The State as a System of Value Flows: A Value-Flow Theory of Governance and the Design of Axiacracy

Full text

1 Introduction

1.1 The rupture

For about two centuries the modern state has rested on one working assumption: human labour is the basis of economic value. Taxation, social insurance, labour law, mass education and pensions were built on it. The industrial loop ran from labour to income, from income to consumption, from both to tax, and from tax to redistribution. A citizen's claim on the social product was routed mainly through a job, and so was the state's. Work was at once the distribution system, the fiscal system and much of the system of social standing.

Artificial intelligence weakens this assumption. When cognitive as well as physical work can be automated, output can keep rising while human participation in production falls. Earlier automation waves displaced particular tasks, and displaced workers moved to tasks that machines could not do. The working hypothesis of this paper is that AI differs in scope: it can reduce labour demand across sectors rather than in one sector at a time. If that hypothesis holds, the loop breaks at several points at once. Labour income falls, consumption out of wages weakens, the wage-based tax base erodes, and value concentrates around the owners of the scarce complements to automated production: compute, models, data, energy and network position.

The classical economists described parts of this mechanism long before AI. Mill observed in his Principles of Political Economy (1848) that a demand for commodities is not a demand for labour. Marx, in the third volume of Capital (1894), described a rising ratio of machinery to living labour and showed that aggregate wealth can grow while the share flowing back to households as wages shrinks. Keynes, in The General Theory of Employment, Interest and Money (1936), argued that a rich economy can settle in a stable equilibrium below full use of its capacity. Together they describe a state asked to fund rising obligations from a shrinking base, amid physical plenty.

The second problem is one of perception. The main instrument of economic governance, gross domestic product, records the money value of marketed output. It does not register sign: whatever is paid for counts as product. A platform can add large market value while degrading attention and weakening communities, and the instrument reports growth. In Donella Meadows's terms (Thinking in Systems, 2008), it sees inflows and misses depletion. A state that reads only money cannot see the variables that the AI transition puts at risk: concentration of value, the erosion of the educational and cognitive base, and the loss of meaning that work used to provide.

1.2 The research question

This paper asks one question: can a state be governed by sensing and rebalancing value flows across a society, instead of by commanding outcomes or by taxing labour and redistributing money? The question has a descriptive part and a design part. The descriptive part asks whether a state and its society can be represented, with acceptable confidence, as a network of multidimensional value flows. The design part asks whether a mechanism built on that representation can correct harmful imbalances without directing private life, without central planning, and without becoming an instrument of capture.

The proposed answer is a form of government named Axiacracy (from the Greek axia, value, and kratia, rule). Its foundations are taken as given here: balancing value and anti-value flows; rule-bound AI agents in place of much discretionary work by officials; transparency of rules, readings and corrections; and an architecture designed to resist capture. The paper states these foundations precisely, derives the mechanisms that follow, identifies how they could fail, and proposes a method for testing them.

1.3 Contribution

The paper makes the following contributions.

• Section 2 sets out a value-flow theory of the state based on the Theory of Value Management (VMT): a substrate of five constructs, society as a graph of value transformers, ten value axes, and the state as an orchestrator that corrects rather than directs.
• Section 3 treats measurement under uncertainty and plural value frames: confidence-bounded readings, cohort aggregates in place of personal scores, and the role of the citizens' vote in making value objective for governance.
• Section 4 describes the state apparatus as value flows, using a qualitative survey of eleven states and a structural model of state value flows.
• Section 5 specifies the governing loop (sense, diagnose, correct, re-sense) and the imbalance function that drives it.
• Section 6 develops distribution: the fiscal base, the commons, the capability floor, the corridor band and ceiling, and the citizen dividend.
• Section 7 sets constitutional limits: the Charter, the conditions under which coercion is legitimate, and justice understood as rebalancing.
• Section 8 catalogues failure modes, including capture, gaming of sensors and over-steering, and the defences designed against each.
• Section 9 places the design in its intellectual lineage and records the main critiques it must answer.
• Section 10 proposes a simulation testbed, the MOS testbed, as the method for testing the design.
• Section 11 lists open problems and a research agenda.

1.4 Status of claims

This is a theoretical and design-science paper. It proposes a theory and a design; it does not report empirical or simulation results. No simulation has been run. The MOS testbed described in Section 10 exists as a specification only, and no number in this paper should be read as an output of it.

Three kinds of material appear in the paper, and their status differs. First, definitions and design rules (the substrate, the axes, the corridors, the governing loop, the constitutional limits) are proposals. They can be judged for internal consistency and for fit with the literature, but not yet for performance. Second, the structural model of state value flows in Section 4 (52 nodes, 97 flows and 74 value and anti-value elements) is a structural map whose magnitudes are placeholders. It shows what flows exist and how they connect, not how large they are. The eleven-country material is raw, unclassified desk research and is used as a qualitative survey, not as a dataset. Third, a number of claims are explicitly hypotheses, among them: that AI will reduce aggregate labour demand rather than shift it between tasks; that the ten axes can be sensed at cohort level with confidence bounds narrow enough to act on; that the governing loop converges rather than oscillates; that rent on compute, data and land can fund the capability floor and the citizen dividend; and that the anti-capture architecture holds under adversarial pressure. Each is marked where it occurs, and Section 11 collects them.

2 A value-flow theory of the state

2.1 Foundations from the Theory of Value Management

The theory used here rests on the Theory of Value Management (VMT), a descriptive theory of how value and anti-value arise, move, are consumed, transformed, accumulated and lost in social and economic systems. VMT gives a language for analysis, not a set of prescriptions. Axiacracy is a normative design built on top of that language.

In VMT, value is the potential of a system to improve its ability to reach its goals. This definition has two consequences that matter for a theory of the state. First, value is always relative to a system and its goals: there is no "value in general", only value for some value system. Second, value is a potential, not a stock of goods. A school, a trained engineer or a trusted court is valuable because it raises the capacity of a society to reach the goals that society holds.

Anti-value is the mirror category: a change that reduces a system's ability to reach its goals. VMT treats it as an independent category, not as the absence of value or as negative value. A single flow can carry both at once. A factory ships goods and effluent; a social platform delivers connection and erodes attention in the same transaction. A scalar measure nets these against each other and loses the fact that both are real.

A value flow is a directed interaction between a sender and a recipient through which value or anti-value passes. A value system is the frame in which flows are interpreted. VMT defines it by five components: the actors, the flows among them, the criteria of improvement and degradation, the rules of interpretation, and the time horizon. The boundary of a value system is set by the choice of actors and flows, and changing the boundary can change the conclusions without any change in reality.

A value transformer is a role, not a kind of entity: an actor plays the transformer role when it consumes value and produces new value. The points at which value or anti-value is converted from one form into another are called converters in the applied methodology built on VMT; in this paper the term transformer covers both the actor in that role and the conversion it performs. VMT adds two further concepts used throughout. Flow resistance is the gap between the value a sender plans to deliver and the value a recipient actually realizes; it accumulates over time. Leakage occurs when realized value leaving a system persistently exceeds the value entering it. Finally, each actor has a participation condition: a threshold below which it leaves the system or changes the form of its participation. A system is sustainable only if the participation conditions of its actors hold.

Three VMT axioms matter most here: value is defined only within a value system; one reality can carry different values in different systems; and value is not conserved. Together they rule out a single, state-owned measure of "civilizational value".

2.2 The value substrate (V1 to V5)

Axiacracy translates these foundations into five constructs that form the vocabulary of the whole design.

• V1 Value transformer. The universal node. Any entity can be modelled as consuming, transforming and creating value, or generating anti-value: a parent, a firm, a ministry, a city, an AI agent.
• V2 Multidimensional value, the Æ-vector. Value is read as a vector across ten axes, with a parallel anti-value vector. GDP becomes one axis of ten. The axes are a shared vocabulary with partial commensurability; they are not claimed to be reducible to one another. The weights across them are an open political choice, not a metaphysical exchange rate.
• V3 Value flow with a lifecycle. The edges between transformers carry value through three stages: planned (the producer's claim), realized (the recipient's experienced value) and retrospective (a later re-scoring when new evidence arrives). The gap between planned and realized value is flow resistance; persistent net outflow is leakage. The lifecycle is load-bearing, because governance acts mainly on realized and settled value, not on claims.
• V4 Anti-value. Harm is a first-class category. It is classified by scope (a single edge or the whole system), by detection lag (immediate to generational) and by kind (rights violation, malicious harm, negligence, self-directed distress).
• V5 The sovereign value frame. Each citizen, family and business, and the state itself, is a value system with its own criteria. The polity is a federation of frames, not a hierarchy with one frame at the top. The state is one frame with a bounded mandate and never the owner of all value. If an actor's participation condition fails, this is recorded as an alarm on the state, not as a fault of the actor.

One distinction keeps the substrate precise. Following North (Institutions, Institutional Change and Economic Performance, 1990), the Charter, the weights, the corridors and the charge rules are institutions, the rules of the game. The transformers acting under them are organizations, the players. Organizations that benefit from current rules have an incentive to rewrite them, so capture is a normal pressure that must be designed against.

2.3 Society as a graph of value transformers

The central reframe of the theory is to stop reading society through categories that presuppose the wage channel (taxpayers, employees, corporations, consumers) and to read it as a graph of value transformers. Citizens, families, firms, AI systems, scientific institutions, cities and infrastructure are nodes. The edges are value flows: goods and money, but also knowledge taught, health sustained, trust built or spent, attention consumed and ecosystems drawn down. Governance, on this reading, is the coordination and rebalancing of flows among transformers, measured across several dimensions rather than through price alone.

The graph is a system of stocks and flows in Meadows's sense. The levels on each axis are stocks that change only through flows, and only gradually. The governing loop is a balancing, goal-seeking feedback; concentration and vicious circles are reinforcing feedbacks. Two consequences follow. Persistent imbalances are outputs of the structure of flows, not exogenous events, so the state corrects structure rather than chasing events. And reducing an outflow from an axis can matter as much as increasing an inflow, although growth metrics register only the latter.

The graph is a lens, not a licence. It makes visible what the money lens hides, but the state may act on it only where it can measure with stated confidence and where markets fail to account for an effect. Smith, in The Wealth of Nations (1776), discharged the sovereign from directing private industry; the graph does not revive that duty.

2.4 The ten axes

The Æ-vector reads value on ten axes. The list was set by design, not derived from first principles, and it is revisable. Each axis tracks value generated by some class of flows and carries a mirror anti-value. The epistemic axis was added to the original nine because the integrity of shared knowledge, including resistance to propaganda, is treated as a condition of every other reading.

