Institutions are in the business of demanding that human beings adapt. A patient must learn the patient portal. A clinician must master the alert routing queue. A welfare claimant must compile three years of bank statements. A worker must absorb the erratic schedule a scheduling algorithm produces.
Most governance frameworks ask whether these demands are legally authorized, procedurally uniform, or technically efficient. Administrative burden research, led by Pamela Herd and Donald Moynihan, measures what the demands cost the people who meet them. This essay asks the question of permission:
Under what conditions, if any, may an institution legitimately consume another person’s finite capacity to cope?
When an arrangement answers that question by silence, it treats people’s capacity to cope as unlimited and free. That silence produces the human crumple zone: the person who absorbs the cost and the blame when a system fails, the way a car’s crumple zone absorbs a crash.
A limited resource
Human beings do not possess infinite resilience. The capacities required to get through an institution’s friction — sustained attention, working memory, emotional regulation, patience, off-hours paperwork, vigilance against automated errors, and care for others — are limited by the body and by the clock. Together they are adaptive capacity. The term comes from resilience engineering, where it means a system’s capacity to stretch when surprise exceeds what it was built for; David Woods calls that graceful extensibility. Here it means the human share of that capacity. Every hour an individual spends deciphering a contradictory bureaucratic notice or manually translating data between two incompatible software systems is an hour subtracted from sleep, caregiving, creative craft, and ordinary life.
Feminist political economists call the background work that keeps people functioning from one day to the next social reproduction. Adaptive capacity is what that work runs on.
When an institution designs a system that leaks errors, it does not eliminate those errors; it routes them downward. Frontline clinicians stay late to reconcile missing lab reports. Desperate parents spend days on hold attempting to correct automated benefit terminations. Because this work happens outside anything the institution measures, the institution’s dashboard reports smooth operation and lean overhead. The cost has not disappeared; it has been externalized onto the people least able to refuse it.
The cost nobody books
An externality is a cost one party imposes on others without paying for it. This cost has a name: the adaptation externality, the coping an institution’s defects demand from the people around it.
The evaluation problem is that good results can conceal this work. Here the concern is what sustained compensation does to the people supplying it. When demands exceed recovery, their remaining capacity falls even while the institution’s reported performance stays level. Eventually the cost of challenging the arrangement may exceed what they have left.
Theory, drawn
A level dashboard over a falling reserve
Observed performance is designed capacity plus human compensation, and the dashboard shows only the sum. The sum holds level. The compensation is drawn each week from people's attention, evenings and sleep, and less comes back than was spent. Below the dashed line, the person no longer has what contesting a decision takes. A demonstration: no data behind the shapes.
Genuine versus manufactured necessity
Not all friction is illegitimate. Living in a material world with mortal bodies, limited energy, and complex coordination requires real effort. The task is to tell genuine necessity from manufactured necessity.
| Dimension | Genuine necessity | Manufactured necessity |
|---|---|---|
| Origin | Irreducible physical scarcity, mortality, natural complexity, open-ended collaboration. | Unreconciled databases, rigid software schemas, departmental silos, budget caps on intake staff. |
| Alternative | No feasible institutional redesign can eliminate the friction without violating physical reality. | Friction could be eliminated by reallocating the institution’s internal budget or automating internal handoffs. |
| Burden distribution | Shared collectively or absorbed through reciprocal, recognized care. | Asymmetric: concentrated downward on the claimant, patient, or frontline worker. |
| Institutional response | Transparent acknowledgment; continuous effort to buffer the vulnerable. | Defensive denial; treating the friction as “fraud prevention” or the subject’s personal incompetence. |
When an institution justifies manufactured necessity by calling it unavoidable, it commits a category error: it mistakes the limits of its own budget for the laws of nature.
What counts as progress
Treating adaptive capacity as a limited resource changes what counts as institutional progress.
Reform usually focuses on making people better at coping: teaching claimants how to navigate convoluted portals, training clinicians in “resilience” and mindfulness, or deploying consumer-facing AI agents to automate the filing of appeals.
While coping tools may provide temporary relief, treating them as progress is perverse. When institutions discover that people have become better at coping, they routinely raise the baseline demand. Better consumer tools are met with more complex administrative obstacles, raising the minimum stamina required to secure basic rights.
Progress is not making people better at enduring bad systems. It is making the system depend less on people with exceptional stamina or skill.
An arrangement is not just because a skilled person can get through it. It is just when someone with ordinary limits — no special stamina, no spare hours, no expertise — can use it without the attempt costing them their evenings.
Architectural constraints: the binding laws
1. The correction obligation (Law IV)
Every delegated automated action creates an institutional obligation to correct its errors. If an automated decision generates a false rejection, the burden of discovering, documenting, and contesting that mistake cannot be externalized onto the person it decides about. The time and adaptive capacity consumed by the correction process must be tracked as an operational debit against the system’s authorization grant.
2. Proportional standing for error-bearing parties (Law VII)
Whoever absorbs the adaptation externality must hold corrective standing capable of forcing upstream rule revision. If an institution relies on frontline nurses or benefit recipients to patch its defects, those actors cannot be confined to advisory committees or customer feedback surveys. They must possess the formal standing to introduce counterevidence that invalidates the operating model’s assumptions.
3. Preserving the conditions of contestability (Law XII)
No system may consume the very resources required to contest it. An appeal process that requires thirty hours of legal research and paperwork from an unhoused person violates Law XII: it wins by attrition. A legitimate system keeps the friction of challenge lower than the friction of submission.
What the person’s time is for
Every demand listed above is paid out of the same budget: sleep, caregiving, work, and everything else a person’s hours are for. The constraints in the previous section cap what an institution may take from that budget.
The design practice sets out how to remove demands, absorb complexity upstream, and handle the remainder. What the redesign frees up returns to the person.