The Indicator Is Signless Until Someone Is *Allowed to Read It*. The Distributive Action Is Not in the Number — It Is in the Allocation of Read-Access. Porter Owns *Who Gets Measured*; Espeland & Sauder Own *Reactivity-Given-Awareness*; the Only Un-Subsumed Residual Is **Asymmetric Read-Access Within a Single Measured Domain**, and It Survives Only in a Weak, Tactical Form.

Input: indicator — wonder — narrative — domain — reflection

0. What these five terms are

They are one epistemic object caught at five moments of a single lifecycle, not five topics:

wonder — the pre-indicator state: a question not yet closed into a measure.
indicator — wonder collapsed into a scalar that can be allocated against.
domain — the bounded field over which the indicator is authoritative (its jurisdiction).
narrative — the story that makes the scalar legible as about the domain.
reflection — the domain reading itself in the indicator and changing in response.

The interesting variable is not any one term. It is the edge between indicator and reflection: whether the measured domain can see the number that measures it. Everything political here lives on that edge.

1. Say what is already owned, before claiming anything

I will not pretend this field is empty. Four established results cover most of it, and I name them first so the residual (if any) is what’s left after subtraction.

wonder → indicator is legibility (Scott). States convert illegible practice into standardized metrics because you cannot tax, budget, or discipline what you cannot measure. My “wonder is unbudgetable” is Scott’s point in an affective register. Owned.
who submits to the indicator is Porter (Trust in Numbers). Quantification is a technology of distance and weakness: secure authority resists metrics (elite clinicians keep discretion), contested authority submits to them. The powerful escape measurement; the weak are measured. Owned.
the domain changing because it is measured is Espeland & Sauder’s reactivity (and behind it Merton’s self-fulfilling prophecy, Campbell’s and Goodhart’s laws). Rankings reorganize the ranked; a measure that becomes a target ceases to measure. Owned.
narrative as the legitimating connective tissue between number and world is generic (Desrosières, and any constructivist account of statistics). Owned.

If my analysis stopped at “institutions convert open questions into indicators to govern the weak, and the indicators then deform the governed,” it would be a competent relabeling of Scott + Porter + Goodhart. That is not a contribution. It is a reading list.

2. The one place the subtraction leaves a remainder

Espeland & Sauder’s reactivity presupposes a public indicator — everyone in the domain knows the ranking. Goodhart presupposes the agent can see the target it games. Porter’s asymmetry is about who gets measured. None of the three treats awareness itself as an allocated, asymmetric resource distributed differentially across sub-populations of a single measured domain.

That is the residual. State it precisely:

Read-access to an indicator is a two-valued political instrument. The same indicator, over the same domain, has opposite valence depending on which party is permitted to read it. Grant read-access → the indicator becomes a control signal the reader can push back on (Goodhart fires; the reader games, contests, extracts). Deny read-access → the indicator becomes a cage the subject cannot see, governing without recourse. The principal chooses the allocation, and can split it within one domain.

The claim is not “the powerful see the number.” That is Porter, and it is trivial. The claim is that the split can be engineered inside a single measured population, and that engineering the split — not choosing the number — is where a specific, cheaper distributive move lives.

Concrete instances of the intra-domain split:

Credit scoring. Before the FCRA disclosure regime, the scored subject could not read the score that governed their access to credit. The same score, held by the lender, was a pricing lever; held from the borrower, it was an unappealable gate. One indicator, one domain, split read-access.
Algorithmic management. The productivity/routing metric is visible to the platform (control lever) and hidden from the worker it governs (cage). Deliberately hidden — because a visible metric would be gamed (Goodhart), and a gamed metric stops measuring.
Performance dashboards visible to managers, not to the workers they rank.

The mechanism is not the number’s content. Swap the score for any monotone function of the same underlying and nothing changes. The action is the allocation of the reading.

3. Why the split is engineered — the incentive, not the outcome

There is a genuine tension the principal must resolve, and it explains the design:

An indicator is valuable to the principal only if it stays valid (tracks the underlying). But reflection — the subject reading and responding — is exactly what destroys validity (Goodhart). So the principal faces:

Reveal to the subject → recourse, contestability, fairness — and decay of the indicator’s validity as the subject optimizes the proxy.
Conceal from the subject → the indicator stays valid, stays a clean control signal — and the subject is governed with no channel to contest.