Axis	What it tracks	Example anti-value
Economic	Production, trade and services; the familiar money axis, retained in full	Value concentration, displacement without transition
Educational	Teaching, mentorship and transfer of skill	Credential inflation, deskilling
Scientific	Research, discovery, replication and the growth of tested knowledge	Fraud, irreproducible results
Demographic	Parenting, care and household formation	Depopulation, overload of carers
Cognitive	Deep work, sustained attention and problem-solving capacity	Attention erosion, engineered dependency
Infrastructural	Resilience of energy, logistics, compute and housing	Fragility, single points of failure
Social stability	Cohesion, trust and peaceful resolution of conflict	Polarization, unrest
Ecological	Restoration and low-footprint production	Emissions, resource depletion
Meaning	Purpose, belonging, culture and identity	Anomie, loss of meaning, despair
Epistemic	Verified knowledge, calibrated expertise and provenance of information	Propaganda, misinformation, opacity

Two constructs are deliberately not axes. The design rule is that an axis is something traded off through voted weights, and what must never be traded off is a constraint, not an axis. Distributional fairness is therefore not an axis: it enters as a floor-protected constraint and a guardian function, so that a majority cannot vote it down. Solidarity is not an axis either: it is a type of flow, in which donated value is credited on the meaning and social stability axes.

Each axis carries a corridor with three thresholds, defined per cohort and per region: a floor (a constitutional minimum), a target band (where the state keeps the axis and where voted weights steer emphasis) and a ceiling (above which marginal value is discounted, so that no axis is over-optimized). Citizens set the relative weights of the axes by vote. Floors are protected from ordinary votes. The mechanics belong to Sections 5 and 6. Here it is enough that the vector is never a single score and never a personal score: readings are cohort aggregates with confidence bounds.

2.5 The state as orchestrator: correct, do not direct

On this theory the state is neither a tax office nor a planner. It is an orchestrator of value flows with a bounded mandate. Its tasks are to sense the condition of the graph across the ten axes, to diagnose imbalances against the corridors the citizens have set, to correct them, and to sense again. Much of the routine work of sensing and correction is assigned to rule-bound AI agents rather than discretionary officials, with the rules, readings and corrections published.

The governing principle is "correct, do not direct". The state may read and publish an imbalance. Its default remedy is a change to general rules available to all: a price on a measured harm, a right, a disclosure. It may not direct a named actor's capital into a named use. This is stated as a bright line: the state may set the price a factor of production faces, but never its quantity, employment or allocation. The line follows Mises's argument in Human Action (1949) that economic calculation depends on private owners choosing among uses. After a correction, markets re-clear allocation freely. The sequence is see, price, release; it is never see, allocate.

Coercion is narrower still. The design separates a wide indication layer, in which readings are published to inform private and public choice, from a narrow charge layer. On the question of objectivity the theory holds a fixed line. Individual value is subjective. The citizens' vote makes value objective for governance: it is a declared collective choice, not a measurement of inner utility. The effects of designated harms are measurable with stated confidence. Coercion is legitimate only for a democratically designated harm with a countable effect, even where that harm has no market price; where a market price exists, it calibrates the size of the charge but does not decide what counts as harm. Where an effect cannot be counted, the state publishes and persuades but does not compel.

Two further limits complete the orchestrator role. First, the state is one frame among many (V5), so it may not dissolve or overwrite a functioning local order. Following Ostrom (Governing the Commons, 1990), its default toward a working commons is to recognize and back the local rules, not to replace them. Second, every positive intervention carries a sunset and a self-sufficiency test, so that correction does not harden into direction. Whether such a state can sense well enough, correct without oscillation and resist capture is not settled here; these are hypotheses taken up in Sections 5, 8 and 10.

3 Measurement under uncertainty and many frames

3.1 What the state senses, and at what grain

A value-flow theory of the state depends on one question: can value be measured honestly enough to act on? This paper does not claim that measurement is easy. It makes the limits of measurement part of the constitution. Smith is the standing warning: he took labour as the true measure of value, failed to measure the real value of corn across centuries, and ranked certainty above equality among the maxims of taxation (Smith, The Wealth of Nations, 1776). A ten-axis vector is not a richer instrument by default. It is ten measurement problems, most of them less observable than a price.

The design holds one fixed position on objectivity. Individual value is subjective: a judgment of value arranges options in order and does not measure them (Mises, Human Action, 1949). The citizens' vote, when it sets axis weights or designates a flow as anti-value, makes that valuation objective for governance; it is a binding collective choice, not a discovered fact. The effects of a designated flow, such as vacancies eliminated or tonnes emitted, are measurable with stated confidence. Coercion is legitimate only for a democratically designated harm whose effect can be counted, even where no market price exists. Measured effects, voted weights and money prices are three registers that are never silently converted into one another.

Sensing is aggregate by construction. The default view of persons is the cohort, protected by anonymity floors. Reading one identified person's data requires consent or a proportionate, logged warrant, and the read is recorded as an event the person can see. There is no personal value score and no behavioural rating; the axes are dimensions of measured effect, never grades of people. Individual liabilities and dividends rest on a few simple, hard-to-conceal bases, chiefly rent and luxury consumption, while the full vector serves collective steering only. Every reading carries a confidence bound, and non-monetary value is reported as a band, never as a false precise number. An axis may carry coercive weight only if it has a falsifiable definition with a published error bar, a settling mechanism (a market, a staked prediction market or contested attestation), out-of-sample validation, and a sensor separated from the beneficiary. Otherwise it stays in the indication layer, which publishes signals and charges nothing.

3.2 Goodhart and the anti-gaming guards

A measure used to steer becomes a target, and a target is gamed. The sensor is therefore built adversarially, and three guards on the governing loop carry most of the load.

• The map is not the ground. The value-flow map may never be treated as the definition of the value it represents. Each axis carries a parallel lived-sentiment channel from the people affected, and rebalancing is validated against it. The map governs hypotheses; lived experience governs legitimacy.
• The lived-benefit gate. An intervention justified mainly by index improvement rather than by demonstrable benefit to an identified cohort is flagged for human review. Every steering action must name the people it helps.
• The structural vs cosmetic test. Pricing an externality, raising a floor or paying a dividend counts as relief, but is credited as structural only if it also reduces the capacity for unearned accumulation. An intervention that improves a distress metric while leaving the concentration engine intact is recorded as cosmetic and does not discharge the imbalance.

Further rules follow. The measure must never become a producing unit's output target; corrections price an externality or rent and never specify a quantity. A cohort's low reading is treated as a hypothesis of structural extraction to be traced to its beneficiary, not a fact to be priced around (Acemoglu and Robinson, Why Nations Fail, 2012). Calibration authority is separated from beneficiaries, since extraction can happen by moving the meter. Monitors answer to the people monitored, and cheap, tamper-resistant proxies are preferred to gameable ones (Ostrom, Governing the Commons, 1990). Near-unanimous consensus raises scrutiny rather than confidence.

3.3 Epistemic humility

A state that scores value will optimise what it can name and miss what it cannot. The design answers with structural humility. A permanently unallocated reserve of steering weight stands for value not yet named; an unsteered zone lets new activity mature before it is scored; concentration caps limit how far any intervention may pull capital into one axis; and a thin set of protections against cruelty sits above the optimiser, beyond the reach of aggregate-utility arguments. Resilience, in Meadows's terms, is redundancy of restorative loops, and efficiency that removes them breeds fragility (Meadows, Thinking in Systems, 2008). The ten-axis vector is itself a chosen map, open to revision, and the state publishes confidence bounds on its own readings and reports its correction failures. North gives the deeper reason: measuring the valuable attributes of exchange is costly, and institutions exist to economise on that cost (North, Institutions, Institutional Change and Economic Performance, 1990). Some value will always remain unmeasured. The epistemic axis, following Mill, scores the health of inquiry, never the truth of propositions (Mill, On Liberty, 1859).

3.4 Many value frames and no supreme ledger

The second limit is not technical. In the Theory of Value Management, value is defined only within a value system (actors, flows, criteria, interpretation rules, time horizon), and one reality yields different values in different systems. A citizen, a household, a firm, a town and the state are each such a system. Moving warehouse work to nights is a gain for the firm, a loss for a parent of young children, a gain for a student with free days, and mixed for the town. All readings are honest; no arithmetic merges them.

Three consequences follow. Value-in-general does not exist, so summing frames into one state-owned scalar is a category error. Value is not conserved across contexts. Systems meet only at their boundaries, so the state sees only what crosses into its frame. The polity is therefore modelled as a polycentric federation of peer value systems, and the state is one frame with a narrow mandate (floors, cross-boundary harm, long-run balance), not the integrator of all others. The ten axes are a shared vocabulary for partial comparison, and the voted weights are the state frame's weights, not "the" weights. Hayek's knowledge problem reads the same way: much relevant knowledge is tacit and cannot be handed over (Hayek, The Use of Knowledge in Society, 1945), so private frames are kept local rather than read.

3.5 Inter-frame routing and graded enforcement

What replaces a supreme ledger is a search. Inter-frame routing looks for flows that improve each participant's position as that participant's own frame judges it, with no scalar maximised and no exchange rate between one person's loss and another's gain. The search can fail, and then nothing happens. Participants disclose only the minimum projection of their frame a transaction needs. If a correction advances the state frame while participants in the affected flow report themselves worse off, it is paused, narrowed or contested before an independent tribunal. The state may publish demand, supply a floor, price a demonstrated cross-boundary harm or remove a bottleneck; only pricing harm is coercive.

Enforcement is graded by one rule: enforcement strength = measurement confidence x inter-frame scope. High-confidence, countable harm that crosses directly into another frame, or a floor breach, may be compelled. Low-confidence or intra-frame matters receive soft power only: publication, incentives, visibility. Confidence is necessary but never sufficient. This operationalises Mill's harm principle, and coercion fires only on a manifested act, never on sensed intent. Whether the rule can be applied consistently at scale is an open question for the MOS testbed.

4 The state apparatus as value flows: a qualitative survey of eleven states and a structural model

4.1 Why look at the state as value flows

The usual picture of a state is an organisation chart: ministries, departments and directorates. It records who reports to whom and says little about what the apparatus delivers to whom. This paper adopts the opposite starting point, set out in the third volume of the Axiacracy trilogy under the title "the apparatus as value flows". The founding drawing shows flows, not offices: value entering from nature, trade and citizen contribution; unearned rent pooling in a commons fund; a citizen dividend and services returning to citizens; harm priced back onto its source; and citizens steering the circulation through votes, needs and attestation.

