Much of the middle layer runs on judgment. How damaged is this participant? How far does an addiction reduce their capacity to reciprocate? Which part of a burden is theirs to carry? Who has any authority at all over a person living on the street? The framework can answer use judgment — but the moment it does, a harder question arrives and will not leave: whose judgment, over what domain, on what evidence, for how long, and subject to what appeal? This is the point where the framework either becomes a governance architecture or quietly collapses into benevolent discretion, which is only soft sovereignty wearing a kind face.
The answer is not better rulers. It is distributed adjudication.
No single actor may decide a participant’s total standing, burden, capacity, or future. Each decision is bounded to its domain, made by the least concentrated competent authority, and remains reviewable.
That is stronger than let the experts decide, and it rests on one line that does most of the work:
Institutions may acquire jurisdiction over problems, not ownership over participants.
Break the question apart
The failure begins with the shape of the question. Ask what should we do with this person? and you have already granted someone total jurisdiction — the whole life becomes one case, handed to one authority. So the first discipline is to refuse the giant question and break it into small, domain-bounded ones. For the person on the street that means asking, separately: Are they medically impaired? Is an addiction reducing decision-making capacity, and for which decisions? Are they a danger to others? Can they manage housing independently? What support do they actually want? What burden are they placing on shared space? Which services are genuinely scarce? What restriction, if any, is warranted — and who is responsible for each of these, one at a time? Different questions belong to different decision-makers, and keeping them apart is what stops any one of them from becoming a verdict on the person.
Standing is not on the table
The first and firmest boundary: standing is never adjudicated. No forum, however expert, gets to decide that a participant no longer counts. That question is constitutional, not discretionary — it is simply not among the things a doctor, agency, court, officer, or algorithm is empowered to rule on. Institutions decide plenty of real things: eligibility for a program, the degree of a specific incapacity, a restriction a danger requires, how a scarce resource is allocated, which treatments are offered. None of those is the question does this person still belong in the ecology? Taking that question off every desk is what stops any office from becoming sovereign over human legitimacy.
Capacity is domain-specific
Is this person capable? is the wrong question, because it is too big. Capacity is not one switch. Someone may be unable to manage complex medication yet perfectly able to choose whom they live with; unable to handle their finances yet entirely able to refuse a program they find objectionable. So the framework rejects the global incapacity judgment and asks only the narrow one — can this person make this decision, at this time, under these conditions? A real deficit then constrains only its own domain, and personal jurisdiction survives everywhere else. Where authority must narrow, it narrows by domain and on a clock, through the bounded-override discipline of supported decision-making and periodic review — never as a single finding that hands one office control of the whole life.
Jurisdiction follows the problem, not the person
No participant belongs to a single authority. The right picture is a set of bounded jurisdictions, each attached to a problem rather than to the person: clinical matters to clinicians, tenancy and its support to a housing body, immediate danger and legal violation to the safety and court systems, program eligibility to an administrative agency, a genuine capacity question to a specialized review forum — and the participant retaining authority over their own life wherever no compelling counter-claim applies. That is distributed adjudication in practice, and it runs on the least-jurisdiction principle: use no more authority than the problem actually requires. Sleeping outside, but not dangerous, justifies nothing over a person’s beliefs, relationships, finances, or medical choices. Lacking financial capacity justifies nothing over where they worship. Being dangerous while intoxicated may justify temporary restraint, but not a permanent loss of self-jurisdiction. Every step past the minimum is custodial drift — support quietly turning into control.
From blame to burden
Where a choice creates cost — the familiar example is smoking — the tempting question is how blameworthy is this person? That is almost always the wrong jurisdiction, and unanswerable besides. The framework shifts from blame to burden: what burden actually exists, where does it land, how preventable was it, who had meaningful control over it, and what intervention is proportionate? Those are usable questions, and they need no moral accounting. They also need no single culprit. A person’s condition on the street may involve personal choices, addiction, childhood harm, a broken housing market, healthcare that failed, displaced work, family collapse, and public policy all at once; trying to assign blame in percentages is both impossible and beside the point. The actionable question is not who caused all of this? but who can reduce this burden now? — which lets responsibility be shared even when causation is a tangle.
