Capture is usually described as something done to an institution by an outside party: a regulator captured by the industry it regulates, a board captured by management. An institution can also be captured by a system it deployed, without anyone intending it. The mechanism is not influence. It is the gradual disappearance of everything the institution would need in order to stop.
How capture happens without anyone choosing it
An institution adopts a system because it does a job better or cheaper than the arrangement it replaces. That is a good reason. Once the system is in place, the old arrangement becomes redundant, and redundancy is a cost. The staff who did the job move to other work or leave. The processes that ran without the system are not maintained, because nothing runs through them. The records that would show what the system decided and why are kept in the system’s own format, because that is where they are generated. The alternatives that were evaluated at procurement are not re-evaluated, because the decision was made.
Each of these steps is rational. Each of them is also the removal of a capacity the institution would need to question, replace, or withdraw the system. After enough steps, the institution can no longer do any of those things at acceptable cost. It has not been coerced. It has spent, in the name of efficiency, the resources that its own contestability was made of.
This is capture in the relevant sense: the state in which an institution can no longer afford to reconsider a decision it once made freely. The system did not seize control. The institution handed over control one economy at a time, and each economy looked, at the time, like good management.
Theory, drawn
The capture ratchet: how an institution loses what it needs to stop
Capture is the gradual loss of what an institution needs in order to stop a system: the staff, the skills, and the alternatives. Law XII protects that reserve, not only the kill switch.
Why the boundary is hard to see
The capacities that capture consumes are, almost by definition, capacities that are not being used. The staff who could run the process by hand are not running it by hand; the system is. The alternative vendor is not being paid; the current one is. The manual review process is not reviewing anything; the system’s outputs are trusted. From inside the institution, these look like slack, and slack is what efficiency exists to remove.
The boundary that Law XII draws is a boundary around unused capacity. It says that some of what looks like slack is not slack but reserve, held against a contingency the institution hopes will not arise, and that the reserve may not be spent even though it produces nothing on ordinary days. That is a difficult thing to defend inside a budget process, because the case for the reserve is always about a future that has not happened, and the case against it is always about a cost that is being paid now.
Older traditions know this problem and have solved it by rule. A hospital keeps generators it hopes never to run. A bank holds capital it hopes never to draw on. An aircraft carries fuel for a diversion it hopes never to make. In each case, the reserve is mandated, because the institution left to itself would optimize it away and then discover, at the wrong moment, why it was there.
What is being preserved
The capacities that matter are the ones the institution would need on the day it decides the system should stop, be replaced, or be made to answer for something. They are not mysterious. Expertise: people who understand the domain well enough to do the work, or to judge whether the system is doing it, without the system’s help. Process: a way of getting the job done that does not run through the system, even if it is slower. Records: an account of what the system decided, why, and on whose authority, kept in a form the institution controls and can read without the system. Alternatives: at least one substitute that has been kept viable enough that switching is a real option rather than a theoretical one. And standing: parties who retain the right and the means to challenge the system’s decisions, so that the question of whether it should continue can be raised by someone other than the people who deployed it.
Each of these is a capability the institution preserves precisely because it hopes never to need it. Their value is not in their use. It is that, because they exist, the institution’s continued reliance on the system remains a choice. The day they are gone, reliance stops being a choice and becomes a condition.
Reversibility at the institutional level
A system can be technically reversible, in that a switch exists and works. It can be operationally reversible, in that the people and processes can absorb the switch being thrown. And it can be institutionally reversible, in that the organization can survive the consequences of having thrown it. Capture is the loss of the third level while the first two may remain intact.
An institution in that state has a working halt control that it will never use, because using it would be ruinous. The control is nominal. Nobody will admit this in advance, and the institution may sincerely believe it retains the option. The belief is tested only when something goes wrong enough that halting is contemplated, and at that point the institution discovers what it has spent.
This is why rehearsal matters and why it is resisted. Rehearsing withdrawal reveals whether institutional reversibility still exists, and the rehearsal is expensive, disruptive, and produces news that nobody wants. But an institution that has not rehearsed withdrawal is asserting a capacity it has not tested, and the assertion is exactly the kind that capture makes false without anyone noticing.
Where the line falls
A consequential system may draw on the institution’s capacities. It may use its staff, its processes, its records, and its authority. It may become important. It may become the way most of the work is done. What it may not do is consume the capacities that would let the institution question, replace, or withdraw it. Those are held in reserve, named, recorded, and reviewed, and their erosion is treated as a governance event rather than as an efficiency gain.
This is a constraint on the institution rather than on the system. The system does not choose to consume the reserve; the institution chooses to spend it. The law therefore obliges the institution to know what its reserve consists of, to decide deliberately whether any part of it may be spent, and to accept that a system whose reserve has been spent is a system the institution can no longer govern, whatever its dashboards say.
What the laws take from this
Dependence is a state variable, so that the institution can see how far it has come to rely on a system and what withdrawal would now cost. Nominal reversibility is not operational reversibility, and the institution has to evidence all three levels rather than pointing at a switch. And no system may erase the conditions of its own contestability, which means that certain capacities are listed, preserved, and rehearsed, and that spending them is a decision with a name.
A reader can reject the political framing of capture and still accept the constraints. They amount to the observation that an institution that cannot halt a system is not governing it, and that the ability to halt is made of things that look, on most days, like waste.