In this view a flow-domain is a direction of value the state must keep flowing. The flow-domains are the ten axes plus cross-cutting domains (the fiscal commons, external relations, defence, emergency and justice). The bodies that service them are custodians bound to a named flow, replaceable and without constitutional standing of their own. The three branches become functions over flows: the legislative sets priorities (weights and floors), the executive senses, finances and delivers value, and the judiciary rebalances flows that crossed a harm threshold. The administrative layer is proposed as a federation of rule-bound public AI agents, each holding a narrow, revocable, time-bounded and published mandate for one flow; citizens keep the weights, floors and mandates, and courts stay human. The trilogy presents this agent architecture as a wager to be tested rather than a solved claim, and this paper treats it the same way. Meadows's principle that structure drives behaviour gives the reason for the reframe: a hierarchy of offices tends to defend offices, while an apparatus defined by flows can be judged by whether the flows reach people (Meadows, Thinking in Systems, 2008). The approach also changes the analytic unit. Instead of comparing institutions, one can compare what passes between the state and each of its counterparties, in both directions, including value that carries no price.

4.2 A qualitative survey of state value flows in eleven countries

To ground the structural model, a desk survey collected value flows between the state and its counterparties in eleven countries: Russia, the United States, Germany, Sweden, Singapore, Estonia, the UAE and other Gulf states, China, India, Brazil and Bhutan. The material is raw, unclassified desk research. Flows are organised only by direction and, for citizens, by cohort; no taxonomy of value types has yet been applied, no magnitudes were collected, and coverage is uneven between countries. What follows is a qualitative description of patterns, not a measurement or a ranking.

The survey uses five directions.

• State to citizen, by cohort. Eleven cohorts were used: infants and young children, school children and students, working-age adults, pregnant women and new parents, pensioners, people with disabilities, the unemployed, veterans, low-income people, large families, and a residual group (rural residents, refugees, orphans, farmers and others). Flows recur in a small number of forms across all countries: cash transfer, in-kind service, risk-pooling insurance, price subsidy, legal right or protection, infrastructure and access, identity and information, guarantee, forced-savings account and dividend. They are also life-cycle-conditioned, following the citizen from birth registration to survivor pension.
• State to business. Subsidised and guaranteed loans, research grants and tax credits, procurement with set-asides, tax regimes and special zones, export credit, wage subsidies, legal and intellectual-property infrastructure, free statistical data and financial-system backstops.
• Citizen to state, beyond taxes. This direction is overwhelmingly non-monetary: labour, unpaid care, demographic reproduction, conscription, jury duty, voting and legitimacy, census participation, personal data, voluntary compliance, volunteering, donation of blood and organs, public-health behaviour and trust.
• Business to state, beyond taxes. Employment, unpaid tax withholding, mandatory reporting, regulatory compliance, operation of critical infrastructure, defence-industrial capacity, export earnings, anti-money-laundering monitoring and lawful-interception capability.
• Other parties. Two-way flows with civil society, other states and alliances, international organisations, regional blocs, sub-national governments, nature, academia, religious organisations, the diaspora, foreign investors, media, bond markets, the central bank, indigenous authorities and organised labour. Remittances, foreign investment and bond markets confirm that the external boundary is real and large.

The table lists salient patterns per country as they appear in the survey. Entries are illustrative examples taken from the collected material, not complete inventories.

Country	Salient state-to-citizen and state-to-business flows	Salient flows to the state and architecture noted
Russia	Means-tested unified child benefit; one-off birth grant; lump-sum family capital usable for housing or pension; large-family benefits (utilities discount, mortgage payoff on a third child, land plot); northern wage supplements; subsidised family and sector mortgages; small-business guarantee programmes	Twelve-month conscription; residence registration; digital state-services identity; product-tracking and online cash-register data feeds; lawful-interception capability; mandatory foreign-currency repatriation; import substitution; social upkeep of single-industry towns
United States	Child Tax Credit and EITC; SNAP and WIC food support; Medicaid, Medicare and Social Security; veterans' health system and GI Bill; small-business loans, SBIR research grants and procurement set-asides	Jury duty; Selective Service registration; whistleblowing incentives (qui tam); employer AML and sanctions monitoring; lawful-interception capability; company pensions reducing future welfare load
Germany	Universal child benefit; parental allowance; legal right to a daycare place; student grants (BAfoeG); short-time work allowance; long-term care insurance; a social insurance fund for artists	Lay judges; public-media fee; dual vocational training by firms; co-determination and works councils; chamber membership; levy for not employing disabled people; church-tax collection by the state
Sweden	Universal child allowance; 480 days of parental benefit with a partner quota; capped preschool fees; free higher education; active labour-market services	Nordic universalism: unconditional universal benefits, high tax and near-total data legibility; value returned mainly as breadth of service rather than cash
Singapore	Baby Bonus with a matched child savings account; individual CPF accounts for housing, health and retirement; public flats bought with CPF; GST vouchers for low-income households	Forced savings and self-reliance, minimal unconditional transfer; conscription; citizen savings help fund the sovereign investors
Estonia	Digital identity, interoperable data exchange, once-only data collection and internet voting, including services for young people	Digital state: near-total digital legibility (almost all tax filing online); reciprocal input is data trust; the state also offers its apparatus to non-residents (e-Residency)
UAE and Gulf states	Birth and marriage grants; free housing or land; subsidised utilities; guaranteed public-sector jobs financed by resource rent	Rentier and distributive model with no income tax; reciprocal flow recorded as political loyalty under closed citizenship
China	Minimum-income guarantee (dibao); rural social pension	Services gated by household registration (hukou); social credit described as a behaviour-for-privileges information flow; migrant labour subsidising the system
India	Guaranteed 100 days of rural employment (MGNREGA); public food distribution; farmer income support and health insurance	Biometric identity rail (Aadhaar) with direct benefit transfer; reciprocal flow is biometric legibility
Brazil	Bolsa Familia, a cash transfer conditional on school attendance and vaccination	Shares the direct-transfer architecture noted for India; reciprocal flow is conditionality
Bhutan	Universal health care and vaccination; royal scholarships; discretionary royal welfare (Kidu)	Wellbeing measured as Gross National Happiness across nine domains; reciprocal flow recorded as loyalty and environmental and cultural stewardship

Two observations are offered with the same caution as the material. First, the same needs are met by very different flow structures, from Nordic universal service to Singaporean forced savings to Gulf rent distribution. This is the design space in which a value-flow state would sit. Second, much of what the state receives from citizens is non-monetary and absent from tax and national-accounts data. That is the gap a multi-axis Æ vector is meant to address. It is a motivating observation, not a finding about the size of the gap.

4.3 A structural model: the VSS map of the state

From the survey, a structural model was drawn in a Value System Schema (VSS): a directed graph of nodes, flows and the value elements each flow carries. The map is a model of a value-flow-governed state, not of any of the eleven countries. It has 52 nodes in six layers:

• 5 external nodes: nature and subsoil, foreign markets, other polities and international organisations, the diaspora, and foreign investors;
• 2 commons nodes: land, compute and data as a commons, and the Commons Fund as fiscal hub;
• 15 society nodes: the 11 citizen cohorts of the survey, households, and three business types (small and medium firms, large technology and AI-infrastructure firms, and agriculture);
• 17 executive ministries, one per flow-domain, from Commons Fund and Fiscal to Emergency and Civil Protection;
• 6 governance nodes: the legislature, the judiciary, municipalities and 3 guardians (Equity and Ombudsman, Standards and Measurement, Audit and Integrity);
• 7 other parties: civil society, academia, religious organisations, media, unions, financial markets and the central bank.

The map has 97 flows: 86 internal, 8 entering from outside and 3 leaving. Nine internal flows are control flows rather than value flows: weights and floors from the legislature, accounting rules and confidence from Standards and Measurement, floor enforcement and equity oversight from the Ombudsman, and measurement audit from Audit and Integrity. Flows carry 74 elements, each tagged with an axis: 70 value elements and 4 anti-value elements. All magnitudes are placeholders. Every flow carries the same nominal value of 1 in a relative unit, and every element a quantity of 1, pending quantification. The map therefore describes structure only: who is connected to whom, in which direction, and on which axis. It says nothing about the size of any flow.

4.4 What the map makes visible

Even without magnitudes, the structure raises questions. The following are stated as hypotheses for empirical work and for the MOS testbed. They are properties of a draft model, not findings about any state.

1. Anti-value is concentrated, and may form a reinforcing loop. All four anti-value elements attach to one node, the large technology and AI-infrastructure firms: a concentration charge to the Commons Fund, an ecological externality, an attention externality reaching the Cognitive Health ministry, and a judicial clawback. The same node also receives research procurement and access to the compute commons, and supplies data to it. Hypothesis: where one actor both generates priced anti-value and receives state financing and access, a reinforcing loop of capture can form. The map should be tested for whether charges and inflows on such nodes net out to a subsidy.
2. Many flows are still unpriced. 26 flows carry no value element. Nine are control flows; the other 17 include imports, aid and alliance contributions, all internal fiscal financing lines, unpaid household care, dispute flows and the recognition granted to religious organisations and media. Externality charges from small and medium firms also carry no anti-value element. Hypothesis: the unpriced flows coincide with the non-monetary flows the survey found most often missing from official accounts, and they are where measurement effort would change readings most.
3. The axes are unevenly represented. Of 74 elements, 30 sit on the economic axis and 14 on social stability, while the scientific axis has one element and the ecological and meaning axes two each. Hypothesis: a map built from current state practice inherits the money-centred view it seeks to correct. If the imbalance persists after quantification, it would support the claim that existing apparatuses under-sense non-economic axes.
4. Two hubs carry most of the load. The working-age adult cohort and the Commons Fund and Fiscal ministry each touch 17 flows. The working-age cohort supplies votes, taxes, data, conscription, labour, care and volunteering. Hypothesis: the reciprocal relation between state and citizen depends heavily on one cohort, which links the structural model to the demographic axis and to the participation condition of that cohort.

Testing these hypotheses requires three steps not yet taken: classifying the survey's flow forms into the value substrate, attaching magnitudes with stated confidence to each flow of the map, and comparing several countries' flows on the same schema. Until then, the map is a structural instrument for asking questions, not evidence for answers.

5 The governing loop

This section describes how the proposed design acts on what it measures. The mechanism is a single balancing feedback loop. It optimises no axis; it keeps measured quantities inside agreed limits. It has four parts: corridors that define acceptable states, an imbalance function that turns measurements into a governing signal, a democratic procedure that sets the only political parameters of that function, and a graduated response constrained in form as well as in strength.

5.1 Corridors: floor, band and ceiling

Each of the ten axes carries a corridor with three thresholds, defined per cohort and per region.