Reciprocity by capacity, not worthiness
Deciding what a participant owes runs into the same trap. Is this person trying hard enough? cannot be adjudicated fairly; it is guesswork about effort dressed as judgment. Replace it with functional questions: can they reliably attend appointments, manage medication, meet basic shared-space responsibilities, take part in their own treatment decisions, comply with safety rules? These are observable and bounded, and they let expectation track real capacity — the commensurate logic — without ever grading a person’s character or touching their standing. It is diagnostic differentiation on the reciprocity side: differentiate what is asked; never differentiate whether the person counts.
Bounded, provisional, reviewable
Discretion cannot be eliminated, but it can be disciplined. Any high-consequence judgment should carry the ordinary machinery the framework already builds into its jurisprudence: an explicit domain, an evidence standard, documented reasoning, a stated duration, a review date, a right of appeal, an independent second opinion, conflict-of-interest rules, and a way to reverse. Three of these deserve emphasis, because present systems get them wrong most often. First, authority expires: a restriction should lapse unless renewed with fresh evidence — ninety days unless reauthorized, not this person is now categorized as incapable — so a temporary condition never hardens into a permanent identity. Second, the hardest cases get plural review, not one expert: forced treatment, guardianship, long-term confinement, or the loss of major support should pass through more than one lens — medical, legal, social, an advocate, sometimes community or family — because every profession has its own blind spot (clinicians over-medicalize, safety systems over-read danger, social workers over-protect, families over-control, algorithms over-classify), and plural review keeps any one of them from becoming the whole truth. Third, the participant carries a right to independent representation and stays present in their own adjudication — decisions made with them, not merely about them: explain, ask what they want, preserve the choices that can be preserved, distinguish a refusal from an incapacity, and revisit later. The word that ties it together is provisionally. Someone decides — but provisionally, within a bounded jurisdiction, on the record, and open to being undone. And throughout, AI may inform these judgments but does not become their arbiter.
The presumptions it runs on
Stated as defaults, the architecture becomes usable at the desk where decisions actually get made:
1. Standing remains intact. 2. Self-jurisdiction remains intact unless specifically rebutted. 3. Restrictions must be narrowly tailored. 4. Restrictions expire unless renewed. 5. Scarcity may constrain allocation, not standing. 6. Burden is assigned as close to its source as possible, without creating disproportionate harm. 7. Appeal must remain possible.
The burden of proof always sits on the expansion of authority, never on the participant defending their own standing.
The homeless person as a jurisdiction map
Put together, the question that felt unanswerable — who has jurisdiction over a homeless person? — turns out to have a clean answer: no one, over the person as a whole. Different actors hold jurisdiction over different interfaces. The city has authority over the use of public space; the health system over clinical treatment; a housing provider over tenancy; the courts over legal violations; and the participant over their own life wherever no bounded counter-jurisdiction applies. The person is not a case owned by an authority; they are a participant touched, at specific points, by several. And that reframing yields a single diagnostic that catches a great deal of abuse:
Is any institution extending its jurisdiction beyond the problem it is actually responsible for?
No single authority
One implementation would betray all of this at a stroke: a central body — a National Office of Standing, say — empowered to assess each person’s ecological worth. That is exactly what the framework forbids. It is constitutional in structure and distributed in execution: it supplies principles, jurisdictional limits, appeal rights, evidentiary expectations, and red lines, and then lets bounded, plural, local authorities do the deciding. No monopoly may form over what matters — least of all a monopoly over who counts. But distributing authority this way raises its own hard question — who then adjudicates among the many deciders, and what keeps the distribution itself from becoming deadlock or quiet capture? That recursion, and where it terminates, is worked out in deciding without a final authority.
Which is the whole point restated. Standing settles who belongs; allocation, reciprocity, and repair handle the practical questions beside it — but every one of them requires judgment, and judgment is where sovereignty creeps back in. Distributed adjudication is the guard:
No one gets to decide the whole person.
Every expansion of jurisdiction has to answer for itself — why this domain, why this much authority, for how long, on what evidence, who can challenge it, and what would return jurisdiction to the participant. Answer those honestly, every time, and judgment stays judgment instead of hardening into ownership.