MEV era

Mempools as dark forests

The block as a slot machine

Every pending transaction is a whisper in public. You sign a swap, it lands in the mempool, and for a few seconds it sits in line—naked, with slippage tolerances showing. MEV (maximal extractable value) is the profit squeezed from deciding who enters first, who gets sandwiched, who is ignored, and which liquidations hit the chain. The first wave of DeFi turned Ethereum into a casino of composable money-legos; the mempool became the lobby where hustlers took notes.

Searchers built bots that lived in this lobby, watching for mispriced trades, under-collateralized loans, and fat slippage on DEX swaps. They raced each other with priority gas auctions, jacking up fees to land before rivals. To an ordinary user, the result felt like a mysterious tax: worse prices than expected, failed transactions that still burned gas, and sudden spikes in fees whenever markets moved.

Sandwiches, liquidations, and ghosts

The classics quickly emerged. Sandwiches: a bot buys before you, you trade at a worse rate, the bot sells after—capturing the spread you thought was yours. Liquidations: a protocol lets anyone repay a risky loan and seize collateral at a discount; bots camp the mempool waiting for the slightest price dip to pounce. Arbitrage: two DEXes drift out of sync, and the first bundle to hit both pools banks the difference. None of this violated the rules; it exploited them.

The mempool earned the nickname “dark forest” because naïve transactions were hunted the moment they appeared. Intents leaked, wallets shrugged, and users paid. The drama was invisible unless you looked at block explorers: sudden inserts of strange transactions before yours, or your swap reverted while someone else’s copy went through.

Early chaos and the MEV vocabulary

By 2020, terms like “front-running,” “back-running,” and “time-bandit attacks” (re-orgs to steal value) left the niche and hit Twitter. Miners realized they could run their own bots or sell blockspace to the highest bidder. Researchers quantified the haul: tens of millions, then hundreds. MEV was no longer a curiosity; it was a line item on chain economics. Every player—miners, searchers, wallets, exchanges—had to decide whether to ignore it, fight it, or professionalize it.

“The mempool is a battlefield. If you show your cards, someone will play them before you.” — Dan Robinson, on the “dark forest”

Auctions, relays, and new chokepoints

Flashbots and private bundles

Enter Flashbots: an attempt to civilize the chaos. Instead of bots spamming gas wars in the public mempool, searchers could submit bundles privately to miners. Miners (and later validators) would run an auction: the bundle paying the most wins, without revealing the strategy to rivals. It reduced chain congestion, but it also formalized MEV as a marketplace. Your sandwich was now an official bid.

During the Merge, Flashbots shipped mev-boost: proposer-builder separation (PBS) in practice, if not yet in-protocol. Builders assembled blocks stuffed with MEV. Proposers (validators) picked the block that paid the most. It worked, but it created new trust anchors: centralized relays that could censor or fail.

Relays, censorship, and OFAC panic

When the U.S. Treasury sanctioned Tornado Cash in 2022, relay operators faced a dilemma: include sanctioned transactions and risk legal fallout, or filter them and risk protocol-level censorship. Some relays chose to filter. For a stretch, more than half of Ethereum’s blocks came from relays that excluded certain transactions. The community panicked: had they traded mempool chaos for a quieter, more censorable choke point?

Alternative relays emerged, builders diversified, and client teams debated enshrined PBS to remove trusted relays from the critical path. The episode left a scar: making MEV tidy can also make it fragile if too few hands control order flow.

PBS debates and the builder bottleneck

Proposer-builder separation solves one problem—bribery pressure on proposers—but creates another: centralization of builders and relays. If only a handful can construct profitable blocks, they become kingmakers. Who watches them? How are they paid? Can they grief proposers by withholding blocks? Ideas sprouted: enshrine PBS in the protocol, require open-source relays, or add cryptographic commitments so builders can’t bait-and-switch.

Meanwhile, other chains experimented: Solana’s Jito added auctions inside validators; Cosmos chains explored MEV auctions native to Tendermint derivatives; rollups weighed whether to allow private mempools. Everywhere, the trade-off rhymed: visibility and competition vs. privacy and control.

“You can’t delete MEV. You can only decide who captures it, and how loud it is.” — Flashbots research call

Users fight back and the future of order

Wallet shields and intent layers

Users and wallets refused to stay prey. MEV-protect RPCs sent trades straight to trusted builders, skipping the public mempool. Simulations and “what you see is what you pay” warnings made slippage explicit. CowSwap and 1inch’s Fusion used batch auctions to clear trades together, making sandwiches harder. Intent-based systems asked users to state goals (“swap X for Y at best price”), then let solvers compete to route them, turning the dark forest into a guided tour.

Even L2s inherited the fight. Optimism and Arbitrum experimented with their own MEV policies; some rollups explored encrypted mempools or sequencing auctions. The message: if you want users, you have to own your order flow story.

Who keeps the spoils?

Once MEV became a revenue stream, everyone argued over its rightful owner. Validators saw it as their yield. Searchers said they did the work. Public-goods advocates proposed “MEV burn” to funnel proceeds into protocol funding. Wallets flirted with rebates to users whose order flow generated profit. The argument was less about morals and more about bargaining power: who can withhold order flow, and who can enforce a split?

MEV also collided with decentralization ideals. If a few builders capture most MEV, their incentives could skew protocol decisions. Conversely, spreading MEV to many actors might fragment security or encourage spam. There was no clean answer—just a slow, public negotiation over who gets paid for arranging the puzzle pieces of a block.

Encrypted futures and quieter rails

Looking forward, researchers chase “MEV minimization” rather than abolition. Encrypted mempools (via threshold encryption) could hide transaction details until they’re sealed, blocking sandwiches but introducing latency and key-management risks. Enshrined PBS could delete the relay trust point while keeping auctions. Projects like SUAVE (a Flashbots experiment) explore cross-domain order flow markets, where solvers compete across chains and rollups to give the best execution without leaking intent.

None of this makes MEV disappear. It merely decides where the noise lives. The era proved a bleak but useful truth: blockchains are markets for ordering. Pretending otherwise leaves users footing the bill in the dark forest. Designing for it—audited auctions, open data, user protections—turns the tax into an explicit line item instead of a lurking monster.