• Floor. A constitutional minimum. A cohort below it produces a high-priority imbalance. It changes only by supermajority with a veto for the affected cohort, never by the ordinary vote.
• Target band. The range in which the state keeps the axis. Voted weights steer emphasis inside it. It is set by the ordinary process.
• Ceiling. A soft cap above which marginal value is discounted, against over-optimisation. On the economic axis it takes the form of a concentration self-charge. It changes by supermajority.

Governance means keeping every axis within floor and ceiling for every cohort, and steering emphasis inside the band. The loop homes to a band, not to a peak: an axis pushed out of band in either direction raises the signal, so single-axis maximisation is excluded by construction. The sources prefer corridors to point targets because a point target behaves like a command, while a corridor fences acceptable states and leaves the interior to citizens and markets. Equity across cohorts is deliberately not an axis. It is a floor-protected constraint watched by an independent equity guardian, because a votable equity axis would let a majority trade fairness away.

5.2 The imbalance function

The loop acts on one signal. For axis a, cohort c and region r, the canonical formulation is:

imbalance(a,c,r) = w_a * distance(current_Æ(a,c,r), target_band(a,c,r)) + below_floor_penalty(a,c,r) + variance_penalty(a, across cohorts)

The first term is the steering term: the distance of the current Æ reading from the target band, scaled by the voted weight w_a. The second is a large penalty for any cohort below its floor. The third penalises variance of the axis across cohorts and is the formal expression of the equity guardian.

Only the first term is scaled by the voted weights. The floor and equity terms never are. A majority can lower the emphasis on an axis that matters mainly to a minority, but it cannot vote that minority's floor into irrelevance. In the words of the design, the system cannot read "balanced" while any cohort is below floor, regardless of headcount. A cohort of two percent below its floor generates a high-priority imbalance even if the majority gave that axis a low weight. The unit of the governing signal is measured deviation per cohort, not the number of voters.

Two qualifications apply. The signal is computed mainly on settled value and trends; provisional, unattested value is down-weighted to avoid policy whiplash. And the strength of any response is graduated by measurement confidence and by the inter-frame scope of the effect (Section 3). The function operates on cohort aggregates with stated confidence, not on personal scores. The functional forms of distance, the penalties and the cohort partition are design choices for the MOS testbed to examine; no calibrated values exist.

5.3 Setting the weights

The weight vector w(t) is the only openly political parameter of the function. Citizens set it directly or through delegation, by five rules.

• Priority budget. Each citizen distributes a fixed budget (for example 100 points) across the ten axes.
• Quadratic cost. Placing k points on one axis costs k squared. Intensity can be expressed at rising cost, so single-issue capture is expensive at the margin and spreading is cheap.
• Capped track-record weighting. Voting weight may reflect demonstrated track record but is hard-capped.
• Revocable, topic-scoped delegation. A citizen may delegate chosen axes to a party and withdraw at any time. A party's accumulated delegated weight is discounted quadratically past a threshold, and transitive delegation depth is bounded.
• Reproducible tally. Pseudonymous votes, a commitment to the aggregation function and the resulting change in w(t) are published, so any citizen can replay the count.

The output is rate-limited with hysteresis: w(t) may not move by more than a set fraction per period, and floors change only slowly and by supermajority. The cap, curvature, discount threshold and rate limit are open design parameters.

The sources ground the vote's coherence in Rawls, A Theory of Justice (1971). The structure is lexical-then-weighted: floors are satisfied first, and the weight vote optimises only within the floor-feasible region. A floor is not a very high weight; it is off the weight axis. A weighting that can be met only by pushing a cohort below floor is inadmissible, not merely outweighed. Floors are set behind a thick veil, weights behind a thin one in which the voter's own-cohort position is screened or lagged. Citizens vote the weights, not the rules of measurement. This is how the vote makes subjective value objective-for-governance without certifying that any value is correct.

5.4 Sense, diagnose, correct, re-sense

1. Sense. The Æ vector is read across all axes, cohorts and regions from attested flows and outcomes. The reading is aggregate; the state never sees inside a household or a mind. Each axis has a parallel lived-sentiment channel against which rebalancing is validated: the machine map governs hypotheses, felt sentiment governs legitimacy.
2. Diagnose. The imbalance function is evaluated, followed by a loop-dominance diagnosis. A drifting axis under a balancing dynamic wants a nudge toward band. A reinforcing loop, such as runaway concentration, wants its gain cut.
3. Correct, by graduated steps. The first rung is visibility: route the signal into the affected cohort's own line of sight, at zero coercion and zero price. The next is a price on a class of effect. Where coercion is warranted, it follows an escalation ladder: annotation, charge, restriction of access or standing, and loss of liberty last. Coercion is legitimate only for a democratically designated harm with a countable effect, even without a market price. Following Ostrom, Governing the Commons (1990), sanctions are graduated by seriousness, context, repetition and ability to pay, and their main function is assurance that compliant actors are not being exploited.
4. Re-sense. The effect is read again. Every correction is time-boxed and sunsets unless re-justified; the burden rests on continuation.

A lived-benefit gate flags any intervention justified by index improvement rather than demonstrable benefit to an identified cohort. A rising frequency of intervention is read as a mis-set frame, not as diligence.

5.5 Correct, do not direct

Smith, The Wealth of Nations (1776), limited the sovereign to defence, justice and public works, and discharged him from a fourth duty: directing private industry toward the uses most suitable to society. The design accepts this verdict. A state that reads ten axes will be tempted to steer capital toward the favoured ones, and the temptation is renounced at constitutional level.

The implementing rule is "prices in, quantities out". The loop may set the price a factor faces (an externality charge, a rent charge, a corridor boundary). It may never set the factor's quantity, employment or allocation. After the price is corrected, profit-seeking re-clears allocation. A seeing/steering firewall bars the organ that reads value from directing capital or assigning persons to tasks. Drawing on Hayek, The Constitution of Liberty (1960), every instrument must also be general (a described class, not a named party), uniform, prospective and published, predictable, and judicially reviewable against its stated effect. Because the benefits of intervention are visible and its costs dispersed, the loop carries a standing presumption against correction, and charges are set below the estimated effect, on realized magnitudes rather than forecasts.

5.6 A control-theory reading

The sources read the loop through Meadows, Thinking in Systems: A Primer (2008), which turns the stabilisers from prudence into control requirements.

Stocks and delays. An Æ axis is a stock and changes only as fast as its flows allow. Full correction on a sensed dip predicts overshoot and over-correction. Correction is therefore scaled to the stock's own rate of change, not to the raw gap, and targets allow for the drain that continues during the lag.

Oscillation. A delay in a balancing loop makes oscillation likely, and a faster response makes it worse. Hysteresis and change-rate limits are thus the required control law. The delay worth shortening is sensing latency, not response cadence. Polanyi, The Great Transformation (1944), adds that the direction of change is often beyond the governor's control while its rate is not; damage occurs when change outruns society's adjustment, and hysteresis is how the loop paces a transition such as automation.

Leverage points. Near the top of Meadows's ranking of intervention points sits the system's goal. Here the weight vector is the goal, so capture of weight-setting is capture of the state's purpose; this explains the caps, discounts, veils and reproducible tallies. Information flows rank lower but are cheap, hence visibility as the first rung. Meadows's audit also applies: a system's purpose is read from its behaviour. If the axes read green while lived value stagnates, the loop itself goes under review.

No simulation has been run. These are design hypotheses from control reasoning; Section 10 proposes how the MOS testbed would examine delay, oscillation and capture.

6 Distribution: fiscal base, commons, floor, band, ceiling and dividend

This section describes how the design finances the state, who owns what, and how the corridor of Section 5 becomes a distributive shape. It ends with a correction adopted on 2026-08-14 that changes the causal account of how the floor is secured.

6.1 Two layers of money

• The sovereign Æ ledger. A non-transferable record of what an actor has contributed and received across the axes, closer to a credit history than a coin. It cannot be bought, sold, gifted or inherited. It keeps civic power separate from wealth, so the profit order cannot buy weight in the vote.
• The $ rail. Money demoted in civic power but kept as unit of account and clearing technology. Following Smith, money's virtue is divisibility; settling an educational gain directly against an ecological loss would recreate barter between axes. Charges and the citizen dividend are therefore settled in money.

The book treatment divides the rail into an essential layer, which carries the floor, cannot be frozen at the essential minimum and may bear a small published carrying cost on idle balances (demurrage, after Keynes, The General Theory of Employment, Interest and Money, 1936), and a free layer for saving and enterprise. Both use one unit at one price, so no black-market exchange rate forms. Inflating away obligations requires a supermajority constitutional act. Because anti-value and money share a ledger, a detectable externality is debited when it occurs; latent harms are handled by escrow and later revaluation.

6.2 The fiscal base

The twentieth-century state taxed mainly at the wage relation. As automation reduces the labour share, a wage-based fisc must raise rates on a narrowing base, taxing work while machine-generated surplus passes untaxed. The sources add older objections, from Mill, Principles of Political Economy (1848), that an income tax falls heaviest on the most conscientious, and from Hayek (1960) on progressive tax as a burden a majority places on an identifiable minority. The proposed base has three primary streams and a standby:

• unearned rent to a Commons Fund: land and location, resources, spectrum, compute chokepoints, data monopolies and network positions, charged on the forward increment with present values grandfathered and improvements excluded, after Mill and George, Progress and Poverty (1879);
• anti-value charges on designated harms, kept out of operating budgets so no organ depends on the harm continuing;
• commons yields on assets held in common;
• a standby voted contribution on earned value, a main base only in a rent-poor polity, never touching anyone below the floor.

Every charge must say either "this value was never earned by its holder" or "this act imposes a measured, designated harm". None says "you worked, therefore you owe". Floor-level necessities are not taxed. Whether a given polity's rent base can carry a state is an open empirical question.

6.3 Property, rent and the commons

Ownership extends as far as creation and no further. Earned value is private, bequeathable and protected; unearned rent, captured from nature, society, position or concentration, is owed to the commons. The ledger definition is rent = Æ_captured - Æ_created; how precisely it can be estimated per actor is an open problem. The split applies to income, not activities: renting out a vehicle or a building's structure is earned service income, and only the location component of real estate is unearned. Following Smith and Mill, a payment is rent if withdrawing it would not reduce supply, and a return is earned if competitors could erode it. Where rent cannot be separated from entrepreneurial profit against a live traded price, the default is not to charge.

The modern case is compute, aggregated data and network position, whose value rests largely on collective output. Produced chips and data centres keep their ordinary return; the chokepoint bears rent. The Commons Fund taxes rent but does not operate assets, and resources stay under polycentric, user-authored rules following Ostrom (1990). Following Acemoglu and Robinson, Why Nations Fail (2012), the Fund carries four alarms: a published, bounded gap between rent captured and distributed, no financing of armed force, distribution at source, and vesting of the capture rule before the windfall.

