The framework already insists that no metric may price a person’s worth — capacity is measured, standing is not — and that a model may reveal possibility without adjudicating it. But it also cannot avoid measuring. To see standing eroding, power concentrating, leverage decaying, and irreversibility approaching, the ecology needs instruments. That is precisely the difficulty: every measurement it builds creates a new surface for classification, control, gaming, and jurisdiction. Measurement is not neutral infrastructure. It is a governance intervention — and the question is never should the ecology measure? but what kinds of measurement are compatible with standing and correctability, and what kinds quietly convert participants into administratively legible objects?
Measure what the decision needs, and no more
The first line is narrow. Measure only what a specific legitimate decision actually requires, and prevent the resulting measurement from acquiring authority outside that decision — measurement inherits the jurisdiction of the decision it serves, which is minimum necessary jurisdiction applied to data. “Measure function, not worth” is the right instinct but not enough on its own, because a functional score becomes a worth score the moment it travels. A note that answers can this person safely manage this medication? is legitimate for that decision; the same note, spread across housing, employment, insurance, and policing, has silently become how competent is this human being? That is the ontological leap the framework has to stop.
A hierarchy of measurement risk
Not all measurement carries the same danger, and the framework should be enthusiastic about some of it. Four tiers, from safest to most dangerous. System conditions — concentration, viable alternatives, appeal duration, leverage decay, cross-border burden — take the systemic state as their object and should be measured aggressively; this is what ecological accounting is for. Institutional performance — how often appeals overturn a body’s decisions, time-to-placement, whether emergency powers are renewed rather than allowed to sunset — exists to make power legible. Participant functional state — decision capacity, health status, accommodation need — is genuinely dangerous and carries the highest burden of justification. And participant ontology — worth, trustworthiness, dangerousness-as-identity, generalized “risk” or “capacity” — is the tier the framework refuses outright. The closer a metric comes to answering what kind of person is this?, the more dangerous it is. The discipline in one line: measure structures aggressively, measure persons reluctantly, and measure worth never.
Measure power more than vulnerability
This is the deepest principle the measurement problem yields, and it runs against the ordinary grain of institutions. The more power an entity holds over others, the more legible its exercise of that power should be; the more vulnerable a participant is to that power, the stronger their presumptive claim to opacity. Today the pattern runs backwards: an applicant for help must expose income, health, family, relationships, and history, while the institution deciding their case stays opaque about how the decision is made, what model scored them, its error rates, and who received their data. The framework reverses that. Standing should not require maximal legibility to power; power should require substantial legibility to those over whom it holds jurisdiction.
It has to be said plainly that this principle is not self-enforcing. It inverts the natural gradient of power — the institutions that should be most legible are exactly the ones with the most leverage to resist audit, and the participants who should get opacity have the least power to withhold themselves. So asymmetric legibility is not something a doctrine can grant; it is a norm the enforcement architecture has to win, and it fails wherever leverage over the powerful has already decayed.
The unmeasured domain
Some things must be able to exist before they are formalized, disclosed, or scored. Private belief, intimate life, emotional life, political thought, spiritual life, aspiration, and unacted-on dissent are presumptively unmeasured — not because they do not matter to the health of an ecology, but because requiring their continuous disclosure would colonize the very interiority the framework exists to protect. If every interior state has to become data in order to count, measurement has swallowed the interior.
This also resolves an apparent tension. The framework treats interiority as the ecology’s distributed sensing layer — the system genuinely needs signal from interior states in order to steer. The reconciliation is that the ecology senses through signal participants choose to emit, in voluntary and usually aggregate or anonymized form, never through compelled disclosure or surveillance. And the protection has a hard boundary: the unmeasured domain shields the participant, never the office. The same categories — belief, deliberation, private reasoning — claimed by a powerful actor are no shield against legitimate audit of how that power is used. A person’s political thought is protected; an institution’s decision rules are not, even when dressed as internal deliberation.
Metrics travel — and travel is where they turn dangerous
A measurement rarely becomes an identity all at once. It does so along a path: measurement → classification → portability → gatekeeping → identity. A bounded, legitimate metric becomes a computational identity when it can travel across domains until the participant can no longer escape its classification — a single score that at once governs housing, employment, finance, healthcare, and mobility has stopped being an instrument and become a verdict on the person. The dangerous step is portability, so the safeguard sits there: a metric’s move into a new domain requires new justification. A medical assessment cannot become employment data merely because it already exists — new domain, new jurisdiction. This is non-cascading standing applied to data: a measurement, like an adjudication, binds only its own domain, duration, and parties, and does not propagate into the rest of a life.
Instruments, not passports
An instrument helps you see something; a passport decides whether you may pass. Metrics drift from the first toward the second, and the constant question a healthy ecology asks is whether a given measurement has quietly become a gate. Once it has, it needs the full adjudicative safeguards — and some metrics, such as personality profiles, loyalty scores, inferred political belief, and predicted future social value, should never become gates at all. Two rules keep the drift in check. Authority is inversely related to scope: a narrow construct can legitimately carry weight, while a broad one that compresses far more reality should carry far less. And once a measure becomes a target it ceases to be a good measure, which is why plural, disagreeing indicators are safer than one canonical number — the framework already prefers a topology over a single score and treats capacity as a compass, not a scoreboard.
What a consequential metric must carry
Before any operative metric earns its place, it should have to answer: what decision does it enable, whose power does it increase, what is the minimum information the decision actually needs, may it travel and should it, how long does it persist, can the participant inspect and challenge it, can another measurement disagree, what happens when it is wrong — and, above all, does it describe a condition or define the participant? If the latter, it stops there. A metric that survives those questions then carries its safeguards with it: an explicit domain limit, an expiration, a bar on secondary use, participant access and correction, independent appeal, and its own uncertainty, because a number stripped of the uncertainty that produced it is dangerously portable and falsely authoritative. And its authority decays — an assessment produced under conditions that have since receded may persist as history while losing its current force over a decision, exactly as jurisdiction itself expires.
The doctrine
No Wasted Geometry measures conditions necessary for governance, not persons for the sake of total legibility. Measurement should be decision-scoped, minimally sufficient, time-bounded, uncertainty-preserving, challengeable, and prevented from acquiring cross-domain authority. Structural and institutional power should be more observable than participant interiority. No composite metric may substitute for judgment about standing, and no metric may determine a participant’s intrinsic worth.
The framework refused, from the start, to let a metric price a single life. This is that refusal extended to the whole enterprise of measurement: watch the structures closely, hold the powerful to account, and leave the person room to exist before being scored.
Measure power, not people — and where people must be measured, measure only what the decision needs, only for as long as it needs it, and never in a way the person cannot escape.