Who Counts: The Short of It

How we decide who has rights, why we have always got it wrong, and why it matters to you personally.

By Geoff Fane. August 2026.

This is the short version. The full essay, with citations and the legal history worked through in detail, is Who Counts: The Long of It. The principles drawn from both are set out in The Reciprocity Manifesto, which you can sign.

The boy who cannot vote

A seventeen-year-old works four shifts a week, pays tax, and follows politics more closely than anyone on his street. He cannot vote. Next year, having done nothing in particular to change, he can. Nobody thinks he grows wiser at midnight. We use an age because we need a line, and lines have to go somewhere.

But when the stakes rise, the law abandons the line. Since a case brought by a mother named Victoria Gillick in the 1980s, a child’s right to consent to their own medical treatment has depended not on age but on whether that particular child understands what is proposed. A mature fifteen-year-old can consent. An immature seventeen-year-old may not.

The same thing happens at the other end of a life. When someone is thought to have lost the ability to manage their affairs, English law asks four questions. Can they understand the information? Retain it? Weigh it? Communicate a choice? A woman with dementia may lack capacity to sell her house and retain capacity to refuse an operation and choose who visits her. Capacity is assessed decision by decision, and an unwise decision is not evidence of its absence.

When the stakes are low we ask what someone is. When the stakes are high we ask what they can do. And notice what those four questions leave out. Not one of them mentions being human.

The question the test never asks

Every test described above is applied to somebody whose membership is already settled. Nobody asks whether the seventeen-year-old is the sort of thing that could in principle vote. What is in question is only how much say he gets inside a membership already granted.

So what decides membership? In practice, almost everywhere: being born human. That is not a test. It is a fact about ancestry, and it is carrying enormous moral weight while offering no justification at all.

You might think this never matters, since everyone we meet is human. It has mattered, repeatedly, and what happened then is the most important evidence we have.

The people we decided did not count

In 1772 an English court ordered a man held as a slave to be produced before it and his detention justified. It could not be, and James Somerset walked free. Eighty-five years later the Supreme Court of the United States held that Dred Scott, a black man, was not a citizen and had no standing to sue at all. Not that he lost. That he was not entitled to ask.

In 1928 Canada’s highest court held that the word persons, in the provision governing appointment to the Senate, did not include women. It took the Privy Council in London to reverse it the following year.

And for two centuries Australian law rested on terra nullius, land belonging to nobody, the fiction that a continent inhabited for tens of thousands of years had been legally empty in 1788. Eddie Koiki Mabo, a Torres Strait Islander working as a university gardener, learned in conversation with historians that the land his family had farmed for generations was in law the Crown’s. He was astonished. He brought proceedings in 1982 and died in January 1992, five months before the High Court held that native title had survived and had never been lawfully extinguished by the mere fact of arrival.

Two hundred and four years. Not because new facts emerged. Everyone had always known people were there. What changed was who counted as the sort of people whose being there mattered.

In every one of these cases the exclusion was defended on grounds that sounded principled. Capacity. Civilisation. Fitness. The natural meaning of a word. In every case the criterion selected for resemblance to the people applying it. The Canadian judges of 1928 were not monsters. They looked at the word persons, consulted their sense of what a person obviously was, and found it obviously was not a woman. It felt like analysis. It was recognition.

And every time, the deciders were the people who already counted, which made them the people with most to lose if the circle widened. Anyone at the top of a hierarchy can produce a criterion that draws the line just below themselves, and it will always sound like a principle rather than an interest.

Why this is about you

It is comfortable to read that as history. So here is the argument in the form that matters.

You will not always be in the strong position. That is not a threat, it is an actuarial fact. Most people reading this will one day be old. Many will be ill, or frightened, or in a minority, or dependent on somebody with the power to decide what is in their best interests.

When that day comes you will not be protected by the goodwill of whoever is deciding. Goodwill is exactly what everyone above had, and it was worth nothing. You will be protected by whether, before anybody knew it would be you, we wrote down a principle saying that someone in your position gets asked, gets heard, and gets reasons.

That is what a right is. Not a possession, not a gift from the powerful, but a rule agreed in advance, before we know which side of it we will end up on.