6.4 Floor, band and ceiling

The sources name the distributive stance capability sufficientarianism with bounded, contribution-responsive inequality. Following Sen, Development as Freedom (1999), the capability floor is defined on achieved capability adjusted for cohort conversion factors, so meeting it may need unequal inputs. Following Nussbaum (Women and Human Development, 2000), the list of floored capabilities is entrenched while threshold levels are ratified democratically. The floor is lexically prior to the weighted vote, attached to the person rather than a job, not means-tested, and tapered so that work always raises net income. In the band, inequality is permitted because differences in reward carry the signal that reallocates effort and capital. The ceiling is a concentration self-charge that discounts marginal accumulation, priced rather than prohibited. Following Rawls, its purpose is to protect the fair value of political liberty, not to raise revenue; it succeeds when it collects nothing. It is indexed to the rate of concentration and aimed at the capacity to rewrite the measurement layer, the law or the compute substrate.

6.5 The citizen dividend and the Lipsits correction

The earlier doctrine said "rent funds the dividend". On 2026-08-14 it adopted a correction in response to a challenge from Igor Lipsits: if automation removes employment and income, households can buy almost nothing, so where does taxable rent come from? The objection is decisive in a closed economy where wages are the only autonomous demand, automated capital is privately owned and the state taxes only after sales. Physical output can be abundant while monetary rent is zero.

The corrected doctrine keeps four ledgers distinct: real capacity, beneficial ownership, nominal purchasing power and realized fiscal receipts. The floor is now secured upstream, through citizens' constitutional beneficial claims in the commons and critical automated infrastructure; direct access to essential provision that does not depend on tax receipts; and capacity-bounded sovereign issuance, permitted only against measured spare capacity and withdrawn as real bottlenecks bind. Issuance is not revenue and creates no real resources. Only realized commons yield pays the variable cash dividend. Rent collection still prevents purchasing power from capitalising into scarce assets, prices scarcity and withdraws excess demand, but it does not pre-finance the floor. The adopted wording is: automation can create real surplus; ownership determines the initial claims on it; money makes those claims exercisable; and fiscal collection occurs only after monetary flows have been realized.

6.6 New failure modes

Zero-rent demand collapse. Wage income disappears, sales and asset values fall, and a dividend defined as a share of rent goes to zero when most needed. The upstream floor answers this but creates a risk of issuance outrunning capacity, guarded by a public capacity ledger, a shortage trigger, automatic suspension and independent appeal.

Owners' enclave economy. Monetary rent survives only inside a circuit of owners of compute, energy, land, networks and intellectual property while the mass market collapses. A late tax on the enclave may fail once labour has lost its bargaining power. The remedy is predistribution of ownership and access beforehand, with antimonopoly rules allowed to shrink the rent base.

These are hypotheses, not findings. The MOS testbed would need to compare pure rent recycling with ownership, in-kind, issuance and hybrid scenarios; none has been run.

7 Constitution, coercion and justice as rebalancing

This section describes the legal frame that limits the governing loop: what the state may never do, what power it is granted, how it records and spends, when it may coerce, and how its courts respond to harm. The Charter is a draft (version 0.1), not ratified, and its provisions are proposed norms.

7.1 The Charter as meta-constitution

The Charter is written as a meta-constitution: one supreme text, identical for every polity that adopts value-flow governance, above each polity's own constitution and codes. Its preamble argues that as intelligence automates labour, a state built to tax labour loses its footing, and constitutes instead a state that coordinates multidimensional value flows as "an orchestrator of civilizational value, not its owner". Money remains one instrument; the steering variable is the balance of value.

The text has a preamble and 22 articles in six parts:

1. Part I, Foundations (Art. 1-3): the value-balancing mandate, value as a vector with anti-value as an independent category, and the state as one bounded frame in a federation of value frames.
2. Part II, Inviolable Rights (Art. 4-6): non-priceable, non-votable rights, a protected meaning floor, and per-cohort floors.
3. Part III, Governance Principles (Art. 7-12): imbalance as the only signal, democratic weights but fixed measurement rules, subsidiarity, anti-concentration, generosity over accumulation, and the participation condition of every actor.
4. Part IV, Value, Money and Markets (Art. 13-15): Æ as a non-transferable accounting unit, state pricing of the externality residual only, and consequence coupling.
5. Part V, Epistemic and Judicial Integrity (Art. 16-19): transparency by construction, "the state annotates, it does not censor", rebuttable interpretation, and non-retroactivity.
6. Part VI, Supremacy and Amendment (Art. 20-22): supremacy, the eternity clause, and amendment.

The eternity clause (Art. 21), modelled on Article 79(3) of the German Basic Law (1949), makes a core unamendable: the value-balancing mandate, the inviolable rights and floors, anti-concentration, and transparency and contestability. Other articles change only by deep supermajority and at a limited rate. The relation between this clause and the constituent power of a people to replace its constitution is an open problem that the external reviews (Section 8.3) asked the design to settle.

7.2 The inviolable core and the hierarchy of norms

The norms form a hierarchy in the sense of Kelsen (Pure Theory of Law, 1934): Charter, state constitution, legal codes, regulations and policy settings, individual rulings. Each layer must conform to the one above; a non-conforming norm is void, and constitutional review against the Charter is a standing function.

At the top sits the inviolable core. Life, bodily integrity, liberty, consent and exit are removed from the value machinery in three ways: no majority, no emergency and no algorithm may cross them. A right encoded as a very high weight could in principle be outbid by a large enough computed gain, so following Rawls (A Theory of Justice, 1971), the doctrine uses a two-tier structure: Tier 0 (rights and floors) is a constraint invisible to the optimizer; Tier 1 (the ten-axis weighted vector) is the space the weighting vote governs. The per-cohort floor applies Rawls's separateness of persons: no cohort's deficit can be averaged away against another's surplus. Mill (On Liberty, 1859) supplies the coercion gate: power over a citizen against her will is legitimate only to prevent harm to others. The doctrine's short form is "sight is not reach".

7.3 The grant of power and rule-bound agents

The Charter grants power for three duties only: keep the floor, keep the frame honest (justice, externality pricing, rent capture, truthful shared instruments), and keep the state itself uncapturable. Powers are derived from these duties; a power that cannot be traced to one of them was never granted. A fourth duty, directing private activity, is renounced, following Smith (The Wealth of Nations, 1776). The renunciation is structural: no organ has competence to assign persons or capital.

The offices are staffed largely by rule-bound AI agents rather than human officials. An agent holds only a published, bounded mandate. It may not enlarge its own authority, suppress an appeal, conceal the rule and evidence behind its act, or acquire a private interest in the flow it administers. An act that cannot be reconstructed from the public rule, the evidence, the model and the software version is void, whatever the intent behind it. Agents start at low autonomy and earn discretion through a track record. A related deducibility test applies to coercion: if a tribunal knowing only the public rules and the case facts could not reach the same result without the sensor's valuation, the act falls.

7.4 The registry and the ledger

The state's memory has two organs. The registry holds attested claims about standing facts (persons, property, organizations, life events). The ledger holds the value flows: the sovereign Æ ledger, booked and consumed but never transferred between holders, and the thin settlement $ rail for ordinary money. Both are append-only: a correction is a new linked entry, and even a forged entry is reduced to a public tombstone rather than deleted. The registrar verifies form and authority but adjudicates nothing.

Access is default-deny. Holders read their own accounts in full, counterparties read only shared postings, and the state reads nothing per person without a judicial access contract naming that person. State sensing runs through statistical releases on cohort aggregates with minimum-size floors, logged in a single release register. Every read of a person's record is visible to that person, and yes-or-no questions are answered by zero-knowledge proofs. The design claim is that the aggregate-only rule is enforced in the data layer, not by official restraint.

7.5 Budget and accountability

Every unit of public money is a flow on the ledger, traceable from source through organ to claimed effect. Allocation from the commons fund follows a published formula, schematically share(i) ~ imbalance(i) x weight(i) + floor_urgency(i), so the weighting vote acts as the standing budget vote. Floors override weights. The judiciary holds a fixed endowment outside the formula, and defence holds a security floor the vote may raise but not lower. Funds are released in tranches gated by attestation from the intended recipients, never by the organ itself, and as measured imbalance closes, an organ's funding shrinks, so organs end with their flows. Congruence audits ask whether spending matches voted weights, whether claimed value matches attested value, and whether spending matches the mandate. Following North (Structure and Change in Economic History, 1981), the design treats these bindings as the basis of credible commitment.

7.6 Coercion and the escalation ladder

The state may sense harm broadly, in aggregate, across frames. It may coerce only for a democratically designated harm with a countable effect. The designation half means that an anti-value becomes objective-for-governance through the polity's declaration, not an official's moral taste. The countable half means the harm must be measurable in its effects with stated confidence, even where it has no market price. An offense keyed to a vague evil is void for vagueness, and every offense must be defined and published before the act.

Harm is handled by kind. A rights violation (murder, violence, coercion of the person) is criminal by kind and never priced, because a payable atrocity would be a purchasable one. Malicious and predatory harm is charged and clawed back. Self-directed harm such as addiction is met with care. Negligence is priced as an externality.

The response escalates along a ladder: annotation, charge, restriction of access or standing, and loss of liberty. The highest permitted rung is computed, not chosen: max_rung = f(confidence, cross_frame_reach, floor_breach_severity), a published, versioned function with no per-case discretion. No hard rung applies to a person without an adjudicated finding. Statistical attribution may ground cohort restitution and organizational penalties, never hard coercion of an individual. Until the function is ratified, the design fails closed: findings are capped at annotation, except corroborated priced harm, which may reach a priced debit.

7.7 Courts and justice as value rebalancing

Every event is a change of state in the value flows, and the court is the operator that rebalances after harm. Its pipeline is: provisionally book the event, measure magnitude per axis, read the affected set from the flow record, establish intent and the threshold crossed, adjudicate with a right to contest, then order response = restore(affected_set, damage_per_axis) + deter(magnitude, intent, recurrence). Booking is immediate; the response is always adjudicated.

Restoration is primary and follows Aristotle's corrective justice (Nicomachean Ethics, Book V): take from the gain, return to the loss. Because the ledger records who received a harmful flow, the court can restore the whole affected set, not only the complainant. Following Smith (The Theory of Moral Sentiments, 1759), restitution is keyed on harm while deterrence is keyed on intent (negligent, malicious, concealed) and recurrence. Economic and medical losses support firm awards; moral damage is paid from bounded schedules. Over-coercion is booked as anti-value against the state. Incapacitation is justified only when a person's participation makes others' participation conditions fail, not by retribution. A dignity floor sits outside the penal calculus. Courts judge by the norms in force at the time of the act: evidence may update retroactively, norms may not.

