Part V — The argument

29. The Five Joints

Each chapter has closed with its own argument: the reason for its rule, the strongest alternative, who bears the cost, and what would change my mind. This Part asks what those arguments leave open when read one at a time. How do the choices fit together? Where does one choice cost another? What commitments sit beneath them all? A consistent arrangement can still be oppressive, impractical or beside the point, and a passing check chooses nobody whose losses matter.

Imagine your household receiving the food it is owed, living in the shelter it is owed, and obtaining the care an ill family member is owed. Someone provides that care, whether through paid employment or an unpaid care role. Then an allegation is made against someone in the household. Whether the allegation is substantiated or defeated, those entitlements remain. This is a hypothetical fulfilment of the promise. It reports no service operating and no food, shelter or care reaching anyone in the world. The formal cases can show an entitlement surviving a finding; they cannot stock this kitchen.

That distinction is the standard for the argument. If the design gives someone a right and no lawful way to seek its fulfilment, it needs correction. If the way exists in law and nobody operates it, the right remains unfulfilled. An accurate description of either failure is still a failure.

The commitments and their instruments need different reasons. Equal standing does not by itself choose two legislative chambers, a collective executive or a particular appointment procedure. Those arrangements need an argument about how they protect people, what they obstruct and whether another arrangement would do the work with fewer burdens, and the chapters give it. Their constitutional status makes alteration deliberate; only the core in Changing the Rules (Chapter 22) lies beyond amendment.

The commitments beneath the choices

Five commitments carry the rest.

What is owed waits on nothing the person must earn. Standing follows a birth or an encounter, before anyone has matched a name to a registry. The floor is a debt to each person, owed on personhood alone. Need decides a real shortage, and productivity, contribution and usefulness are forbidden keys. Work, its refusal and its pay sit above the floor, and a prisoner remains among the persons it is owed to. The lineage runs through Henry Shue's basic rights, the levels of duty the United Nations committee adopted and Amartya Sen's entitlements, and What You Are Owed (Chapter 3) marks where the design parts from them: the committee's duty to provide waits on a finding that a person cannot provide for themselves, and this floor asks nothing of the person owed.

A fact reaches a person only by a route written for it. The record admits only declared kinds of entry. The conclusions that confine, restrict or discredit someone must follow from recorded premises by a named finder. A use of a record is bound to one reviewed holding, a credibility finding restricts the one act it concerns, severity permits a harsher place and never chooses it, and what a person tells a clinic travels to no enforcement office. Philip Pettit calls a person unfree to the extent that another can interfere in their choices at will, even if the other never does.^[1] These routes withdraw the capacity rather than trusting it to stay idle.

No party certifies its own case. Delivery is attested from the recipient's side by a witness other than the provider. A breach of custody conditions is found by people who neither hold the person nor are held. The shield's unrelated-prosecution finding comes from reviewers outside the case, whose eligibility two further offices must show. Appointments come from divided sources, and a scarcity allocation's attesters stand apart from the claimants it compares.

Power is current; answerability lasts. Authority needs a positive record tied to its source and end, and ends when that record does. Answerability survives recall, so the person who exposed an official keeps the protection the exposure earned.

Silence decides nothing. An absent entry is silence rather than a finding. Silence extends no power: custody authority nobody renews lapses, and a declaration's powers end with it. Silence closes no claim: when a responding office stays silent, its duty moves to a separate alternate. Silence approves nothing: an amendment is certified only on positive readings, and a missing review holds a Guardian's pause for no longer than its own record allows. And silence becomes no finding about a person: the freedom that remains is written as duties on a named body rather than read from what is missing.

Elizabeth Anderson asks what people need to function as equal citizens, and allows functionings that need an income to be conditioned on work, which this design refuses.^[2] The five commitments are my answer to her question under that refusal: a person who loses every political contest must remain a member who can live, dissent and seek correction.

What the design adds

None of these ideas is new on its own, and the chapters credit their sources. The book's contribution is carrying four of them through a whole constitution and testing each against the rules as written.

The first is keeping entitlement, duty, evidence and delivery apart as a principle of design. A constitution usually states the entitlement and leaves the rest to courts and administrations. In the case that opens Part I, South Africa's Constitutional Court declared what the state must do and ordered housing for nobody. Here each part has its own rule: the entitlement names what is owed, the duty names who owes it and where, the evidence names who may attest that it reached someone, and delivery remains a fact about the world that the rules never assume. A failure can then be located. A debt with no bearer, a bearer with no evidence route, and evidence that attests a receipt while the service stays unreliable are three different defects, and each has its own remedy.