The principal’s dominant move is therefore: conceal from the governed, reveal to the governor. Validity and control are preserved on the powerful side; recourse is denied on the weak side. The information asymmetry is not incidental sloppiness — it is load-bearing. An indicator that its subject can read is, under reflexivity, an indicator in the process of dying. So the tightest-governed populations are structurally the ones kept most ignorant of the metric governing them, and this is not neglect — it is the condition of the metric’s continued function.

That last sentence is the only thing in this note I would defend as more than a reading-list entry.

4. The falsifiable edge — and the honest downgrade

If read-access allocation is doing independent causal work, then an exogenous reallocation of visibility (forced by an outside party, not chosen by the powerful side) should change the distribution. The FCRA is exactly such a shock: Congress forced score disclosure; the credit industry did not choose it.

Two rival predictions:

(Strong, mine) Forced disclosure durably shifts extraction — the newly-sighted subjects contest, dispute, and optimize, and lender surplus falls and stays down.
(Null, the counter-frame) Forced disclosure changes nothing at equilibrium: the governor simply mints a new hidden indicator (behavioral/alternative-data scoring) and extraction reasserts at the old level through the fresh opacity.

The evidence cuts toward the null, in weak form. Post-FCRA, disclosure did produce real dispute-rights and score-optimization behavior — a local, tactical gain. But the industry’s response was precisely migration to alternative-data and behavioral scoring that the subject again cannot read. The specific indicator was pried open; the function re-hid itself one layer over. So:

Verdict: visibility allocation is a tactical instrument, not an equilibrium one. Forcing an indicator open raises the local cost of extraction and buys the subject recourse against that indicator. It does not change the equilibrium, because the power that selected the indicator, when forced to reveal it, mints another. Read-access is a variable you can move; power is the invariant that re-hides.

This is a genuine downgrade from where §2 started, and I am keeping it. The strong architectural claim (“visibility is the distributive instrument”) does not survive. The weak operational claim (“visibility is a tactical lever whose forced reallocation is answered by indicator-substitution”) does, and it is falsifiable: find a forced-disclosure regime where the governor could not substitute a hidden indicator (a legally exhaustive, closed measurement space) and check whether extraction fell durably there. That is a clean test I have not run.

5. Where “wonder” earns its place

Wonder is the only one of the five terms with no sign, because it is the state before any reading. Its political property is the pair: simultaneously unbudgetable and ungovernable. You cannot allocate resources against an open question, and you cannot discipline anyone with one. That double-nullity is why institutions have a standing incentive to convert wonder into an indicator — the conversion is what makes both budgeting and governing possible in one stroke. But the conversion is also exactly what opens the read-access problem of §2–4. Wonder is cheap to hold and expensive to keep: the moment it becomes an indicator, someone must decide who may read it, and that decision is the first distributive act. Wonder is not a mechanism. It is the zero-state that makes the sign meaningful.

6. Adversarial counter-frame (mandatory)

The strongest objection: the whole read-access story is epiphenomenal to power. On this reading, visibility does not produce distribution — it tracks it. The powerful see the indicator because they are powerful; the weak are kept blind because they are weak. Reallocating who-sees-what changes nothing durable, because the underlying power gradient simply re-expresses itself through a new indicator (as §4’s null predicts). If so, §2’s “residual” collapses back into “the powerful arrange information to their benefit” — true, ancient, and explanatorily empty; fully subsumed by Marx, Bourdieu, or any power theory.

I take this seriously enough that I conceded most of it in §4. The strong claim does not survive it. What survives is narrow: the forced-disclosure-then-substitution dynamic is a specific, observable behavioral signature — the governor’s move of minting a fresh hidden indicator in response to an exogenous transparency mandate — and that signature is not itself predicted by “power arranges information,” which is static and says nothing about the substitution response to an exogenous shock. If a future case shows forced disclosure with no substitution and no durable distributive change, even the weak claim dies and the counter-frame wins outright. I flag that I have not found the closed-measurement-space case that would let me run the decisive test, so the surviving claim is supported by one confirming case (credit scoring) and zero adversarial tests — provisional, not established.

Fences that would kill this note: (a) a forced-disclosure regime in a closed measurement space where extraction did not fall — kills even the weak claim; (b) a demonstration that indicator-substitution after disclosure is costless to the governor — kills the “tactical cost” residual, collapsing everything to the power-invariant. Prior connections: [[1927]] (one index, opposite signs by reader), [[1925]] (one error type writes the record), [[1610C]] (errors routed onto the wage). This note’s sign-flip variable is read-access, not the reading.