Two posts back, this series covered what an EIP actually is — a design document, adopted or ignored by client teams, no vote involved. This post covers the most famous one ever shipped. When someone says "the EIP" without a number attached, they usually mean this: the proposal that rewrote how Ethereum prices its own block space. It went live in the London hard fork on August 5, 2021, and it remains the largest change to Ethereum's economics in the network's history.
EIP-1559 is Ethereum's transaction fee mechanism. It replaced a blind auction with a protocol-computed base fee that every transaction must pay — and that fee is destroyed, not paid to anyone. Users add a small tip for the block proposer on top. That's the whole design in two sentences, but the reasons behind each choice are where the interesting mechanics live.
One misconception is worth killing immediately, because it shaped years of disappointed commentary: EIP-1559 was never designed to make gas cheap. Fees on Ethereum are a function of demand for limited block space, and no fee mechanism changes that. What it changed is how the price gets discovered — replacing guesswork with a posted price, and systematic overpayment with a refund.
The Problem: A Blind Auction
Before London, Ethereum priced transactions with a first-price auction. You attached a gasPrice bid to your transaction, and miners — this was pre-Merge — filled blocks with the highest bidders. Simple, and bad in practice.
The problem with first-price auctions is that you're bidding blind. You can't see the whole mempool, you don't know what everyone else is bidding, and the penalty for guessing low is a transaction that sits for hours. So wallets padded their estimates, users overpaid to be safe, and fee estimation became a genuinely hard prediction problem that every wallet solved badly in its own way. Auction theory has a name for this failure mode — first-price auctions push bidders to overpay relative to the true clearing price — and Ethereum was running one every twelve seconds.
The Mechanism
EIP-1559 restructured the market around three moving parts.
The base fee is computed, not bid. Every block has a base fee determined entirely by protocol rules from the block before it. Blocks became elastic: the gas limit roughly doubled, and a target was set at half of it. If the previous block was fuller than the target, the base fee rises — capped at 12.5% per block. Emptier, and it falls by the same logic. Sustained congestion compounds that 12.5% exponentially, so the base fee climbs fast enough to price out excess demand within a few dozen blocks. The result is a posted price: your wallet doesn't guess what to bid, it reads the current base fee off the chain.
The base fee is burned. This is the design decision people fixate on, and the reasoning is easy to miss. If the base fee went to the block proposer, the proposer would have an incentive to manipulate it — stuff blocks with their own junk transactions to push the fee up, or quietly refund users off-chain to undercut the mechanism. Burning the fee makes it incentive-compatible: nobody collects it, so nobody profits from gaming it. The burn exists to make the price credible, and the supply effects — several million ETH destroyed since 2021 — are a side effect of that security argument, not the goal. Whether that side effect makes ETH deflationary depends on conditions covered elsewhere.
The tip pays for inclusion. With the base fee burned, proposers would have no reason to include transactions at all, so users attach a priority fee — a tip — that the proposer keeps. Under normal conditions it's small, a fraction of a gwei to a couple of gwei, because it only has to beat the other transactions waiting, not win an auction.
From the user side, this arrives as the type-2 transaction format: you set a maxFeePerGas (the most you'll pay in total) and a maxPriorityFeePerGas (your tip), and the protocol charges base fee plus tip and refunds the difference. Overpayment stopped being the default failure mode.
Where the Constraints Live
The hard constraint: EIP-1559 is a pricing mechanism, not a scaling mechanism. When a popular mint or a market crash floods the network, the base fee does exactly what it's designed to do — rise until enough people give up. Fees during those windows are still brutal. The mechanism changed how the price is discovered, not how high it can go. Why fees get high in the first place is its own topic.
There's also a lag built into the 12.5% cap. The base fee takes blocks to catch up with a sudden demand spike, and during that gap, inclusion briefly becomes a tip auction again — first-price behavior reasserting itself in bursts. The design accepts this: the cap is what makes fees predictable block-to-block, and short auctions during shocks are the price of that predictability.
And the mechanism prices exactly one resource — gas — in one dimension. Computation, calldata, and storage growth all cost different things to the network, but they're bought with the same unit. That flattening is a known limitation, and it's where the current design pressure is.
What's Changing
The clearest signal that the mechanism worked is that Ethereum keeps copying it. When EIP-4844 added blob space for rollups in 2024, it didn't invent a new fee model — it launched a second, independent 1559-style market, with its own base fee, its own target, and its own burn. Ethereum now runs two of these markets side by side, and the direction of research is more: separate 1559-style pricing for compute, data, and state, so a surge in one resource stops repricing the others. Multidimensional fee market proposals exist in draft form; none is scheduled.
The other live variable is the gas limit itself. Validators signaled it up from 30 million to 36 million in early 2025, with further increases under active discussion — and because the target is defined as half the limit, every increase moves the fee mechanism's equilibrium with it.
What Would Confirm This Direction
Additional resources getting their own 1559-style markets in future forks. The blob fee market continuing to function as blob demand grows. Tips remaining a small fraction of total fees outside genuine demand shocks — that's the observable evidence that the posted price, not the auction, is doing the price discovery.
What Would Invalidate or Change It
Sustained periods where tips rival or exceed the base fee under ordinary load would mean the mechanism is failing at its one job. Documented proposer strategies that profitably manipulate the base fee would break the incentive argument the burn rests on — this has been studied since before launch, and nothing practical has emerged, but it's the standing threat model. A future fork replacing the burn would change the economics outright.
Timing Perspective
Now: EIP-1559 is the water Ethereum swims in — every wallet, every L2 fee estimator, every transaction. The checkable facts are all on-chain: the current base fee, the block-by-block adjustment, the cumulative burn.
Next: Gas limit increases and the blob market's behavior under load. Both shift the mechanism's operating point without changing its logic.
Later: Multidimensional fee markets. Direction is broadly agreed in research circles; nothing is scheduled.
Boundary Statement
This post explains the fee mechanism — what EIP-1559 changed and why the burn exists. It is not an argument about ETH's value; a shrinking supply is not a price thesis, and nothing here supports one. It also isn't fee-optimization advice. The mechanism works as described; what it means for any particular participant is outside this scope.
