One lane, and no way to say you can wait

A post on Ethereum Research this week makes an argument that is easy to state and awkward to answer: the economic value of a transaction depends on when it executes, transactions differ enormously in how that value behaves over time, and the protocol represents none of it.

The examples do the work. A liquidation loses most of its value within a slot or two. A treasury settlement holds steady for hours and then falls to zero at a cutoff. An oracle update is worth something as a function of how long it has been since the last one. Three different shapes of value against execution time, and all three arrive at the same queue.

What Ethereum offers is one homogeneous lane, one stream per twelve-second slot, and the only signal a transaction can send about time is a scalar fee. There is no slower and cheaper option with a deadline attached. A transaction that could comfortably wait either bids as though it were urgent, or bids low and accepts an open-ended wait with no guarantee at the end of it.

Two consequences follow, and the author's point is that neither is a shortage of blockspace. The first is that urgent demand ends up outbidding patient demand that is bidding as though it were urgent, because nothing in the fee distinguishes them. The second is that some demand is excluded rather than delayed: a transaction whose value decays slowly might be worth executing later at a lower price, but with no way to express that it must pay today's price or not transact at all. It drops out, and the drop-out is invisible in the fee data.

The framing matters more than any specific mechanism proposal, because it reclassifies a familiar complaint. Congestion that is read as insufficient capacity may instead be an absence of temporal differentiation — and adding capacity to a system with one lane and one price does not create the service that patient demand was looking for.