The second is a set of constraints on how facts flow into consequences. The method names six: inputs closed by name, conclusions nobody may write directly, records read only for their purpose, endpoints nothing reads, no confinement for an absence, and findings bound to their subject and case. The thesis in one line: a right is protected by stating it and also by limiting what the rest of the law may do with facts about the person. These are checks on the written form of rules, and a rule written another way to the same effect can pass them.

The third is testing a design with paired edge cases. The child with nobody is the person public power has not acted upon; the prisoner is the person it has. They fail in opposite directions, and the book runs its rules against both. The pair shows where each depends on someone else acting: a held person is owed humane care, voice and review of the holding by duties that custody itself raises, while the child nobody has come for depends on somebody beginning, which is why the rights advocate owes initiation without a request.

The fourth is a handful of mechanisms: the one thing a sentence takes, which is movement; answerability that survives recall; severity as a ceiling on the place of confinement rather than a selector; controls that defeat a claim of scarcity made from price, budget or refusal; silence that never counts as approval; and the firewall between services and enforcement. Each is argued in its chapter, and each is small enough to adopt elsewhere without the rest.

Designing from the extreme case

The obvious objection to building from the child with nobody and the prisoner is that ordinary life then bears machinery meant for the worst case. An Ordinary Week (Chapter 7) answers it by following one adult through work, a rented home, a clinic, a child at school, an election, a dispute with a landlord and a police stop. Two entries give that adult personhood, the floor and the ballot, and most of the week touches nothing further. The rules engage where a record would decide something about the person, and there they ask for writing before touching: a stop that only identifies a human being restricts nothing, and a search needs a record of its own.

The weight falls on whoever seeks to act against someone. Help takes effect on the record of the one actor who gives it, with an independent reviewer owing prompt review; a restriction, a finding or an appointment giving power over another person waits for its attesters and reviewer. So a landlord seeking an eviction and an office seeking a restriction bear procedure, and a person asking for help mostly does not. The nearest exception is a wage the employer alone attests: it waits for an attester authorised for the worker, who may be the worker, to record the payer, the kind and the basis. Where procedure lands on the person the rules serve more heavily than that, the design has failed in one of the ways Where This Could Fail (Chapter 30) ranks.

Designing from the extreme case also finds defects an average case hides. A floor tested on a registered adult with a home and a job would never have shown that standing must come before identity, or that a request is too much to ask of someone who cannot make one.

Valuation

The first joint is how a society values people without ranking them. The choices meeting here are the floor that ignores contribution, pay and contribution above it, need as the key to a real shortage, a credibility finding that restricts a single act, and separate claims for the environment, the commons and animals. The commitment beneath them is that a person's worth is never a quantity the constitution computes.

History warns against both halves of the problem. Owen's New Harmony community began in 1825 and dissolved in 1827. His later labour exchanges sought a common measure in time, but the hours on their notes incorporated market valuations of labour and materials; calling a unit an hour removed no judgment about unlike products and tasks.^[3] Justin Yifu Lin argues that Chinese collectives retained a right to withdraw until 1958, and that its removal undermined the agreement sustaining cooperation. He presents evidence consistent with that hypothesis rather than a settled account of the agricultural crisis. The constitutional question he raises is whether essentials may be used to make someone accept a judgment of their contribution.^[4]

Pay above the floor is left to democratic law, and communities set it differently. MONDRAGON's industrial beginning dates to 1956; its 2024 annual report records an average workforce of about seventy thousand, and its 2021 account describes a wage scale of six to one between maximum and minimum salaries.^[5] For comparison, the Economic Policy Institute's projected estimate for the largest 350 American firms by sales in 2025 is about 325-to-1, measured against an industry average for production and nonsupervisory workers. The dates, populations and pay measures differ, so the two figures illustrate different distributions and measure no effect of cooperative governance.^[6]

I confer no constitutional status for contribution. People can honour care and creation through expression and association, and a public service can account for the acts it relies on. The strongest alternative is an acknowledgment with no legal consequence, which lets the public see a contribution while owing everyone the same minimum. But a badge acquires social meaning even when the rules never spend it, its absence can stigmatise someone whose work went unrecorded, and withdrawing it for one dishonest act would condemn unrelated teaching or care. Where a wage or supplement rests on false evidence, that claim has its own correction; a general register of esteem adds no remedy for the payer, the worker or a dependant. This choice gives up a common public acknowledgment and leaves neglected work exposed to social disregard, which labour rights, care duties and contestable compensation meet claim by claim. Evidence that a public acknowledgment met a concrete unmet need without becoming a status hierarchy would reopen it.

