The Slowest Interface in the Market

The Footprint Became a Ledger closed a three-memo thread by showing that each memo existed to shut the specific gap the previous one had left open on purpose. The five memos this piece reviews do not have that shape. They do not close gaps. They climb. Each one takes the claim the previous memo made and moves it up a level — from a single action, to a single resource, to a single firm's loop, to a market, to the mechanism inside the market — and each one carries the Steward up with it. That is the property worth being precise about, because the arc's discipline in raising the Steward's position is exactly what exposes the one thing it never specified: the channel through which a human reaches a position that high.

Reversibility replaced frequency as the gate

The Threshold Was Never Priced for Irreversibility opened the arc by finding a variable the existing threshold model had never scored. The Intervention Threshold calibrates by frequency — roughly one intervention per hundred executions at T1 — and frequency is the right variable when a wrong decision costs a retry. It is the wrong variable for anything that cannot be retried. The memo introduced the Physical Intervention Threshold, an Intervention Threshold calibrated by the reversibility of an action rather than by frequency alone — set by the cost of a single wrong action, not how often it occurs. It was careful to separate that from the Authorization Trap: the trap is an operator who cannot remove a gate that no longer serves a function; the Physical Intervention Threshold is a gate the task itself requires.

The evidence it used is worth recalling for what it shows about the Steward, not the agent. In the MHS deployment the memo cited, a compiled recovery routine ran 700 unattended trials, and the same agent paused for human confirmation on nearly every new physically risky action. The memo read both halves as one policy: release what is proven, hold what is not. What it did not ask is what the human on the receiving end of those confirmations was looking at, or how fast they could answer. The gate was specified. The person behind the gate was assumed.

Addressability, then execution, then allocation

From Agentic Commerce to Agentic Markets moved from the action to the resource. Its claim was narrow and it said so: not repricing, which has run for decades, and not visibility, which EDI and marketplace APIs have supplied nearly as long, but reallocation — a unit of inventory or an hour of capacity addressable by an agent that can commit against it. An Agentic Market is a market in which supply-side resources — inventory, production capacity, logistics capacity — are represented by agents that continuously price and reallocate them in response to changing conditions, rather than agents acting only on behalf of buyers. The memo was explicit that this is a projection, not an observation, and that every reallocation runs inside the gate #119 built.

Executable Economy moved from the resource to the loop. It removed the planning cycle and kept the exception gate, and it named the order of operations that keeps the removal safe: addressability first, calibration second, speed last. An Executable Economy is an economic system in which pricing, inventory, and production signals translate directly into physical and digital action, without a human planning cycle mediating between the signal and the response. The memo's own verdict drew the boundary: this does not mean production runs unattended. It means the plan stops being the thing the rest of the system waits for. Something else is still waited for. The memo did not say what.

When the Market Decides What Gets Made moved from one firm's loop to the market. Market Computation is the condition in which market allocation — what gets produced, not merely what gets sold — is performed by continuous computation across agents rather than by periodic, human-mediated price-setting and planning. This is the memo where the Steward's position visibly changed. Inside a single loop, the Steward resolves exceptions the threshold escalates. At market scale, the memo argued, the Steward must also govern the allocation rule that generates every commitment the threshold gates — a second stewardship surface above the first. It also recorded, honestly, that Arco has no term yet for the economic position of owning that rule. That is one gap the arc left open by name.

The mechanism, and what it was built to be legible to

Allocation Becomes Computational went one level deeper into #122 rather than one level higher, and it is the memo that makes the second gap visible. It separated price from allocation — two buyers can face the same price and still compete for the same 100 hours of line capacity — and it specified what a market must represent to compute an allocation rather than react to a price: what resource exists, what portion is available, who can claim it, under what conditions, and what commitments already stand against it. Market state, not market information.

Every item on that list is a requirement about legibility to agents. The resource must be addressable to an agent. The state must be machine-readable so an allocation rule can compute against it. The commitment must be recorded so the next agent's query returns a current position. The arc, taken whole, is a specification for making a market legible to the systems that run it. It is thorough. It is also one-sided, because the one participant in that architecture whose legibility it never specified is the Steward — the human every escalation in the arc terminates at, and the human #122 placed above the allocation rule itself.

The Steward moved up twice and the channel never moved

This corpus has said before that the interface matters. The Steward Cannot Govern What They Cannot See established that the Audit Surface is a design requirement, not a byproduct. The UI That Runs the Business made Steward Experience a discipline held to the same rigor as customer-facing product. The Steward's Blind Spot showed what disengagement from that surface does to the autonomy metrics that depend on it. Those three memos established that the Steward's interaction surface is load-bearing. None of them said what form it must take, or why one form and not another. They were written about a Steward governing one business's exceptions at that business's tempo.

This arc changed the tempo and the altitude without revisiting the surface. The market state #123 specified recomputes on every relevant change. The allocation rule #122 handed to the Steward generates commitments at agent latency across every participant. The gate #119 built escalates irreversible physical actions to a human who must answer before the action proceeds — and the more irreversible the action, the more the escalation is not a report to read later but a decision to make now. At each level the arc raised the Steward, the volume of state the Steward must perceive grew and the time available to perceive it shrank. Nothing in the five memos asks how a human, reading at human speed, takes in market state that changes faster than it can be read, or how that same human puts a decision back into a loop that is already running.

That is the gap. The arc made every other participant in the market fast. It left the Steward as the one interface still operating at the speed of a person reading a screen — and it placed that interface at the point in the architecture where the cost of a slow or blind decision is highest. Every memo in the arc assumed the Steward could see what the system just did and say what it should do next. None of them examined whether the surface through which the Steward does either was built for the position the arc moved the Steward into.

The Operator's Verdict

This chapter is complete as a projection of what a computed market requires from its resources, its loops, and its allocation rule. It is incomplete as a projection of what that market requires from the person who governs it. Five memos specified how a market becomes legible to agents. Zero specified how the market's own state becomes legible to the Steward at the tempo the market now runs, or how the Steward's judgment gets back into the loop before the loop has moved on.

An operator building toward any of what this arc projects should notice that the Stewardship Model's viability now depends on an interface question the model has never been made to answer. The gate, the loop, and the rule are all specified. The channel between the human and all three is not. That is where the next chapter begins.

Technology changes what the market can compute. The Steward's channel determines what remains governable.