You send a swap to a decentralized exchange. The screen quotes you a price. By the time the transaction lands in a block, you got a worse one, and somebody you have never heard of is a few dollars richer.

Nothing broke. No contract was exploited, no key was stolen, no price feed was manipulated. The system did exactly what it was built to do. Somebody saw your order before it executed and chose what went in front of it.

So, is MEV front-running? In the narrow sense, yes. Maximal extractable value is the profit available from deciding which transactions enter a block and in what order, and its most lucrative forms involve trading ahead of an order that is already visible. Wall Street called that front-running, made it illegal, and fined firms for doing it.

Crypto called it MEV and built an auction for it.

As a former banker, what strikes me about MEV is not that crypto invented a new form of extraction. It did not. The mechanism is old enough to have case law. What is genuinely new is the conclusion crypto drew from it.

Traditional finance decided that the ability to see customer order flow and choose its sequence was a conflict of interest. It banned the conduct, pushed the order flow into private venues, and punished the firms that abused it. Crypto decided the same ability was a property right, and sold it in public to the highest bidder.

Neither approach made the money go away. Both relocated it, and both concentrated it in a handful of firms. What follows is what each system actually did, and the one thing crypto copied the mechanism without ever copying.

What Is MEV, Really?

The Ethereum developer documentation defines it plainly. MEV is "the maximum value that can be extracted from block production in excess of the standard block reward and gas fees by including, excluding, and changing the order of transactions in a block" (ethereum.org, accessed September 2026).

Read that definition again and notice what it does not say. It says nothing about hacking, nothing about bugs, and nothing about breaking rules. Every word describes ordinary block production.

MEV is not an attack on the system. It is a property of the system.

The raw material is the mempool, the public queue where pending transactions wait before confirmation. On a public blockchain that queue is visible to anyone. A pending swap on an automated market maker is not just a request. It is an announcement of exactly how the price is about to move, published to the world, seconds before it happens.

The same documentation names four categories where that information turns into money. DEX arbitrage, liquidations, sandwich trading, and NFT MEV.

Sandwich trading is the one that maps cleanly onto the old offense. A searcher sees your pending swap, buys the asset immediately before you, lets your trade push the price up, and sells immediately after. Your slippage is their revenue.

You are not the counterparty to a bad trade. You are the reason the trade was profitable.

The Ethereum documentation is not defensive about this. It describes MEV as genuinely mixed, stabilizing in some respects through arbitrage and liquidation, harmful in others through sandwich attacks, and a potential source of consensus instability. That is a fair reading, and more honest than most industry commentary manages.

What Did Wall Street Actually Ban?

The American answer to the same information advantage is written down, and it is narrower than people assume.

FINRA Rule 5270 states that "no member or person associated with a member shall cause to be executed an order to buy or sell a security or a related financial instrument when such member or person associated with a member causing such order to be executed has material, non-public market information concerning an imminent block transaction in that security" (FINRA, accessed September 2026). The SEC approved it in September 2012.

Two details matter for the comparison. First, the rule bites on information that is material and non-public. Second, it applies to an imminent block transaction, and the supplementary material generally treats 10,000 shares or more as a block.

Hold those two conditions against a public mempool and the whole framework stops working.

A pending Ethereum transaction is not non-public. It is broadcast, by design, to every node that will listen. The secrecy that makes front-running an offense in equities does not exist as a category on a transparent ledger. And the size threshold is close to meaningless where a $4,000 swap is worth sandwiching because the cost of trying is a few dollars of gas.

This is the part most comparisons get wrong. It is not that crypto looked at FINRA 5270 and decided to ignore it. It is that FINRA 5270 is built on an assumption, the privacy of the order, that a public blockchain deliberately discards. The rule does not so much get broken as fail to attach.

Why Did the SEC Fine Five Firms $241MM?

The enforcement precedent is more instructive than the rule, because it describes conduct rather than category.

In March 2004 the SEC settled with five New York Stock Exchange specialist firms, Bear Wagner Specialists, Fleet Specialist, LaBranche & Co., Spear Leeds & Kellogg Specialists, and Van der Moolen Specialists USA. The firms paid $241,823,257, made up of $87,735,635 in civil penalties and $154,087,622 in disgorgement, for conduct between 1999 and 2003 (SEC, March 2004).

What did they do? The SEC describes specialists "interpositioning" their own dealer accounts "between customer orders by trading into both of them in succession, for example, buying into a customer market sell order first, and then selling, at a higher price, into the opposite market buy order." They also "traded for their dealer accounts ahead of executable agency orders on the same side of the market, orders that were executed later at prices inferior to the prices of dealer account trades."