The joint's trade-off is between separateness and judgment. Keeping claims apart makes officials judge where a single score would calculate. A scarcity manager compares resource-specific need, a court weighs the river's claim and the fish's claim without letting either buy the other, and the river case holds immediate human water and ecological repair as two duties at once. A human floor beneath an ecological ceiling is Kate Raworth's doughnut, whose ceiling draws on the planetary boundaries Rockström and colleagues proposed in 2009.^[7] A single total of human benefit, ecological gain and animal loss would be easier to administer, and it would hide which protection was surrendered and who must answer for it. The cost of separateness falls on the people given lower priority and on whoever depends on an official's judgment being sound; reasons, challenge and an unmet claim recorded as failure bound it. Evidence that the permitted distinctions reproduce social disadvantage would defeat the argument, however specific to the resource they claim to be.

Rotation

The second joint is who does necessary work and who holds necessary office when nobody is compelled to. The choices meeting here are the freedom to refuse and leave work without losing the floor, minimum-service duties that bind a provider or the bargaining parties and never a named worker, authority tied to its source and end, a government removed only by electing its successor, caretaker limits, custody that lapses when nobody renews it, and duties that move to an alternate when the responding office stays silent.

The kibbutzim make the first half a question to investigate rather than a verdict. Abramitzky's account concerns roughly 270 communities and examines shared commitments, selection, monitoring, incentives and exit, and Leviatan reports that by 2011 roughly three in four kibbutzim had adopted differential salaries. Their history shows durable cooperation and pressure to change its terms. It isolates no single cause of reform, and it leaves open whether an unconditional minimum requires equal pay.^[8] The design permits higher pay for difficult work, training, predictable hours and collective bargaining, and public, cooperative and private provision under common rights.

Rotation of public power answers a different problem: an officeholder who can make their own continuation indispensable. Current power depends on its own lawful mandate, scope and end; recall or succession ends authority without ending answerability for its use, and a caretaker holds continuity powers without the ordinary mandate. When an election cannot be held, the electoral body owes it at the first lawful opportunity and the outgoing chamber sits only within caretaker limits until its successor meets.

The two halves share a commitment: nobody acquires a person through their need, and nobody acquires an office through the failure to replace them. They also share a cost, and it falls on third parties. Refusing conscription can leave a service unfilled, and the person relying on it bears the gap. Refusing continuation lets authority lapse, and the person the authority protected bears that: when custody lapses for want of a renewal, the injured person and the public lose protection they were owed. I accept both because each alternative extends a claim on a person through need or silence. Where the design can, it answers with a duty owed before the end: review before custody authority expires, with a predeclared alternate when the ordinary body is certified silent. Staffing it leaves to conditions, recruitment and training, and whether those suffice is the second book's to show. Evidence that services fail at those transitions calls for repair of the continuity route and better conditions; automatic extension would answer a different question.

Coercion

The third joint is what the design does to a person it confines. The choices meeting here are a sentence that takes movement and nothing else, the vote kept outside the sentence, the prisoner kept among persons, severity proved in the case as a ceiling on the place, the shield scoped to the case and dated by the charge, credibility findings restricted to the act, custody that lapses without current authority, and breaches found from outside the walls. The commitment beneath them is that coercion takes one thing, on grounds proved in the case, for as long as current authority lasts.

The chapters' evidence points in more than one direction, and together it explains the shape. In Norway, where criminal cases are assigned to judges at random, Bhuller, Dahl, Løken and Mogstad compared defendants whose sentences depended on the judge they happened to draw: imprisonment reduced reoffending for defendants who had not been working before prison, and those who had been working lost employment.^[9] John Pratt ties the Nordic combination of low imprisonment and humane conditions to the equality of the welfare state,^[10] and Gabriel Chin argues that the restrictions following conviction elsewhere have revived civil death under another name.^[11] Read together, they suggest that confinement under humane conditions can help some people while the losses that follow a conviction can outlast the sentence. The design keeps a bounded power to confine and strips the collateral losses: standing, the floor, the ballot and political voice survive the sentence, and no further loss follows without its own grounds and authority.

The trade-offs inside this joint all fall on someone injured or at risk. The shield and lasting answerability protect a discloser from retaliatory confinement, and a person injured in an unrelated case waits while a finding is sought. Lapse protects against power extended by silence, and a person protected by the confinement loses it when nobody renews. The severity ceiling keeps a prediction from opening the harshest place for a minor offence, and a person exposed to repeated lesser harms gets protection, support and remedies without secure confinement on that ground. In each case I answer the cost with a duty and a separate route rather than with more coercion: review owed before authority ends, protective measures on their own grounds, and compulsory non-carceral remedies. The abolitionist case, which The One Thing Taken (Chapter 27) answers, would remove the power altogether; I retain it for the case where an adjudicator finds that nothing less restrictive can protect another person. That is an argument for keeping the power available; whether any particular use is necessary is for the adjudication in that case.