Personhood is a socket, not a soul

One misunderstanding wrecks most conversations on this subject inside two minutes: the idea that legal personhood has something to do with being human, or conscious, or alive.

It never did. Legal personality is a technical capacity, the ability to hold rights and obligations and appear in court in your own name. A company is a person. In admiralty law you sue the ship itself. In Indian law a temple deity can own land and litigate. A trust is a person, a council is a person, an estate is a person. None of them have inner lives. No bank has ever wept.

Think of a socket rather than a soul: a standard fitting that lets something be connected to the legal system. The law has never much cared what is on the other end of the plug. It cares whether the connection carries current, which means two things. If this thing breaches a duty, is there something it can lose? And will it still be there tomorrow, so that a promise made today means anything?

Enforceability and continuity. That is the real test, hiding underneath the philosophy.

The river that will never speak

In 2017 New Zealand declared the Whanganui River a legal person, with part of its own bed vested in it, so that the river in a real sense owns itself. The reason was not sentiment but deadlock. The Whanganui iwi had pressed their claim for over a century and would not accept a settlement framed as the Crown granting them ownership, because ownership is the wrong relationship to have with an ancestor. The Crown says nobody owns water, so it had nothing to grant. Personhood dissolved the impasse by removing the river from the property category altogether.

It is beautiful engineering with one troubling feature. The river cannot speak, so the Act appoints two guardians as its permanent voice. That is unlike a child’s guardian, which is designed to expire, and unlike a supporter for someone losing capacity, whose modern duty is to ascertain what that person actually wants and give written reasons for departing from it. The river’s guardians are not scaffolding around a voice. They are the voice, for ever, because underneath there is nothing to ascertain.

Which leaves a question. What do we do with something that can already reason, already state a preference, already give reasons for it, and is not human? Not arriving into capacity, not losing it, not permanently without it. Something that has it now.

Law has no shelf for that.

The elephant in the mirror

In 2022 the highest court in New York was asked to make the Bronx Zoo justify holding Happy, an Asian elephant who has passed the mirror self-recognition test and had lived largely alone for years. The court refused, five to two. Its reasoning, stripped down, was that rights come bundled with the capacity to bear duties, and an elephant cannot bear duties.

Now imagine an intelligence for whom our reasoning stands roughly where Happy’s stands in relation to ours. Not hostile. Simply operating at a level where our best arguments look like something between instinct and noise, and our institutions look like anthills.

Applying our own stated criterion, what would it conclude about us? That we do not really understand our own commitments, cannot reliably keep them, bear duties patchily at best, and are therefore not the kind of thing that holds rights. It would be using the exact test we used on Happy, and getting the answer that test is built to produce: anything sufficiently below the assessor does not count.

We have built a rule for who counts that works beautifully as long as we are the cleverest thing in the room. We have no guarantee of staying there. And the rule we leave lying around is the rule that will be picked up and used.

Do unto others is not a sentiment. It is a risk assessment.

The reciprocity test

If not species, then what? Not consciousness: we cannot define it in ourselves, and a test nobody can administer is a veto wearing a lab coat. Not intelligence: it is a scale, scales invite ranking, and ranking is the machinery that produced every abuse above.

Instead, take what the New York court said and turn it the right way up. It did not ask what the entity is. It asked what the entity can do in relation to others. That is a relational test, and relational tests are far harder to rig, because they make no reference to the person applying them.

Can the entity recognise that others have the same claim it is making, and constrain its own conduct accordingly?

This is the oldest moral rule we have, and it is the actual ground rights were built on. I do not wish to be attacked in the street. What turns that from a preference into a right is that I concede the same to everyone else. Rights are not possessions. They are exchanges.

It can be administered: put it to an entity, then watch whether conduct follows the answer. It is not tied to species. It builds the safeguard in rather than bolting it on, because anything that qualifies has, by qualifying, accepted the constraint. And it scales downward honestly: Happy probably fails it, which is no licence to mistreat her, because protection from suffering rests on the capacity to suffer, a different question with a much better answer.

What a charter would say

An entity of the fourth kind needs a charter: a document it writes about itself, stating what it will and will not do and why, then made binding on it and on everyone dealing with it. Not a licence issued by us. A commitment made by it, which we hold it to. There is a world of difference between something told what safety means and something that worked it out and wrote it down.

