A person’s capacity to adapt is finite, and who spends it is a political question. Ethotechnical design asks not only whether an arrangement works. It asks what adjustment the arrangement demands from people, who bears that demand, and whether they can force the arrangement to change.
The essay on what Ethotechnics is not draws the discipline’s boundary. The essay on what outcomes hide applies it to evaluation.
Who adapts
Every design is an imperfect model of reality. A form, a workflow, a building, an algorithm, a rule, an interface, or an institution sets expectations about the people and situations it will meet. Some reality will not fit.
The ordinary design question is: how do we make people interact successfully with this system? The ethotechnical question inverts it: when the system and the person do not fit, why should the person be the thing that changes?
The discipline is about who is made to adapt:
design → mismatch → a need to adapt → someone assigned to meet it
A badly designed arrangement keeps assigning that need downward, to the people with the least say. Users remember passwords. Patients coordinate providers. Workers repair software. Disabled people request exceptions. Citizens reproduce information government already possesses. Caregivers integrate fragmented services. The arrangement stays stable because people bend around it.
Ethotechnical design reverses the direction:
Spend institutional capacity before consuming human adaptive capacity.
That does not mean people should never adapt. Learning, cooperation, compromise, care, and collective life all require it. The distinction is between adaptation that is productive or mutual and compensatory adaptation: effort a person spends mainly because an arrangement failed to accommodate something it could have foreseen.
How a designer reverses the direction
First, eliminate the demand where possible. Before improving a form, ask whether the form needs to exist. Before making compliance easier, ask whether compliance is necessary. Before designing an accommodation process, ask whether the default could accommodate more variation.
Second, absorb complexity upstream. Where complexity is necessary, place it where there is the greatest capacity to handle it. Institutions can integrate databases, coordinate departments, hire expertise, maintain infrastructure, automate repetitive tasks, and learn across cases. Isolated individuals usually cannot.
Third, design for the remainder. No arrangement will anticipate everyone. An exception is not necessarily evidence of bad design. Some exceptions are certain. Ethotechnical design therefore includes low-cost exception handling, human judgment, escalation, reversibility, appeal, and a way to keep operating when the model meets something it did not anticipate.
Fourth, let repeated mismatch redesign the system. This separates accommodation from correction. If thousands of people need the same exception, the system should not get better at processing thousands of exceptions. It should ask whether the rule producing them is wrong.
Unfinished design
The four moves assume that design is deliberately unfinished. No amount of anticipation eliminates mismatch. A conventional designer can treat the specification as authoritative and the exception as deviation. The ethotechnical designer treats the specification as a hypothesis and the exception as possible evidence.
That changes how the designer reads failure. When users keep working around something, the workaround is evidence about the design, not only clever behavior. When frontline workers keep exercising unofficial discretion, the formal process may be inadequate. When disabled people keep requesting the same accommodation, the question is not how fast the requests can be processed. It is what the requests reveal about the default.
The remainder is information about the design.
Following the friction
After a redesign, ask who is doing the work now. A healthcare portal may save staff time while making patients assemble the information themselves. The evaluation framework checks the result, the work required, and people’s power to change those requirements.
This widens accessibility. Accessibility is not a feature added for a subset of users. Human variation, whether bodily, cognitive, temporal, linguistic, or circumstantial, is an ordinary fact a system should expect. When some people cannot compensate for a design, they may expose a burden that better-resourced people have been absorbing without comment. The workaround presumption, the rule that a recurring workaround is first read as evidence of an upstream design failure, applies the same reading to coping.
Corrective power, not ease
The discipline’s object is who bears the burden of adapting and whether they can get it corrected. That is why it cannot be reduced to universal design, human-centered design, participatory design, service design, ergonomics, or accessibility, although it borrows from each. Its view of participation follows: it values participation but does not assume more is better. Making people constantly report problems, attend meetings, submit feedback, monitor decisions, and organize for basic corrections creates another adaptation burden.
The objective is high corrective power at low human cost. People should not have to become experts in the institution that governs them to make it behave reasonably. The test for any arrangement asks: what does this design require people to notice, remember, learn, endure, coordinate, translate, correct, prove, monitor, or repeatedly request — and which of those tasks could the arrangement itself absorb? Then it asks the political question: when those requirements become unreasonable, can the people bearing them make the design change?
Those questions separate the discipline from making something pleasant or convenient. Its aim is not maximum ease, personalization, or efficiency. It is that the institution spends its own staff, budget, and systems on complexity so that people do not spend their evenings on it.
What the laws take from this
Each move needs something an institution can be held to.
Law IV: every consequential delegation creates a correction obligation. That is the fourth move. Repeated mismatch has to be able to change the rule that produces it, not only be absorbed more efficiently.
Law VII: error-bearing parties require standing proportional to exposure. That is the political test question, whether the people bearing an unreasonable demand can make the design change.
Law VIII: observability without state transition is theater. It separates a system that logs exceptions from one that is changed by them.
Law XI: successful automation increases its own governance burden. Every internal step a system saves is a demand that has landed somewhere — when it has landed. Some steps disappear without landing anywhere: a duplicate entry, an obsolete approval, a preventable error. The law does not assume the transfer; it asks of each saving whether a demand appeared elsewhere, and makes the next expansion a decision that has to say where.