The pair of stress cases shows what custody adds. The prisoner and the child are owed the same floor, and neither a sentence nor a birth record is evidence of provision. Custody adds duties: humane care, recorded voice, review of the holding and continuity at release. The child's protection is owed by an office that must begin without being asked. A record that stopped changing once a person was held would reproduce what the case opening Part IV found: people held for years because nobody moved their case. The design's answer is authority that must stay current. Custody under a sentence lapses unless renewed, and detention before trial needs prompt judicial review that nobody has to request. A lapse ends the lawful authority without releasing anyone: release is a further act, and a person still held is owed humane care and review of the holding until it happens.

Capture

The fourth joint is an institution that obeys the visible procedure while answering to someone else. It can appoint people with different titles who answer to the same interests, accept a false attestation, or leave a person's encounter out of the record. More signatures can make the account look stronger while making it harder to challenge. The choices meeting here are the record closed by name, delivery attested from the recipient's side, uses bound to their holding, divided appointments that an independently reviewed finding of appointment control can withhold, outside reviewers and witnesses, help that takes effect on one actor while harm waits for review, and the firewall between services and enforcement. The commitment beneath them is that no party certifies itself and every adverse conclusion has a finder who can be named.

Ownership settles none of this. Cooperative, public and private providers face the same labour rights, equality duties and scrutiny, and concentrated private power takes duties when a finding shows a particular dependency, essential function or gatekeeping power. Elizabeth Anderson describes most workplaces as private governments, whose authority over workers they have no say in, and K. Sabeel Rahman argues that firms controlling infrastructure many others depend on should be regulated as public utilities.^[12] The function- specific duties follow that line, and neither author supplies the finding a particular case needs.

Standing attaches before identity reconciliation. That matters because the record problem reaches essentials directly. Drèze and his coauthors report mass cancellation of ration cards in Jharkhand for want of an Aadhaar link, following a March 2017 order, and Muralidharan, Niehaus and Sukhtankar estimate that between one and a half and two million legitimate beneficiaries lost access at some point during the reforms, much of the harm coming from how the transition was managed. Cards and people are different units.^[13] The family whose case opens this book was among those whose card was cancelled. Under this design a failed match stops nothing that personhood alone grounds, and a service may not turn a disputed identity into a reason to withhold the floor.

The joint's trade-offs are between price and barrier, and between speed and review. Separate attesters, a responding office, an alternate, public reasons and a challenge route raise the price of a false record, since each is a named person who can be found out; they leave its possibility, which Where This Could Fail (Chapter 30) ranks among the larger risks. Help taking effect on one actor's record reaches people before review, and help given wrongly binds a provider or public body until review withdraws it. Adverse acts wait, because an adverse act done wrongly is harder to undo than help given wrongly, and help given on one record never becomes the completed record that other rules read.

The last trade-off is between privacy and coordination. A broader linked record could reveal an unmet need without making someone repeat their account at every door. The design binds each record to its purpose instead, and its cost falls on people who need several services, cannot keep explaining their circumstances, or depend on someone else to communicate. The initiation duty and accessible assistance meet part of that cost. A broader coordination route would deserve adoption if it reduced missed care without permitting unrelated reuse, conditioning essentials or weakening correction.

The consequential records remain the constitutional source and the facts given to it. The source can be inspected and its represented consequences executed. A change still requires constitutional justification; passing a repository check is not democratic approval. Whoever operates the system must never turn a successful calculation into authority to enact its inputs. Explicit grounds, separated functions and contestable records make a consequence answerable to something more precise than an official's discretion. They leave false evidence false and an unwilling institution unwilling, and leaving those judgments to trusted professionals needs evidence and challenge too. The explicit route is justified only where it makes refusal and correction more accessible without making a complete file the price of help.

