Method reference

What this framework asks instead of the usual debates

The usual debate asks questions like: Is the system biased? Did leaders mean well? Is the policy efficient? Does the right exist on paper? Are the ethics principles strong enough? Those questions matter, but they often stall because they stay too close to official language and too far from lived conditions.

This framework asks a more structural question: what does a person have to carry, prove, endure, or risk in order for the arrangement to keep functioning as advertised? That shift moves attention from stated permission to usable options, from values language to actual constraint, and from isolated bad actors to recurring patterns.

Those three moves make a method for reading institutions through burden, livability, recourse, delay, standing, and maintenance rather than through public justification alone.

The failure-behavior claim

Beyond who controls them, what rules they announce, or whether people may object, institutions should be evaluated by their failure behavior. Ask where error is absorbed, what it costs the absorber, whether that error counts as evidence, and which confirmed failures can actually limit or end power.

The claim holds even after conceding that nearly every part of it has a precedent. One structural question shows up in administrative burden, bogus consultation, automated denial, heroic frontline labor, investigations that never conclude, institutional apologies, model errors, memorialization, procedural exhaustion, and regulators captured by the entities they regulate. When the system is wrong, what has to give: the institution’s claim to continue, or the person underneath it?

A framework can make those cases structurally related and say what institutional change would distinguish a real remedy from a cosmetic one. Doing so is a meaningful contribution even where none of its component intuitions is unprecedented.

The framework's strongest version is a theory of catastrophe prevention through preserving correction capacity. The catastrophes it sees best are generated by systems that keep operating after reality has started proving them wrong. Reality diverges from the model, people absorb the difference, and the institution reads successful compensation as evidence the model still works. Seventeen essays' compensation vocabulary, Hollnagel's resilience engineering, and the frontline-adaptation cases behind the crumple zone all name the same loop. What resilience engineering calls work-as-done, this framework reads as telemetry: when people repeatedly have to save a system, those saves count as evidence that the system is failing. Correction is cheap only while the absorbers' capacity holds.

The shortest way to see the whole framework

The burden-visibility diagram shows why a system can look efficient while its costs are hidden: it counts the completed outcome and ignores the phone calls, waiting, caregiving, masking, and workarounds that produced it. The good-vs-bad-insulation diagram separates the insulation that protects a decision-maker from improper pressure from the insulation that shields it from consequence. Read together they answer the two questions the framework starts from: what the system counts, and who is allowed to stay unchanged.

The inferences this framework refuses

The recurring failure the archive attacks is an inference rather than a particular institution. A fact is true, and a conclusion about what should happen gets smuggled out of it. Each conversion below is a move institutions make routinely, and each is invalid. A fact about capacity, necessity, reliance, adaptation, provision, or credentials does not support the conclusion it is routinely asked to carry.

1. Useful becomes entitled. Relevance does not become jurisdiction: an institution can be competent, predictive, or deeply relied on without acquiring a right to govern adjacent parts of a life.
2. Necessary becomes justified. Dependence does not imply subordination: needing an employer, a state, a platform, or a system never justifies giving it progressively broader command.
3. Adapted to becomes consented to. Adaptation is not consent: functioning inside a bad system because the alternatives are worse is not evidence the arrangement was endorsed.
4. Successful becomes competent. Observed success can conceal institutional failure: outcomes systematically overstate competence when invisible human effort prevents failures from surfacing.
5. Provided becomes relieving. Formal provision is not practical relief: an institution can officially supply help while exporting the scheduling, coordination, and troubleshooting onto the person supposedly helped.
6. Exit on paper becomes meaningful exit. How much refusal costs decides how much weight consent can carry.
7. Dependence becomes indispensability. Necessity can be self-produced: a system can restrict alternatives, force adaptation, and then cite the resulting dependence as proof of its indispensability.
8. Technically possible becomes socially warranted. Capability does not create permission, and expertise in one domain does not yield authority outside it.

The framework's own inference is the mirror image of these refusals. Sometimes two claims that both look true collide, such as a formally neutral rule and a systematically unequal experience, or an apparently efficient institution and exhausted employees. Treat the collision as evidence that the representation is missing a variable, before concluding that one side is lying. Ask what must be true of the underlying system for both observations to hold. Then look for the mechanism, such as identical rules imposing unequal compliance burdens, ambiguity allocated asymmetrically, or unmodeled cases charged a hidden adaptation cost. The archive's concepts come from that both-true test. Each one names the hidden relation that explains how both observations can be true.

What this framework cannot do

Architecture is frozen politics. Who gets standing is a political settlement. So are what counts as evidence, which findings must be acted on, who bears delay, whether an appeal pauses the harm, and who can take away whose authority. Instead of replacing power analysis with mechanism design, the framework asks where power has been encoded into the mechanism. The economy the mechanism lives in is part of that analysis, since who profits from an institution’s errors and whether correcting them is priced above avoiding them decide whether a corrigible design survives.

Corrigibility is a necessary limit on legitimate power and falls well short of a full theory of justice. A perfectly corrigible deportation regime could still pursue an immoral policy. Substantive commitments (equality, liberty, welfare, anti-domination) still have to specify what institutions should pursue.

An impossibility result is a structural claim. The term fits only when, under the institution’s rules, no allowed input can produce the outcome. For example, an oversight board that can investigate but has no way to force the conduct to stop cannot stop it with the powers it has. It does not fit a prediction that a reform probably won’t work.

The framework takes no side on how much public power there should be and is strict about how that power handles its errors and who carries their cost. A highly redistributive welfare state, national health service, public housing authority, industrial planning agency, or powerful regulator can satisfy it better than a privatized market regime. The constraint is that power, however large, must still change in response to confirmed evidence of its own mistakes.

The framework applies symmetrically to private power. Employers, insurers, platforms, landlords, banks, hospitals, and model providers also allocate persistence, route errors downward, impose stamina tests, and preserve authority after demonstrated failure. Where private systems have no internal way to be corrected, the framework makes the case for more public intervention.

The opening claims are easy to accept, so the steps beyond them have to be flagged. Hidden burdens are nearly universal, but what follows from them is not. A reader may agree that burdens exist and are distributed by power, reject the redesign a conclusion proposes, and still carry the diagnostic intact. The framework keeps that possible by being honest about its own escalations. When an argument moves from an observed relation to a structural theory, it should say so. Keeping the exits open stops the method from turning into the persuasion it refuses to be.

What the framework builds toward

The limits name what this framework cannot do. The diagnosis also makes positive claims, compressed into one sentence: build institutions that increase what people can safely depend on while reducing what that dependence lets others demand from them.

Dependence creates duties before it creates any authority. The more people rely on an institution, the stronger its duties of reliability, care, accessibility, explanation, and repair should become, and the weaker the case that their reliance gives it authority over them. Dependence alone never gives anyone the right to command.

Social provision is freedom's cost structure. Housing, healthcare, income security, and public services matter partly because they lower the material penalty for saying no to an employer, a platform, a profession, a family arrangement, or an administrative demand. Provision makes refusal survivable, and refusal keeps provision answerable.

The positive ideal is interdependence built on strong institutions with little power to compel: institutions that provide a lot, decide narrowly, can be corrected while they run, and leave people plenty of practical freedom. The opposite of domination is interdependence arranged so that people gain security without giving up too much control over their lives.

The framework's philosophical leverage is a set of separations. Dependence need not imply subordination. Capacity need not imply jurisdiction. Provision need not imply paternalism. Expertise need not imply rule. Consent counts for less as the cost of refusing rises. Institutional strength need not imply diminished liberty. Classic debates assume these move together, but the framework treats each as an independent design question.