The companion essay on friction as accidental governance is about constraints that existed as side effects of incapacity: a review happened because the system could not decide by itself; a cooling-off period existed because the system could not act instantly; an exception route existed because the system could not handle the case. Those constraints were doing constitutional work, and nobody built them to.
This essay is about what happens when the incapacity goes away. The accidental constraint was never named as a constraint, so nothing protects it. The capability that arrives does not merely add an option; it dissolves a structural limit, and with it the safeguard that limit happened to be providing. What was a wall becomes a door policy, and door policies are much cheaper to argue out of existence than walls.
The default is deployment
Organizations treat a newly possible capability as a decision already made. The question shifts, almost without anyone posing it, from should this exist to under what conditions should we deploy it. The conditions are real work — that is what the standards on this site are for — but the work happens entirely downstream of a decision that was never taken. A capability that can infer emotional state, health risk, employability, political preference, or behavioral intent arrives with the implicit answer to its own question attached.
The non-conversion principle says a capability does not become authority on its own: a tool being reachable is not permission to use it, and a pilot going well is not a rollout. That principle keeps capability and authority in separate objects so the widening of one cannot silently become the widening of the other. But it presupposes that a grant decision is what happens next. There is a decision upstream of the grant that the principle does not by itself force anyone to consider: the decision that this capability should remain outside institutional decision-making altogether, whatever its accuracy, whatever its efficiency, whatever the competitors are doing with it.
What refusal protects
The case for non-use is not prudishness about technology. It is jurisdictional. The question a deployment review asks — can this system be stopped, explained, appealed, and repaired — has a prior sibling that the standards assume has already been answered: should this be a place where institutional decisions happen at all.
Some inference is like that. A model that estimates a person’s emotional state from their face can be accurate, audited, and appealable, and the arrangement can still be illegitimate, because the legitimacy question is not about the artifact. It is about whether the institution may hold that inference about a person at all, and whether any amount of procedural polish around it is enough to answer for holding it. The standards can make an arrangement governable. They cannot make every operation one an institution ought to be conducting.
This is the strongest form of bounded jurisdiction. The ordinary form says: greater technical capability does not produce greater legitimate authority, so each widening needs its own justification. Non-use says: some capabilities should remain unavailable to institutional decision-making even when they work, because the work they would do is not work an institution can be made answerable for doing.
Why it is hard to hold
A refusal has no constituency. It produces no product, no efficiency claim, no metric improvement, no case study. The person it protects is a person something did not happen to, and nothing happened to nobody is not a headline anyone collects. Meanwhile the ratchet runs: a system that works gets used more, trusted more, extended further, and each step of extension is small enough to look like conditions on deployment rather than the answer to the deployment question.
That is why non-use has to be built as a standing decision rather than held together by whoever happens to be paying attention. The capability catalog and the authority surface exist as separate objects precisely so that we can and we may cannot collapse into each other, and the discipline’s expansion rules exist so that widening is always somebody’s decision. Non-use is the same machinery pointed the other way: a recorded refusal, with its basis, its scope, and the conditions under which it would be revisited. It is not a shrug. It is a decision with the same standing as a grant.
Prohibitions replace incapacity
As capability accumulates, the structural version of a safeguard is replaced by the political version. When the system could not read faces, there was no decision to make about face-reading; when it can, the absence of face-reading has to be maintained by someone, on the record, against pressure from every direction the capability’s value comes from. The friction-as-accidental-governance argument shows what disappears when friction is optimized away. This argument shows what has to be built to replace what incapacity used to do for free: named boundaries, held against the temptation of their own removal.
The law this answers to is the one about preserved capacities. An institution keeps some capacities precisely because it hopes never to need them, and treats their erosion as a governance event rather than an efficiency gain. Non-use extends the same posture one level up: there are capabilities the institution preserves in its catalog and declines to operationalize, and the preservation is deliberate, recorded, and revisable rather than accidental and therefore disposable.
The measure
A discipline that only asks how to use capability safely will always find a way to use it, because the question contains its own answer. The fuller question is what a deployment is for, and one honest answer to that question is that the capability should not enter institutional reach at all. A mature technical civilization is one that can hear that answer, act on it, and keep the refusal on the record — not one in which every capability that becomes possible eventually becomes practiced, and every safeguard around the practice becomes negotiable.
What the laws take from this
The method does not require a reader to accept that some capabilities should never be operationalized. It takes three constraints that give the refusal structural weight. From Law I it takes that capability and authority live in separate objects, so that a recorded non-use — with its basis, scope, and review conditions — has the same standing as a grant rather than the status of an omission. From Law XI it takes that expansion is a decision, not a consequence of success, which is what keeps “conditions on deployment” from quietly becoming the answer to “should this exist at all.” From Law XII it takes that the capacities contestability depends on are preserved on purpose, and the same posture extends upward to capabilities the institution declines to operationalize. An engineer who rejects the essay can adopt all three on narrower grounds: a refusal that is not recorded is not a decision, a widening that is not a decision is a drift, and a safeguard that protects only what is already built protects nothing that was never built.