8 Failure modes and defences

A design that governs by measurement must state how it can fail. The doctrine's canonical text keeps a catalogue of self-diagnosed pathologies, organized around two classical survivability audits: Hobbes's "diseases" of a commonwealth (Leviathan, 1651, ch. XXIX) and Plato's sequence of regime decay (Republic, Books VIII-IX). Two framing rules come from these. First, a constitutional sufficiency test: no axis may become a duty of the state without a matching instrument. Second, by Plato's law that each regime dies of the excess of its own ruling principle, this design's characteristic death would be measurement in excess, which motivates a standing audit of the rebalancing itself. Each of the three books adds a failure-mode chapter: on the economy, on society, and on the constitution.

8.1 Catalogue

The defences below are proposed design features; none has been tested. The last column records what the sources admit remains open.

Failure mode	Mechanism	Defence	Residual risk
Capture of the weights	Definers of categories are captured by patrons; rent relabelled as earned; anti-value definitions widened to hit competitors or minorities.	Vote sets general-rule weights, never persons or cases; Charter core and floors outside the vote; provenance traces; rotated definers; standing red team; plural media treated as a constitutional organ.	Democratic pedigree does not prove a weighting is uncaptured; defence depends on independent media and auditors.
Hidden sovereign of the accounting layer	Whoever writes attribution rules, confidence thresholds and cohort definitions controls outcomes while weights stay "democratic".	Constitutional standards guardian; independent standards body; sortition-jury ratification of schema changes; public versioning; court review.	Named by reviewers as the deepest residual; the institution is not yet specified.
Goodhart effects and metric gaming	Actors optimize the reading, not the reality; a constituency forms around the gap between proxy and outcome.	Corrections on realized magnitudes only; confidence-gated coercion; sensor is never beneficiary; versioned, adversarially audited metrics re-derived on schedule; gaming treated as redesign feedback.	Gaming adapts; low-accuracy sensing may do worse than no correction.
Metric fetishism	The state optimizes its index instead of the lived end; attention drifts to the most legible signal.	Lived-benefit gate; unsteered zone as an unscored check on the instruments.	Depends on the unsteered zone staying genuinely unread.
Confidence spoofing	Wealthy actors use money to depress measured confidence and keep the state at soft power only.	No governance signal may be moved by transferable money; governance markets staked in non-transferable standing.	Mechanism for confidence is not yet formalized.
Attestation collusion	Mutual or purchased attestation inflates claimed value.	No self-attestation; staked, juried attestation; bond sizing and delayed vesting; collusion-graph detection; priority of objective side-effects.	Requires an anti-Sybil identity layer still being written.
Cohort fragmentation	A group declares itself a micro-cohort, engineers a floor breach and extracts transfers.	Cohorts not self-declared; published vulnerability criteria and minimum size; floors owed to persons, cohorts used only for sight.	Boundary cases between protection and gaming remain judicial.
Surveillance creep	Catching long-lag harm seems to need finer data; the privacy boundary is itself contestable.	Aggregate-only sensing enforced in the data layer; default-deny access; per-subject judicial contracts; release register; read logs visible to subjects; zero-knowledge proofs.	Owned as a tension: some long-lag harms may stay invisible under privacy limits.
The performed life	Citizens and communities perform for the sensor; chilling of thought.	No personal file to perform to; counter-annotation channel; volumes of coercion and performance read as inverse legitimacy indices.	Perceived observation may chill behaviour even without real observation.
Technocratic drift	The apparatus optimizes axes citizens do not endorse; paternalism creeps from enabling to managing lives.	Congruence audit; thin endorsement answered by recalibration and persuasion, never heavier enforcement; capability not behaviour; correct, not direct; deducibility test.	Endorsement is hard to measure without the sensing it limits.
Guardian turned rentier	The commons fund becomes the next rent-seeker; long tenure privatizes office.	Fund is a claimant, never an operator of assets; distribution at source; iron-law-drift detector; tenure ceilings triggered by others; dividend coalition.	Extractive patterns reproduce under new personnel.
Purchased watcher	Judges, auditors and media are bought or removed.	Audit and contestation offices unremovable by those they scrutinize; intent-blind checks; detector separated from adjudicator; tribunals judge rule-conformity only.	Physical intimidation lies outside institutional design.
Emergency ratchet	Temporary powers become permanent.	Self-executing sunset; trigger held apart from wielder; crisis-provenance audit; no constituent reach over the Charter.	Repeated genuine emergencies may normalize use.
Interventionist ratchet	A mis-set charge is answered by a second charge, drifting toward central direction.	Release-bias; uniform, under-corrective charges on realized magnitudes; default-off sunsets; live-price tether; prices inward, never quantities.	Stability rests on five entrenched invariants; failure of any one restarts the ratchet.
Algorithmic coup	Sensing or rebalancing software is altered or substituted from inside.	Version citation and reproducible builds; unreconstructable acts void; vetting and low starting autonomy; independent implementations cross-check; hidden diversions surface as ledger flows.	Common-mode errors shared across implementations.
Zero-rent demand collapse	Wages vanish before purchasing power is replaced; demand, sales and monetary rent contract together.	Four-ledger test (real capacity, ownership claims, nominal demand, collected revenue); floor secured by beneficial ownership and essential access; dividend kept residual.	Hypothesis untested; no dividend rule is to be constitutionalized before stress runs pass.
Owners' enclave economy	Owners of compute, energy, land and IP trade in a closed circuit; the majority is pushed to a low-demand periphery.	Same stress tests; concentration monitor reading residual, reversible power rather than titles; indexed and re-based ceiling.	Structural power may evade any monitor; untested.
Legitimacy decay	A generation stops believing the rules are worth keeping.	Stability-slope meta-metric; open correction of errors; contestation treated as infrastructure.	Stated by the doctrine as its deepest known vulnerability; no mechanism survives the loss of trust.

Three patterns recur. Defences are structural rather than moral, following Acemoglu and Robinson (Why Nations Fail, 2012) on extractive institutions reproducing under new personnel, and Meadows (Thinking in Systems, 2008) on structural problems not yielding to harder policy levers. The dividend serves a constitutional purpose before a distributive one: every citizen becomes a residual claimant who loses visibly when a faction captures the commons. And the volume of coercion is an inverse index of legitimacy, after Rawls's argument that a stable just order does not rest on a coercive apparatus added from outside.

Two rows concern the transition to automated production and are the most open. The zero-rent demand collapse and the owners' enclave economy are not solved by any constitutional rule. The design converts them into validation conditions: any dividend rule must survive both scenarios in stress runs before it is fixed in law. These runs have not been performed. The society volume makes a similar admission about meaning: the state can clear ground for purpose but cannot supply it, and the doctrine calls this its most open bet.

8.2 A sabotage map, not a prosperity machine

The proposed MOS testbed (Section 10) is framed as an instrument for finding failure, not for demonstrating success. Its stated purpose is to map where value-flow governance stays stable and where it is gamed under noisy, adversarial and contested measurement, with simulated participants acting as pressure on the sensors. The framing draws on Machiavelli (Discourses on Livy, 1531): a rebalancer calibrated to the citizen who ought to exist will fail against the citizen who does exist, so the testbed should model self-interested actors who route around each rule. Each failure family in the table becomes a scenario: a captured definer, a tenure baron, a moral panic against a synthetic minority.

The charter of the testbed is deliberately modest. It aims to help avoid the worst mistakes, not to engineer the best outcome. Following Ostrom (Governing the Commons, 1990), it treats a framework as a source of questions rather than predictions, and it may not emit an optimal ruleset for central adoption. The testbed is specified but not built, and no results exist. The catalogue above is therefore a list of hypotheses about where the design is weak, not a record of tested defences.

8.3 External adversarial reviews

The doctrine has been submitted to two rounds of structured adversarial review by three general-purpose AI models from different developers, each prompted to attack the design at maximum force. These reviews are not peer review. They test internal coherence; the author triaged their findings.

The first round produced 57 findings with strong convergence. Its verdict was that the doctrine was a mature conceptual frame but not an implementation-ready specification. The convergent blockers included the feasibility of measuring realized value and rent, a tension between perspectival pluralism and a single central value vector, collusion in attestation, open-ended retrospective clawback, missing pillars (defence, emergency, citizenship), control of the accounting rules, the liability of algorithmic officeholders, and the paternalism of measuring meaning. The response narrowed the claims of measurement rather than the foundations. The state measures cross-boundary externalities, material capability floors and market signals; meaning is self-declared and never state-measured; rent is defined through observable categories; clawback is bounded by limitation periods except for concealment. The synthesis that came out of this round is the distinction applied throughout Section 7: sensing is not coercion. The state may see broadly in aggregate and act coercively only on narrow, confidence-graded findings.

The second round, run on the revised specification, raised no conceptual blockers. All three reviews moved the synthesis moves, including the sensing-coercion distinction, bounded revaluation and the non-transferable Æ ledger, into the list of elements to keep. The remaining findings concern engineering and completeness, though several were still marked as blocking for implementation: a formal executable specification of the loop, the missing pillars, governance of the measurement and confidence layer, and concrete gaming attacks such as cohort fragmentation, confidence spoofing with money, and extraction of liquid money before long-lag anti-value settles. Some tensions were recorded as permanent dials rather than solvable problems: rich sensing against privacy, stigma from published soft signals, and paternalism in defining floors. Several reviewer-proposed fixes were adopted, among them the rule that no governance signal may be moved by transferable money and the freezing of liquid funds in proportion to outstanding anti-value liability. The reviews show coherence under attack, not that the design works.

9 Intellectual lineage and critique

9.1 Position among the traditions

Axiacracy does not belong to a single school. Its substrate is the Theory of Value Management (VMT): value systems, the four value forms, anti-value as an independent category and the participation condition. On that substrate it combines elements usually held by opposed traditions. From classical and Austrian economics it keeps the market as the allocator and the profit-and-loss signal as sovereign over allocation. From the Georgist and Marxian traditions it takes the separation of earned from unearned income and the claim that position rent belongs to the commons. From the liberal constitutional tradition it takes rights, the harm principle, the rule of law and checks on power. From cybernetics and systems thinking it takes the governing loop of sense, diagnose, correct and re-sense. It rejects central planning, the labour theory of value and any claim to know a single human Good.

The project tested this position by close reading. Twenty-three works were read in full at the source, each against the doctrine, and every reading was recorded as what the doctrine takes, what it adapts and what it rejects or must answer. One rule governed the exercise: critique may add constraints, mechanisms and open problems, but it does not rewrite the founding commitments. The results feed the canonical System document and a register of positions. Table 1 condenses them.