The State

The fifth joint is what kind of state carries the duties. This proposal specifies a republic, not a community exempt from public law. History offers no single answer about the state's role. Eden Medina's account of Chile's Cybersyn describes an economic coordination project using an existing telex network under Allende's government; the coup ended it, so its later performance is unknown.^[14] Auroville's governance sits inside a statutory framework: in Auroville Foundation v. Natasha Storey in 2025, the Supreme Court of India held the Governing Board entitled to constitute its development council without reserved places for the Residents' Assembly, distinguishing consultation on the master plan from authority to implement it.^[15] Switzerland's WIR, founded in 1934, is a complementary credit arrangement inside a national legal and monetary system, and its persistence identifies no single reason for the state's tolerance.^[16] Kerala's People's Plan, launched in 1996, shows a state enabling local decisions: Franke's account describes devolution of a third or more of the development budget and training for more than 100,000 people, evidence of an organised programme rather than of what any one feature caused.^[17] Elinor Ostrom found that long-enduring commons shared rules fitted to local conditions, collective choice, monitoring and accessible conflict resolution, observed practices rather than guarantees.^[18]

The criteria for the state's institutions are equal political voice, the ability to decide and to replace a government, protection against concentrated power, and continuity of the rights the state owes. The chapters argue each institution, and the pattern across them is how I resolve the conflicts between those criteria. Where continuity of rights meets another criterion, continuity wins: the core lies beyond amendment, a crisis suspends no right, and the floor takes priority over any fixed limit on borrowing. Even the ability to replace a government yields to continuity, since a government falls only when its successor is chosen. Where protection against concentration meets identifiable responsibility, protection wins: the executive is collective, the common government is divided among tiers, and reviewers come from divided sources. Equal voice gives way at one point: a territorial chamber, whose objection to ordinary law is suspensive.

Each resolution has a cost with a name. The voter looking for whom to punish finds a council; the person sent between tiers finds each denying responsibility; a later majority finds the core closed to it; a region finds its weight unequal. The duties that bound those costs, a named bearer for each floor duty, continuity passing up a tier on a certified failure, reasons and challenge, are argued where they apply.

Nor does the book's own democracy and wellbeing calculation settle any of this. In the bundled data for 141 equally weighted countries, the correlation between electoral democracy and average life evaluation falls from about 0.5 to about 0.2 after adjustment for log income per person. Income might help explain both; democracy might affect income; other causes might affect all of them. Adjustment cannot decide among those accounts, and controlling for something democracy changes can remove part of the effect one hoped to measure.^[19] The group averages of closed and electoral autocracies are close together in that snapshot, while electoral and liberal democracies are further apart; different countries compose each group, so the difference estimates nothing about a country changing its government. The exploratory dispersion calculation is narrower still. Absolute residuals mix departures above and below a fitted line of country means, and measure neither the worst lives within countries nor a guaranteed lower bound. The estimated association with electoral democracy is negative after income adjustment. It is one estimate from one snapshot: another index could return a nonsignificant estimate, and the contrast between a significant and a nonsignificant result would not itself show a difference between them, as Gelman and Stern explain.^[20] Comparable observations over time, defensible causal assumptions and measures of people's worst circumstances would support a stronger investigation.

My reason for equal voice is that people subject to public power should have an equal part in directing it. My reason for the floor is that needing essentials should give no one a means to demand obedience, Pettit's non-domination again. These are commitments to defend, and the country data hides no causal finding behind them.

The proposal leaves an operating burden to meet. No provider form, household relation, contract or delegated function removes the public responsibility. Offer, accessible provision, receipt, challenge, continuity and remedy remain distinct requirements. A recorded duty does not perform the next act, and independently attested receipt does not show a service system's capacity. The task of What It Would Take is to make an operating account answer those demands with resources and evidence. It cannot turn a constitutional defect into a staffing question or make an unmet entitlement count as fulfilment.

What the argument asks

These arguments support choices with specific costs, and they add up to no empirical finding that the whole society works. The constitutional claim is that a person does not have to earn standing or essentials and that public power must answer for the burdens it imposes. Its means remain open to challenge where a better arrangement can protect the same commitments.

The named people in the formal cases have no recorded inner lives. Giving Nell fear or Hano relief would add evidence the cases do not contain. Their limited records test consequences; they neither portray complete human lives nor stand in for people who have lived under this proposal. The kitchen is imagined. The historical cases are documented experiences under other arrangements, and none offers testimony from a society operating these rules.

The claims registry records the empirical claims, sources, dates, units and caveats. The democracy calculation includes its script and bundled inputs so a reader can repeat it. The registry's claims are public domain; source publications and the data snapshot keep their own terms. A locator makes evidence inspectable; it leaves its truth to be judged. Different denominators and disputed findings remain part of the argument.

For the child who cannot initiate a claim, the test is help begun without a request and provision that reaches them. For the person confined, it is a restriction with lawful limits, retained rights, review and actual release when authority ends. A rule can state those demands and expose a consequence that violates them. A person still has to meet them.

The next chapter ranks the ways this design could fail, and what would show each failure.