The Two Jobs That Used to Be One

You hold the slot. A block proposal lands in your validator's lap, the clock is ticking, and every transaction sitting in the mempool (the public waiting room for unconfirmed transfers) is yours to arrange however you like. Before Proposer-Builder Separation existed, that arrangement was the job, and the person doing the arranging quietly had their thumb on the scale for every sandwich trade, every arbitrage, every limit order shoved to the back of the queue.

Proposer-Builder Separation, PBS, cuts that single job cleanly in two. Specialized builders assemble the block contents. The proposer, meaning the validator holding the slot, picks the most valuable block from a list of sealed bids and signs it. The proposer never inspects the transaction order before committing. That's the whole mechanism. The consequences, though, ripple into every corner of validator economics.

Why Ordering Is Worth Fighting Over

Transaction ordering is not a neutral technical detail. It's money.

A DeFi trade that moves a liquidity pool's price creates an opportunity for whoever gets placed immediately before or after it. That opportunity has a name: Maximal Extractable Value, MEV, which is the extra profit a block producer can extract purely by controlling sequence. Flashbots' MEV-Explore data puts cumulative extracted MEV in the hundreds of millions of dollars since tracking began, with single-block peaks sometimes exceeding $5 million during volatile sessions.

Under the old model, a validator with genuine technical sophistication could run her own MEV extraction, collecting that surplus on top of standard block rewards. One who couldn't just left it sitting there. The result was a bifurcated validator set: a small group of capable operators capturing significant extra yield, and everyone else effectively subsidizing them by ignoring the same opportunities.

PBS formalizes a separation that already existed informally through MEV-Boost, the middleware most Ethereum validators run today. Think of MEV-Boost as PBS implemented in the garage before the factory got involved: it replicates the same builder-proposer split outside the core consensus layer. Roughly 90% of Ethereum blocks, at various points since the Merge, have been built through MEV-Boost relays rather than by the proposing validator itself.

A Concrete Scenario

Take two validators: Alinta and Bashir. Identical hardware, identical amounts of staked ETH, the same number of proposal slots assigned over a year.

Alinta runs MEV-Boost and connects to three competitive builder relays. Bashir builds blocks locally, ordering transactions by fee only. In a slot where a large DEX arbitrage is sitting in the mempool, a builder inside Alinta's relay network spots it, constructs an optimal block that captures the arbitrage, and passes 80% of the surplus to the proposer as a bid. Alinta accepts the highest bid without ever inspecting the internals. Her effective yield for that slot is noticeably above the base reward.

Bashir's locally built block misses the arbitrage entirely. Standard priority fees. Same stake, same slot count, meaningfully different annual return.

That gap is the incentive PBS creates, and it's not subtle. Validators are pushed toward accepting external blocks because refusing them is expensive. The real competition shifts from the validator set to the builder market.

What Actually Gets Concentrated

Here is where the honest analysis gets uncomfortable, because PBS doesn't eliminate centralization pressure. It relocates it.

Under naive pre-PBS conditions, sophisticated validators centralized. Under PBS, sophisticated builders centralize. Competitive block building requires low-latency mempool connections, fast simulation of transaction bundle combinations, and enough capital to absorb failed strategies. Those resources are not evenly distributed, and the numbers show it. At various points, the top three builders have accounted for over 80% of blocks, according to Flashbots' published market-share data. That is a striking concentration in a system that holds decentralization as a core value, and anyone who glosses past it is doing you a disservice.

The relay layer adds a separate trust assumption worth naming. Relays sit between builders and proposers, verifying that a block's bid is honest without revealing the contents prematurely. A validator trusting a relay is trusting that relay not to lie about block validity or suppress bids strategically. Enshrined PBS, the version that bakes this logic directly into the protocol with cryptographic guarantees rather than social trust, is still in research and design phases.

So the realistic picture: PBS genuinely improves fairness across the validator set, compresses the yield advantage that technically sophisticated validators once held, and makes MEV revenue more predictable for everyone. It does not make the block production supply chain trustless. It trades one trust question for another, moving it from "do you trust your validator" to "do you trust the builder and relay ecosystem."

Anyone telling you PBS fully solves MEV is selling something. Anyone telling you it changes nothing is ignoring what the yield distribution data actually shows. The shift is real, the improvement is real, and the new attack surface is equally real. What PBS did was make the game more honest, not smaller.