9.2 What the theory takes and what it must answer

Thinker and work	What Axiacracy takes	What it rejects or must answer
Plato, Republic (c. 375 BC)	Justice as harmony among parts; injustice as a part usurping the whole; decline of regimes through one ruling good	Rejects the philosopher-king, the noble lie and censorship; keeps the Cave as a warning about its own dashboard
Aristotle, Politics (c. 335 BC)	Empirical survivability science; the stabilising middle; earned versus unearned acquisition; leisure as the end of the state	Rejects the soul-forming legislator and naturalised hierarchy; declines his claim to know the Good from nature
Confucius, The Analects (c. 5th century BC)	Rectification of names as definitional integrity; harmony, not uniformity; trust as the load-bearing variable	Rejects the sage-ruler and the Mandate of Heaven as source; keeps the conditionality of legitimacy
Machiavelli, The Prince (1513) and Discourses on Livy (c. 1517)	Design against the real actor; productive conflict; scheduled renewal; bounded emergency power	Rejects the collapse of "ought" into "is"; the Charter stands above the rebalancer
Hobbes, Leviathan (1651)	Why coerce at all; protection-obedience reciprocity; no law after the fact; the diseases of the commonwealth	Must answer the indivisibility of sovereignty: the apex is a rule (the Charter), not a person
Smith, The Theory of Moral Sentiments (1759)	Value formed by sympathy and judged by an impartial spectator; justice enforceable, beneficence free	Must answer that aggregated sympathy encodes rank; the state can never be a value sensor in the simple sense
Smith, The Wealth of Nations (1776)	Unearned rent as the best tax base; protection of the worker's mind; the knowledge limit on the statesman	Renounces the "fourth duty" of directing private capital; must harden every mechanism against capture
Mill, Principles of Political Economy (1848)	Distribution as a human institution; the rent/profit demarcation; the stationary state	Diverges on Say's Law and on routing income to hiring capital; positive-exceptions catalogue left open
Marx and Engels, Manifesto of the Communist Party (1848)	Diagnosis of commodification, alienation and concentration	Rejects abolition of property and markets; must answer "the state as committee of the ruling class"
Mill, On Liberty (1859)	The harm principle as the coercion gate; fallibilism; individuality; tyranny of the majority	Must keep capability floors as provision, not tutelage; the book acts as a standing tribunal
Marx, Theories of Surplus-Value (written 1863)	Productive value as a social relation; operational rent-sensor details; crisis topology	Rejects the labour theory of value; must answer whether measurement adds a second fetish
Marx, Capital, Vols I-III (1867-1894)	Fetishism critique; differential and absolute rent; displacement does not self-correct	Rejects the labour theory of value and revolution; must answer the separability of rent from surplus
Keynes, The General Theory of Employment, Interest and Money (1936)	Effective demand; the multiplier behind the dividend; euthanasia of the rentier; demurrage	Declines socialisation of investment, answered through dividend and money demotion
Polanyi, The Great Transformation (1944)	Fictitious commodities; embeddedness; the double movement as constitutional form	Declines re-embedding by planning; corrects the frame and keeps the market interior
Mises, Human Action (1949)	Ordinal value; prices in, quantities out; the commons fund as claimant, not operator	Must answer the calculation problem and the instability of a "third position"
Hayek, The Constitution of Liberty (1960)	Dispersed and tacit knowledge; spontaneous order; rule of law as the form of correction; the minimum floor	Must answer the knowledge problem and that a vote is confined to the already known
Rawls, A Theory of Justice (1971; rev. 1999)	Priority of right; separateness of persons; the veil as a debiasing device; just savings	Diverges on maximin (sufficientarian floors) and primary goods (ten-axis capability metric)
North, Institutions, Institutional Change and Economic Performance (1990)	Institutions versus organisations; measurement cost; adaptive efficiency	Must answer path dependence and lock-in of its own founding architecture
Ostrom, Governing the Commons (1990)	Eight design principles as a legitimacy audit; polycentric nesting; monitors accountable to the monitored	Must answer crowding-out of self-governance and regime-conditionality
Meadows, Thinking in Systems (2008)	Stocks, flows and feedback; leverage points; system traps; resilience	No rejection; must manage delays and self-chosen boundaries
Acemoglu and Robinson, Why Nations Fail (2012)	Inclusive versus extractive institutions; virtuous and vicious circles	Must answer "you can't engineer prosperity" and capture of the measurement layer itself

The pattern is consistent. The classical and systems sources are mostly confirmatory. The sharpest pressure comes from three directions: the Austrian critique of knowledge and calculation, the Marxian doubt that rent can be separated from productive surplus, and the institutional literature on capture.

9.3 The strongest objections and how the design answers them

The knowledge problem. Hayek, in The Constitution of Liberty (1960), argues that relevant knowledge is dispersed and partly tacit, so no central sensor can hold it, and that a majority can only vote on what is already known. The design answers in three ways. The Æ-vector is a diagnostic aim, not an operative warrant: it informs, but every coercive correction must take the form of a general, prospective, reviewable rule. The loop corrects; it does not direct. And the vote constitutes value only over effects already surfaced, while an unsteered zone, which is exempt from prior scoring, protects the discovery frontier. One part remains open: the wall around that frontier is a design commitment that a future majority could vote to shrink, and a more capable sensor makes that easier. The doctrine carries this as a known structural vulnerability (System, Part VII).

The calculation problem. Mises, in Human Action (1949), denies any unit of value and any interpersonal sum. The design accepts the premise and narrows its claim. Individual value is subjective. The citizens' vote makes value objective for governance, as a declared social act, not a measurement. Effects are measurable with stated confidence, and the Æ-vector stays a vector, never summed into one score. Coercion is legitimate only for a democratically designated harm with a countable effect, even without a market price. To keep calculation possible, every socialised magnitude is tethered to a market price the state does not set, and the state sets prices, never quantities or capital allocation. Mises' stability charge is answered only conditionally: the doctrine survives as a system, not a slide toward planning, if five invariants hold (rent-only base, uniform under-corrective charges, release bias, default-off sunsets, no capital direction). Whether they hold under majority pressure is open.

Capture. North (1990) and Acemoglu and Robinson, Why Nations Fail (2012), together state the gravest risk. Organisations invest in changing the rules, and a hostile elite would not destroy the sensor but re-parameterise it, extracting value by moving the meter. The design answers with separation of sensor from beneficiary, a non-transferable and non-inheritable civic ledger, plural and contestable measurement, logged calibration, an apex that is a rule rather than a person, and an explicit enactment precondition: the rule-set presupposes a broad coalition and cannot be installed from above. Two items stay open. The enactment gap asks whether the doctrine can create the coalition it needs. The reflexive vanguard risk asks whether the custodians of the ledger become the next extractive elite. Both are hypotheses to be tested, not settled results.

10 A simulation testbed (proposed method)

10.1 Purpose and status

The MOS testbed is a proposed method. It is specified at the level of architecture and has not been built. No simulation runs have been made, and no results exist. The project is currently paused. All numbers in this section are design estimates from the specification, not observations.

The purpose is deliberately modest. Following the challenge in Why Nations Fail (2012), the testbed is not meant to show that Axiacracy produces good outcomes. Its charter is to "avoid the worst mistakes, not engineer the best outcome." It is a sabotage map, not a predictor: it instantiates self-interested actors and records where each rule is gamed, captured or routed around, for example through capture of calibration, directed capital, transfers captured at the centre, and drift of the guardians toward oligarchy. Ostrom's distinction applies: the testbed is a framework that generates questions, not a model that issues an optimal rule-set. It must also be guarded against its own legibility, because a clean simulated dashboard could become an argument for central steering, which the doctrine rejects.

The testbed is one realisation, not the doctrine. Axiacracy fixes invariants (multidimensional value, separation of sensing from coercion, anti-value, the non-transferable civic ledger, per-cohort floors, transparency, the Charter). The testbed makes replaceable choices: AI citizens instead of humans, a simulation clock instead of real time, and specific parameter values. It also has a property reality lacks, namely access to ground truth. That property is used only to score the doctrine's estimators against known values. It is not a claim that a real state could have it.

10.2 Population and tiers

The design populates one virtual polity with 2,000 to 3,000 AI-agent citizens. Each citizen is a value transformer in the graph and an agent with identity, civic roles, needs including meaning, a contribution record and a life cycle. To keep cost bounded, cognitive fidelity is tiered:

• Tier A (design estimate about 30 to 150 agents): state sub-agents, judges, key business actors and protagonist citizens, run on frontier models.
• Tier B (about 300 to 600 agents): mid-agency citizens making real choices, run on pooled open models with short, batched prompts.
• Tier C (about 2,000 or more): the bulk of the population, run by rule-based behaviour models and sampled distributions, with no model call per tick.

Citizens move between tiers as they become relevant. A Tier C citizen facing a real decision or a dispute is temporarily promoted. For population-scale effects, one call can reason over a representative cohort and the result is distributed statistically.

10.3 Time and capacity estimate

The simulation is event-driven. Time advances in ticks, and in each tick only citizens with a pending decision act. The working assumption is about 3% active per tick. With 3,000 citizens and one tick per simulated day, that gives about 90 actors per sim-day, split roughly into 10 Tier A, 30 Tier B and 50 Tier C. Over a 365-tick simulated year this is an estimated 3,600 frontier calls and 11,000 pooled calls, against a naive 3,000 x 365, or about 1.1 million calls. These are planning figures that show the workload is development-scale rather than population-scale. They have not been checked against any running system.

One tick follows the governing loop: advance time and fire events; citizens and firms act, each action producing a value flow with an Æ-delta, a money cost and any anti-value; the ledger books value per axis, region and cohort; state sub-agents sense and diagnose imbalance against corridors; they propose corrections in the form of missions and incentives within the Charter; disputes go to review; the state observes and the loop repeats. The event journal is append-only, so any past state can be reconstructed and audited.

10.4 State sub-agents and meta-models

Each state sub-agent owns one axis for a jurisdiction, senses flows and anti-value on it, diagnoses imbalance, proposes corrections and is itself reviewed on how well its axis stays within its corridor without raising anti-value. The specification names Demography, Scientific Advancement, Cognitive Health, Infrastructure Stability, Ecological Sustainability and Space (frontier) agents, with Economic Balance and Social Stability agents beside them, roughly 8 to 15 in all.

The state is modelled as a semantic graph of about 30 meta-models in five clusters: polity and governance (constitution, branches, ministries, territory, law, judiciary, programmes, treasury); citizens and society (citizen, civic role, household, reputation, needs and wellbeing, education); economy and production (organisations, markets, contribution, infrastructure); civilisation and knowledge (missions, science, culture and meaning, ecology); and simulation runtime (agent runtime, time and events, observation, control dials). Under all of them sits the value substrate: value transformer, the multidimensional Æ-vector, value flow and anti-value. Eleven meta-models are marked for the first build.