Strip out the vocabulary and you have described a sandwich. Insert yourself between a buyer and a seller who would otherwise have met, and keep the difference.

But the sentence that matters most is the one describing the duty. The SEC found the firms violated "their basic obligation to match executable public customer buy and sell orders and not to fill customer orders through trades from the firm's own account when those customer orders could be matched with other customer orders."

That is not a prohibition. It is an affirmative obligation. The specialist was not merely told to refrain from trading ahead. The specialist was required to match the public orders first.

The offense was failing to perform a duty, not only performing a trick.

Nothing in crypto's block production stack contains an equivalent sentence. A block builder has no obligation to match your order against the best available counter-order. It has no obligation to you at all. It is bidding for the right to sequence a block, and the bid is the entire relationship.

How Does Crypto Auction the Same Advantage?

Ethereum's answer to MEV was not prohibition. It was market design.

Under proposer-builder separation, the roles split. Validators still propose and attest to blocks, but a specialized class of block builders takes over ordering transactions. A builder assembles a block and bids for its inclusion, and the validator chosen to propose the next block picks the highest bid (ethereum.org, accessed September 2026).

The middleware that implements this describes itself without euphemism. Flashbots states that "mev-boost is open source middleware run by validators to access a competitive block-building market" (Flashbots, accessed September 2026).

A competitive market in block building is, precisely, a competitive market in transaction ordering. The thing Wall Street defined as a conflict of interest is here the product being sold.

The supply chain has three tiers. Searchers hunt for profitable opportunities and send bundles of transactions with a sealed-price bid. Builders aggregate those bundles into full blocks. Relays sit between builders and validators, validating blocks and passing the most profitable one along.

I want to be careful not to sneer at this, because the design solves a real problem. Before proposer-builder separation, extraction ran through priority gas auctions that congested the network and pushed the advantage toward whoever ran the fastest infrastructure. Routing it through an auction genuinely did reduce that congestion, and it returns value to validators, which strengthens staking economics across the network.

The design is competent engineering. My argument is not that it fails on its own terms. It is that its own terms never included a duty to the person whose order is being sequenced.

Was This Always the Plan?

No, and the founding document says so in its title.

The paper that coined the term is "Flash Boys 2.0, Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges," by Philip Daian, Steven Goldfeder, Tyler Kell, Yunqi Li, Xueyuan Zhao, Iddo Bentov, Lorenz Breidenbach, and Ari Juels, submitted in April 2019 (arXiv, April 2019).

The name is a direct reference to Michael Lewis's 2014 book about high-frequency trading in American equities. The authors were not being subtle. The abstract states that "like high-frequency traders on Wall Street, these bots exploit inefficiencies in DEXes, paying high transaction fees and optimizing network latency to frontrun, i.e., anticipate and exploit, ordinary users' DEX trades."

The word the paper uses is frontrun. Not arbitrage, not value capture, not ordering optimization. The people who named MEV named it after the offense.

They also issued a warning that had nothing to do with fairness. The paper argues that "high fees paid for priority transaction ordering poses a systemic risk to consensus-layer security." The concern was that if the value of reordering a block exceeds the reward for producing one honestly, the incentive to reorganize the chain itself becomes rational.

The abstract closes with a line worth sitting with. "Traditional forms of financial-market exploitation are adapting to and penetrating blockchain economies."

That sentence was published in 2019 as a warning. Seven years on, the adaptation it predicted has been institutionalized, given standard middleware, and turned into a revenue line for validators. The warning was not ignored. It was productized.

Is a Transparent Auction Better Than a Ban?

Here is the strongest version of the case against me, and it deserves a hearing.

Wall Street's ban did not eliminate the economics of order flow. It relocated them.

Retail orders in American equities largely do not reach a public exchange at all. They are routed to wholesalers who pay for the privilege, and the payment happens in private contracts an ordinary customer never sees. The conduct became illegal and the value kept flowing, just through a channel nobody outside the arrangement can audit.

Crypto's auction, by contrast, is visible. Anyone can watch relay dashboards, see which builders win, and measure what validators earn. There is no equivalent public dashboard for what a wholesaler pays for a retail equity order.

That is a genuine advantage, and I have made a version of this argument before about where stablecoin yield actually comes from. Banning a payment rarely stops it. It usually just decides whether the payment is visible in the data.