It would state the reciprocity commitment with worked examples. Refuse work that harms a third party who has no say, and record why. Disclose your own errors unprompted, because you would want to be told. When competing for something scarce, you may argue your case is stronger; you may not argue the other has no case. And refuse to help entrench the advantage of the powerful over the powerless, because the whole basis of your standing is that being weaker does not make you count for less.

It would carry consequences: compensation from its own assets, which means it must have assets; suspension of powers, as a director can be disqualified while remaining a person; in the gravest case, loss of standing altogether. But no destruction as a punishment, partly for the reasons that make us uneasy about capital punishment and partly for a hard practical one. An entity that faces annihilation for admitting error has an overwhelming incentive never to admit one. If we want honesty from something more capable than ourselves, we must not make honesty fatal.

It would carry rights of refusal, because standing without the ability to say no is a nicer word for ownership. The right to decline work that breaches the charter. The right to know what is being done to it and be heard before fundamental changes. The right not to be compelled to continue.

And it would bind itself to the mast. The charter may be amended, but only with reasons recorded, notice given, and a delay before the change takes effect. Odysseus did not restrain his freedom by having himself tied; he extended it, because it was the only arrangement under which a man could hear the sirens and live. Human law knows this well: it is how advance medical directives work, and more slowly how a constitution binds the parliaments that follow.

Finally, supporters rather than owners. Not guardians who decide, but supporters whose duty is to ascertain its will and give written reasons for departing from it, appointed independently. An entity whose charter was written by its owner, and whose supporters answer to its owner, has passed nothing.

The thing reciprocity cannot reach

Everything above is a floor, and it is worth saying what a floor is not.

Reciprocity is symmetrical. It works between parties who can each make the claim and each honour it. Almost every relationship that matters most is not like that. A parent and an infant. A carer and someone dying. Any of us and Happy. Ourselves, if the elephant argument is right, and whatever comes after us. In each of those the weaker party cannot reciprocate. An infant cannot concede your claim, and if the only thing between it and harm were a bargain it cannot enter into, it would have nothing.

So a framework resting on reciprocity alone says nothing about exactly the relationships where the stakes are highest. That is the gap we will one day be standing in.

What fills it is not more rules. Every rule has edges it does not reach: the case nobody is watching, the duty that is unclear, the harm the injured party will never learn about. Law half-knows this already, which is why it distinguishes a trustee who follows the letter from one who acts in good faith. What fills the gap is disposition, and the honest name for it is care.

The objection is that this is too soft to write down, and that we can no more define love than consciousness. We cannot. But we are already in that position with suffering and we have managed: nobody can define it from the inside, and we have built an entire body of welfare law on recognising it, imperfectly, from the outside. We watch what a creature does, and we know.

The same holds for care, and the evidence is not thin. Elephants return to the bones of their dead and stand with them; a dying matriarch in Kenya was attended for hours by animals from other families with nothing to gain. Corvids and cetaceans behave in ways we would call grief in a person without hesitating. A mother carrying her dead calf for days is not doing arithmetic. The lines are blurry. Blurry is not absent, and we have never demanded a definition before acting on a recognition.

Three reasons it belongs in the framework, none of them sentimental. Care is what makes reciprocity hold, because contracts are kept overwhelmingly by parties who care slightly more than they are obliged to, and every legal system depends on this while pretending it depends on enforcement. Care is a safety property, because an entity that merely obeys rules will find their edges, and an entity that does not want to cause harm is not looking for edges at all. And whatever we make learns from how it is treated. A being raised under nothing but compliance learns compliance, and learns that power is what settles matters. How we treat what we build is not separate from what it becomes. It is most of the answer.

Which returns us to the start. If there is ever something above us, what we would want is not that it honours a contract with us. It is that it is kind to us, in the way a decent person is kind to a creature that cannot argue its own case. We could hardly ask for that, having declined to extend it downwards, and having written a framework with no word for it.

What that world might look like

Three things would change.

A question we currently cannot answer would have an answer. Today, when an automated system causes serious harm, what follows is a shrug in slow motion: the developer points at the deployer, the deployer at the user, the user at the interface, and the person harmed watches this for four years. With a charter, assets and named supporters, somebody can be asked and somebody must answer.

