDF candidates:

Wesolowski's VDF: Shows high efficiency for the proposed use case. Verification time remains low (~700ms) even for computation times ranging from 15 to 60 minutes. Proof sizes are manageable (<516 bytes).
Pietrzak's VDF: Verification time is logarithmic relative to computation time, making it less suitable for this specific application where fast verification is required for all nodes.

Historical MEV Data

Analysis of historical data from Layer-2 protocols (Polygon, Arbitrum, Optimism) and Flashbots datasets reveals:

While maximum MEV profits can reach millions of dollars, the overwhelming majority of opportunities are small (often < $100, and frequently < $1).
Claiming these opportunities often requires complex, multi-step transactions (e.g., 4-6 transfers for arbitrage).
Conclusion: A delay of 15–30 minutes (feasible with current VDFs) would effectively neutralize the vast majority of existing MEV threats, specifically non-predictable ones, as the cost and complexity of the delay outweigh the small profits of most opportunities.

Usability Considerations

The paper acknowledges that introducing delays impacts usability.

Suitable Applications: Scenarios not requiring immediate market response, such as NFT minting or registration.
Unsuitable Applications: The mechanism is not suitable for defending against MEV attacks that rely solely on back-running, as its focus is on front-running attacks. Additionally, it may not be suitable for decentralized exchanges (DEXs) and applications sensitive to price movements where users need to respond quickly to market conditions.
Deployment Flexibility: The mechanism can be implemented at the smart contract level rather than the ledger protocol level, allowing it to be applied selectively to specific DeFi applications.

Significance and Claims

The paper claims to offer a trustless defense mechanism against MEV that relies solely on the security of the ledger and cryptographic primitives, avoiding the need for trusted third parties or complex ordering protocols.

Theoretical Contribution: It establishes a formal link between VDF delays and the prevention of input causality violations, proving that standard security properties (safety, liveness, equilibrium) can be preserved or enhanced in the presence of MEV.
Practical Contribution: By analyzing historical data, the authors argue that the mechanism is realistically implementable and sufficient to prevent most non-predictable MEV cases, as the "slow and steady" approach effectively filters out low-value, high-frequency attacks.
Limitations: The authors modestly note that the mechanism relies on practical security estimates of VDFs (as rigorous theoretical lower bounds for efficient evaluation are still lacking) and is not a panacea for all DeFi applications, particularly those requiring high-frequency trading or those vulnerable solely to back-running.

In summary, the work proposes that by intentionally slowing down transaction finalization via verifiable delays, distributed ledgers can become more robust against extractable value attacks, trading a degree of immediacy for enhanced security and fairness.