Part I
The Center and the Crossing
Aa
What the center cannot see, it cannot govern by pretending to see.
Computational systems now act at scales and tempos no person can follow continuously, but speed is only the most visible part of the difficulty. The deeper limit is positional. Much of the knowledge needed to judge a consequential act exists where the act meets a particular person, practice, institution, or moment, and some of it is generated only in that encounter. A center can establish common rules, compare cases, prohibit bounded harms, and study patterns that no local participant can see. It cannot make every local circumstance one of its own experiences.
Every administrative order nevertheless has to generalize. It creates forms, categories, models, and reports through which distant events become governable. James Scott's scientific forest remains the emblem because the plan did not fail by seeing nothing. It saw timber with remarkable clarity. What disappeared were the relations that did not answer the fiscal question: undergrowth, soil, minor species, customary use, and the slow ecology on which the visible yield depended. The danger begins when a representation that was built for one purpose acquires the authority of the whole case.
The knowledge problem is therefore neither an argument for a blind center nor a romance of local judgment. Local knowledge can preserve prejudice, favoritism, and error as readily as wisdom. Central knowledge can expose patterns, discipline arbitrary variation, and protect people captured by the institutions nearest to them. The constitutional question is how these forms of knowledge meet without either claiming an authority it has not earned.
Verification cost enters at that meeting point. Where checking is expensive, governance tends to gather inside institutions large enough to absorb the overhead, and those institutions acquire both the information and the authority that follow from being trusted. Where checking becomes cheap, claims can travel with more of their evidence and people can test some exercises of power without rebuilding the issuing bureaucracy. Yet cheap verification chooses no political direction. It can distribute standing or perfect surveillance, expose the institution or expose the person, open a trust monopoly or make exclusion nearly frictionless.
The decisive moment is the crossing. A score becomes a denied loan, a model output becomes a medical decision, a discrepancy becomes a debt, and an identity relation becomes permission to work. Call that crossing the membrane. It may be distributed across a model, policy engine, data broker, interface, and reviewing office, with each institution able to describe its narrow contribution while no one owns the act as a whole. Identity, settlement, communication, and access can then become forms of conditional tenure, administered through systems the person depends upon and cannot inspect on comparable terms.
Part I follows two objects: the report that reaches the center and the consequence that reaches the life. It asks what is lost as knowledge travels upward, what must remain attached as authority travels downward, and which institution can change the result when the abstraction no longer fits the case. The answer will not be a sovereign capable of seeing everything. It will be an order in which the center knows the limits of its representation and the crossing leaves enough evidence for the person at the far side to answer back.
Once that limit is admitted, another question becomes unavoidable. If neither central command nor isolated local judgment can carry the whole work of coordination, what institutions have divided that work before, and what happens when computation changes the cost of doing so?
Part II begins with the forum and the firm.
The center need not know everything. It must know the limit of what it knows, and make the crossing answerable.