Every ungoverned decision point charges rent, and it is paid in attention.

The recurring cost

An unsettled question is not answered once; it is answered every cycle forever.

Consider a question the hierarchy settles: whether a favourable expectancy read may lift deployment above what the gate permits. Left unsettled, that question arrives at every deployment decision in which the two readings disagree, which is not a rare configuration. Each arrival demands the same reasoning, reaches a conclusion of variable quality, and consumes attention that was needed for the questions genuinely requiring judgement that cycle. Settled in advance, it arrives once, is answered once, and thereafter costs a glance. The arithmetic of that difference over a year of cycles is not marginal, and it is invisible in any measurement of decision quality because the saving shows up as decisions that were never made.

Where the quality goes

Decisions made under pressure are worse, and the pressure is highest where it matters.

The second cost is that the recurring decision is not made under neutral conditions. It arrives at the deployment moment, with positions open, results fresh, and the market doing whatever it is doing — conditions under which recency weighting is strongest and the argument for expansion is most available. So an unsettled question is systematically resolved in the direction the moment favours, and the moment favours expansion after good stretches and contraction after bad ones. Settling it in advance moves the decision to a time when none of those pressures apply and when the person deciding is reasoning about the general case rather than about a particular cycle they have money in.

The compounding effect

Depletion is cumulative, so the last decisions of a session are the worst ones.

Attention spent on one resolved-in-principle question is not merely lost — it is unavailable for the rest of the session. A trading day contains a finite number of genuine judgement calls, and every ungoverned decision point competes with them for the same resource. The consequence is an asymmetry in when errors occur: the questions handled late in a session, after the negotiations have accumulated, receive less than they deserve. Removing recurring questions from the surface therefore does more than save the time they took. It protects the quality of every decision that comes after them, which are disproportionately the ones the system could not have made in advance.

Why the hierarchy is the instrument

Ranking is what converts a recurring negotiation into a lookup.

A rule alone does not do this. An instruction to be careful about expectancy during drawdowns still requires the operator to determine how careful, this time, given this reading — the negotiation persists in a smaller form. What removes it is a stated precedence: the gate outranks the tactical layer, so the tactical layer's enthusiasm has no route to the ceiling and there is nothing to weigh. The same applies to the tie-break direction when standing is equal, and to each layer's explicit list of what it may not override. Each of those converts a question that would otherwise be argued into a fact that is read. The hierarchy is not primarily a safety mechanism. It is a device for retiring questions.

What is deliberately left unsettled

Not everything should be retired, and the ones kept open are the valuable ones.

It is worth being clear that the goal is not to eliminate judgement. Trade selection stays discretionary. Whether a setup qualifies, which branch fits the structure, whether conditions favour participation at all — these are left open because they are where the operator's skill actually lives and because no rule could settle them well. The design decision is about allocation: retire the questions with defensible general answers, so that the finite supply of attention lands on the questions that genuinely require it every time. A system that governed selection and left deployment to judgement would have the allocation exactly backwards, and would feel considerably more free.

• An unsettled question costs its answer every cycle, not once.
• The recurring answer is composed under the conditions least suited to producing it.
• Retire the questions with general answers; spend the attention on the ones without.

The key idea

The benefit that never appears in the results is the one worth building for.

Governance is normally justified by the losses it prevents, which is real and measurable and probably not the largest effect. The larger one is that an operator running a settled system arrives at each cycle with most of the questions already answered and their attention intact for the few that are not. That advantage will never show up as an attributable improvement, because it consists of errors that were never made in decisions that were never taken. It is nonetheless the thing an operator notices first when they start running a hierarchy properly, and the thing they miss most if they stop.