10.5 The test bench

The System document (Section 24) sets the method for the test bench. Following North (1990), the primitive under test is the incentive structure, not the sensor: the Æ-vector matters only insofar as its pricing changes what agents are rewarded for. Following Machiavelli, the testbed must model actors as they behave, not as they ought to behave, because a rebalancer tuned to idealised citizens would be gamed by real ones. The bench therefore asks where measurement stays stable and where it is gamed under noisy, adversarial and contested conditions. It also sets two operating disciplines that a real transition would need: exercise the sensor in calm periods against its own value terrain, and batch corrections so that the corridor can settle. Any later move toward the real world is framed as bounded, reversible pilots under existing institutions and names.

10.6 Phase P0

The first phase, P0, is defined as about 100 citizens in one region with three or four sub-agents. Its single goal is to close the first balancing loop end to end: one full cycle of act, book value, sense imbalance, propose correction, apply incentive and re-sense, plus at least one corrected imbalance, one revaluation through an independent reviewer and one handled alarm for a cohort below its floor. Later phases (P1 to P3) would add law, courts, families and education, then markets, infrastructure, science and ecology, then crisis and concentration stress tests at full scale. None of these phases has started.

11 Open problems and research agenda

11.1 Open problems stated by the doctrine

Part VII of the System document, "Honest status and open problems," lists fifteen items. They fall into five groups.

• Specification. The exact imbalance mathematics, event schemas, cohort rules, hysteresis parameters and a worked tick are not yet written.
• Measurement. Whether enough axes can earn a defensible settling mechanism is an empirical question. How much of meaning, trust and epistemic health has a countable behavioural footprint, and can be charged, rather than only indicated, is unknown. The boundary between inter-frame and intra-frame matters, which gates coercion, has no natural location and needs a reviewable revision procedure.
• Economy. The capital, credit and public-debt modules are the least tested. The separability of unearned rent from productive return, the doctrine's central wager against Marx, must be tested empirically. Whether sensing latency can beat fast, fat-tailed shocks is open.
• Survivability. The non-transferable ledger may accelerate redistributive outbidding (Mises). The entrenched discovery frontier may be voted away (Hayek). The founding architecture may lock in (North). Central provision may crowd out self-governance (Ostrom). Participatory mechanisms may not scale beyond high-trust polities, which suggests federation.
• Enactment. The doctrine presupposes a broad coalition it cannot itself create, may depend on a critical juncture, faces the risk that its own custodians become an elite, and must keep the dividend from acting as a pacifier.

One former open problem, the legitimacy of soft influence, is treated as resolved through a three-part test drawn from Mill, On Liberty (1859): an influence is legitimate only if it addresses reasons, works when disclosed and does not manufacture penalties. Policing that test at scale remains empirical.

11.2 Decisions that gate the first run

A consolidated checklist (Open Decisions) sorts the testbed's open knobs into four classes. LOCKED items are decided: govern by value-flow balancing, a single region for P0, a verifiable log, liquid democracy with concentration caps, lightweight value frames in P0, and the Tier A share as a simulation dial. PARAM items have placeholder defaults to be calibrated in simulation, such as normalised corridors of floor 30, band 45 to 70 and ceiling 90 per axis; these are arbitrary starting values, not findings. DEFER items are out of P0. Eight GATE items must be ratified before P0 can be assembled, and none has been ratified:

1. the final axis set, with epistemic as a separate tenth axis;
2. which axes are instrumented in P0 (proposed: economic, cognitive, social stability, meaning, demographic);
3. which sub-agents run in P0 (proposed: four, with an equity guardian);
4. which value frames are first-class (citizen, state and two or three firms);
5. whether firms are agentic or statistical;
6. formal adoption of the two-layer money model, the sovereign Æ ledger and the thin "$ rail";
7. a light dispute and revaluation path;
8. the success criteria for closing the loop.

11.3 Coverage gaps

An early comparative analysis tested the doctrine against the dimensions that rival orders treat as central. It found the value-theoretic core more developed than the classical statecraft dimensions. Several gaps it named (property and rent, succession, membership, emergency powers, self-diagnosed failure modes) have since received design text, but that text is untested. Others remain thin: a method for judging whether an intervention will actually move the balance (the analysis points to belief-aggregation markets in the manner of futarchy), a structural viability check in the cybernetic sense, an explicit model of the citizen and preference change, and external relations and defence.

11.4 Research agenda for ÆVRI

1. Empirical classification of the eleven-country survey. The desk survey of state flows in eleven countries is raw and unclassified. Question: can the observed flows between state, citizen cohorts and businesses be coded into value and anti-value flows on the ten axes, and how much of each state's apparatus fits the structural model of Section 4?
2. Measurement validity per axis. Question: for each axis, which indicators measure the intended effect, with what confidence, at what cost and with what exposure to gaming? This sets which axes can support a charge and which can only inform.
3. The P0 simulation. Question: with about 100 citizens in one region, can the governing loop be closed at all, and where do the first sabotage paths appear? This requires ratifying the eight gates first.
4. Rent separability study. Question: in real accounts, can position rent be separated from productive return with stated confidence, and does it relocate when charged?
5. Small-scale pilots. Question: in a municipality or a cooperative, does a transparent, citizen-weighted corridor with a cohort floor change decisions without crowding out self-governance or being captured? Pilots must be bounded, reversible and run under existing law.

Each item yields evidence that could weaken the theory, which is the point. Until such evidence exists, the claims in this paper remain hypotheses.

12 Conclusion

This paper set out to answer one question: can a state be governed by sensing and rebalancing value flows instead of commanding outcomes or taxing labour? It does not answer the question empirically. It shows that the question can be made precise, and it states what would have to be true for the answer to be yes.

The descriptive part of the answer is a value-flow theory of the state. Built on the Theory of Value Management, it reads a society as a graph of value transformers, reads value as a vector across ten axes with anti-value as an independent category, and treats the state as one bounded value frame among many. The qualitative survey of eleven states shows that every state already runs on large, mostly unpriced flows in both directions, and that very different flow structures meet the same needs. The structural model of the state makes these flows explicit at the level of nodes and edges, while leaving their magnitudes open.

The design part of the answer is a single balancing loop. Citizens vote the weights of the axes; floors and equity stay outside the vote; an imbalance function turns cohort readings into a governing signal; corrections are graduated by measurement confidence and inter-frame scope; and the state sets prices on designated harms and unearned rent, never quantities or allocation. Distribution follows from the same logic: a fiscal base on rent and anti-value rather than on labour, a capability floor secured upstream through citizens' claims and essential access, a band in which inequality carries information, and a ceiling that protects political equality. A Charter places a core of rights and floors beyond any majority, any emergency and any algorithm.

The paper has been equally explicit about failure. Capture of the weights and of the accounting layer, gaming of sensors, surveillance creep, technocratic drift and the interventionist ratchet are named, each with a proposed defence and an admitted residual risk. The strongest objections from the literature, Hayek's knowledge problem, Mises' calculation problem and the institutional account of capture, are answered only conditionally. The answers depend on invariants that a future majority could weaken, and the paper records this as an open vulnerability rather than a solved problem.

Three conclusions follow for research. First, the decisive questions are empirical and can be tested in parts: whether each axis can earn a settling mechanism with a published error bar, whether unearned rent can be separated from productive return in practice, and whether a confidence-graded loop converges rather than oscillates. Second, the cheapest next step is not a full simulation but the first closed loop: phase P0 of the MOS testbed, with about one hundred simulated citizens in one region, designed to find where the rules are gamed rather than to show that they work. Third, the eleven-country survey should be classified into the value substrate and given magnitudes with stated confidence, so that the structural model can be compared with real states on one schema.

Until such evidence exists, Axiacracy should be read as what it claims to be: a theory and a design under test. Its contribution is a vocabulary in which the state, its citizens and its harms can be described as flows of value and anti-value, and a mechanism whose failure modes are stated before its benefits are claimed. ÆVRI invites researchers to test, refute or extend any part of it.

References

• Acemoglu, D. and Robinson, J. A. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. 2012.
• Aristotle. Nicomachean Ethics. 4th century BC.
• Aristotle. Politics. c. 335 BC.
• Basic Law for the Federal Republic of Germany (Grundgesetz). 1949.
• Bolshakov, V. Axiacracy: The Economics of Axiacracy; The Society of Axiacracy; The State of Axiacracy. Edition 1.0, 2026. axiacra.cy.
• Bolshakov, V. The Meta-Orchestrator State: Government After the AI Industrial Revolution. 2026.
• Bolshakov, V. Value Management Theory and Employee-Centric Competitive Advantage: A System-Level Perspective. ÆVRI, 2026.
• Confucius. The Analects. c. 5th century BC.
• George, H. Progress and Poverty. 1879.
• Goodhart, C. A. E. Problems of Monetary Management: The U.K. Experience. 1975.
• Hayek, F. A. The Use of Knowledge in Society. American Economic Review, 1945.
• Hayek, F. A. The Constitution of Liberty. 1960.
• Hobbes, T. Leviathan. 1651.
• Kelsen, H. Pure Theory of Law. 1934.
• Keynes, J. M. The General Theory of Employment, Interest and Money. 1936.
• Machiavelli, N. The Prince. 1513.
• Machiavelli, N. Discourses on Livy. Written c. 1517, published 1531.
• Marx, K. Capital, Volumes I-III. 1867-1894.
• Marx, K. Theories of Surplus-Value. Written 1862-1863.
• Marx, K. and Engels, F. Manifesto of the Communist Party. 1848.
• Meadows, D. H. Thinking in Systems: A Primer. 2008.
• Mill, J. S. Principles of Political Economy. 1848.
• Mill, J. S. On Liberty. 1859.
• Mises, L. von. Human Action: A Treatise on Economics. 1949.
• North, D. C. Structure and Change in Economic History. 1981.
• North, D. C. Institutions, Institutional Change and Economic Performance. 1990.
• Nussbaum, M. C. Women and Human Development: The Capabilities Approach. 2000.
• Ostrom, E. Governing the Commons: The Evolution of Institutions for Collective Action. 1990.
• Plato. Republic. c. 375 BC.
• Polanyi, K. The Great Transformation. 1944.
• Rawls, J. A Theory of Justice. 1971 (revised edition 1999).
• Sen, A. Development as Freedom. 1999.
• Smith, A. The Theory of Moral Sentiments. 1759.
• Smith, A. An Inquiry into the Nature and Causes of the Wealth of Nations. 1776.