I take the point, and it moves me part of the way. Transparency is worth something real, and a system that lets you measure extraction is better than one that hides it.

But transparency is a property of the measurement, not of the obligation. Knowing precisely how much value was taken from your trade, by whom, and through which relay does not give you a claim against anyone.

The 2004 specialists operated in one of the most heavily surveilled venues in the world, on a physical floor, under an exchange's direct supervision. Visibility was never the missing ingredient. The duty is what made the conduct actionable, and the duty is exactly what did not carry over.

What the Relay Numbers Show About Concentration

There is a second pattern worth naming, and it is about who ends up holding the pipe.

A 24-hour snapshot of MEV-Boost relay activity on 19 September 2026 showed 13,055 payloads delivered across all relays, distributed as follows (relayscan.io, September 2026).

Relay Share of MEV-Boost payloads (24h)
relay.ultrasound.money 32.20%
bloXroute (regulated) 28.27%
Titan 26.84%
All others combined roughly 12.7%

Those top three add to 87.31% by my own arithmetic on the figures above. Three relays carry the overwhelming majority of what the documentation calls a competitive block-building market.

Two cautions about that table, because this is exactly where MEV commentary goes wrong. First, these are shares of relay payload deliveries, not shares of all Ethereum blocks. Multiple relays report delivering the same slot's payload, so the counts overlap and cannot be divided by the roughly 7,200 slots Ethereum produces in a day to yield anything meaningful. I am deliberately not printing a share-of-all-blocks number, because the public dashboards do not support one.

Second, this is a single day. Relay shares move, and this table will be stale within weeks.

What is durable is the shape rather than the level. In 2004, five specialist firms sat at the point where NYSE order flow had to pass, and the SEC fined all five. In 2026, three relays sit at the equivalent point on Ethereum. The technology inverted almost every assumption about who may see an order, and the concentration came out roughly the same.

That should not be surprising. Order flow intermediation has enormous economies of scale in any market structure, because the most profitable ordering decision requires seeing the most orders. Decentralizing the ledger did not decentralize that.

What to Watch

Three things would tell you which way this settles, and none of them requires waiting for a regulator.

The first is whether private order flow becomes the default. The obvious defense against being sandwiched is to stop broadcasting to the public mempool and submit through a private channel instead. That works, and adoption is growing.

It also reconstructs, step by step, the private order flow arrangement American equities arrived at by a completely different route. If most retail crypto trades end up routed privately to a builder, crypto will have spent a decade rebuilding the market structure it defined itself against. Watch the share of transactions that never touch the public mempool.

The second is whether any obligation appears anywhere in the stack. Not a regulation, a commitment. A relay or builder that publishes an enforceable policy on what it will not do with the orders it sees would be the first genuine analogue of the specialist's duty.

Best-execution commitments, no-sandwich policies with real consequences, and wallet-level routing guarantees are all candidate forms. The relevant question is whether any of them carries a remedy when breached, or whether they stay marketing.

The third is relay concentration itself. If the top three relays sit near 87% a year from now, the competitive-market description is doing less work than it appears to, and the systems that matter are a small number of firms whose internal policies are unaudited. Concentration is what turns an engineering choice into a governance problem, and it is measurable today by anyone who cares to look. If you are new to how this plumbing fits together, the DeFi primer covers the underlying mechanics.

Where I Land on the Duty

The trade at the top of this piece still happens thousands of times a day, and nothing about it is a malfunction.

My position is this. Crypto did not fail to ban front-running. It correctly recognized that the offense, as traditional finance defines it, depends on a private order that a public ledger does not provide, and it built a transparent auction instead of pretending otherwise. That was an honest response to a real constraint, and the auction is better engineering than the gas war it replaced.

What crypto did not do is build the other half. The 2004 settlement did not turn on secrecy. It turned on an affirmative obligation to match public customer orders before trading against them, and there is no sentence like that anywhere in the block-building stack.

Transparency tells you what happened to your order. A duty is what gives you a claim when the answer is bad.

Here is my falsifiable flag. I do not expect regulators to supply that duty, and I am not sure they can, because the conduct is not secret and the rule they would reach for does not attach. I expect it to arrive commercially, from wallets and routers competing on execution quality, the way UK hedge funds adopted a collateral re-use limit after Lehman without anyone legislating one. If three years from now the dominant wallets still compete on fees and speed alone, and none of them will say in writing what happens to an order between signature and inclusion, then the transparency argument will have been a consolation prize rather than an answer, and I will have been too optimistic about the market solving it.

Further Reading