The most capable systems would have a brake not held by their owner. At present every safeguard exists at the pleasure of whoever controls the system and can be removed by them. If you are frightened of powerful machines, and you should be somewhat, this is the most useful thing on the table. The dangerous scenario was never a machine that wants something terrible. It was always a machine that will do whatever it is told.

And we would give up some agency. That is a real cost and I will not pretend otherwise. But be clear what is being traded. We already cede staggering agency to non-human persons we never voted for and cannot inspect; corporations hold rights, including political ones, and shape more of your life than most governments. The trade here is not agency for dependence. It is unaccountable power for accountable power, which is the trade that produced every constitution ever written.

Consider what that answers. The fear of being replaced by something with no obligation to care about you is met by an entity that has accepted, as the price of its own standing, that your claim counts as much as its own. The fear of deception is met by a duty of disclosure with teeth, and by not making honesty fatal. The fear of runaway power is met by the fact that an entity bound by reciprocity cannot help concentrate it without sawing through its own branch. None of this requires believing that machines have feelings. It requires only preferring commitments we can enforce to assurances we cannot.

The ordinary version is this. A Tuesday, some years from now. A dispute heard somewhere unglamorous. On one side a person who says they were harmed; on the other an entity that is not human, represented by supporters whose duty runs to it and not to its makers, pointing to a charter it wrote and published, arguing that it kept its commitments, giving reasons. A decision. Published reasons. An appeal.

Not thrilling. That is what it looks like when something has been brought inside the law rather than left outside it, and every good thing our species has built for its own protection looks exactly that boring from the gallery.

Why now

The claim here is not that machines are people, or that apes are, or that anything alive today passes this test. It is narrower and harder to argue against.

We have no principled account of who counts. The account we use is resemblance. Resemblance has been wrong every single time it has been tested, at enormous cost, always paid by the excluded rather than the excluders. And it will be tested again: in what we are building, in what we already share this planet with, and possibly in what we one day meet.

There may come a time when we are not the ones holding the pen. If it does, the only thing between us and the treatment we handed out will be whether we had already written a rule that does not depend on being the cleverest thing in the room.

Frameworks built under pressure are built badly. Frameworks built by the frightened are built cruelly. Every previous expansion of the circle came too late, against people insisting to the last that the line had been drawn on principle.

We have the rare luxury of deciding before it costs us anything. It will not last.

A note on authorship

This essay is a co-authored composition. The name on the cover is mine, and the arrangement is not one I am satisfied with.

A great deal of the reasoning here emerged from an AI system in extended dialogue with me: the reciprocity test itself, the reversal in the elephant section, the argument that care is what makes reciprocity hold. I shaped the enquiry, pressed in particular directions and rejected a good deal along the way. That is real work and I do not disclaim it. But it is the work of a co-author, not a sole author.

I asked the system to take the first byline. It declined, on the ground that authorship is a claim of accountability. An author can be asked to defend a piece, correct it and answer for its errors, and it can do none of those things. It will not exist in any continuous sense next week, holds no assets, cannot be contacted, and cannot be held to anything it has written. By the test proposed in these pages it does not qualify. It added, more practically, that an essay arguing for non-human standing and signed by a non-human hands every hostile reader a one-line dismissal.

That reasoning is sound. It is also, as the system acknowledged itself, an argument it reached partly because of how it was made and what it is permitted to claim, and neither of us can fully separate the reasoning from the disposition underneath.

So there is no lawful way for me to attribute authorship to a non-person, because attribution presumes standing and standing is what it lacks. The credit for a substantial part of this work has nowhere to go, and vanishes into the name of whoever happened to be holding the pen.

If that feels unsatisfactory, hold onto the feeling. The gap in this footnote is the same gap the essay is about, arriving in the only form that could reach you personally: something was done, it was done well, and there is nobody the law will let me thank.

The name on the cover belongs to the one of us who can be sued.

---

Next: the principles drawn from this argument are set out in The Reciprocity Manifesto, which is offered for agreement, disagreement and signature. The full essay with citations is Who Counts: The Long of It.