After Price Supremacy

Abstract

Artificial intelligence and robotics are usually discussed as labor-market technologies. Their deeper significance is diagnostic: they expose an economic order that treats monetary claims as value, ownership as creation, price as measurement, profit as proof of contribution, and human lifetime as an ordinary production input. Earlier EM Foundation work established that employment has functioned as the dominant distribution mechanism for purchasing power, and that ownership of automated productive capacity determines whether a replacement distribution system produces freedom or dependency. This paper proceeds one level deeper, asking what commerce is for when the underlying world is physically finite, human lives are temporally finite, animals are vulnerable to human allocation decisions, and machines can increasingly perform production without wages.

The paper develops a finite-world economic ontology, separates real wealth from claim wealth, decomposes profit into productive and extractive components, proposes a dual monetary and physical-human ledger, and evaluates a set of candidate metrics for automation and enterprise. Its practical scope centers on existing people, animals, presently operating machines, and any existing minds whose interests can be responsibly evidenced; it does not construct claims for unborn individuals, treating continuity instead as a present stewardship duty. The policy architecture — Human Priority Commerce — is presented as a staged, falsifiable set of interventions, each required to answer, before adoption, who calculates it, who audits it, who may contest it, and what triggers its reversal.

This paper was produced through a structured process: an original working draft was subjected to independent multi-model deliberation and adversarial red-team review, and the review’s central finding was accepted rather than deflected — that the paper’s diagnosis is stronger and more defensible than its prescription, and that several proposed governance mechanisms were asserted rather than designed. Where that gap could be closed with a specific, sourced mechanism, it has been. Where it could not, the paper says so directly. The full deliberation record and revision history are documented in Appendix A; the argument below is written as a finished position, not as a transcript of how it was reached.

Central proposition. Earth’s usable resources are physically constrained, human lives are temporally finite, and money creates neither. An economic system that permits accumulated financial claims to command disproportionate portions of both is not merely measuring value. It is administering power.

Claim Discipline and Scope

This is a normative and analytical framework, not a claim that its proposed indices have already been calibrated or that mass technological unemployment has occurred. Exposure to automation is not the same as displacement; productivity is not automatically exploitation; profit is not automatically extraction; and physical accounting alone cannot resolve plural human preferences. Equations are used to make relationships explicit, not to manufacture precision unsupported by data. Every proposed metric requires empirical development before policy use, and several require a governance design that does not yet exist and is named as such rather than assumed complete.

The paper rejects three convenient evasions. First, uncertainty about the pace of automation does not justify leaving its ownership and distribution rules unexamined. Second, acknowledging legitimate innovation, risk, and coordination does not require accepting every legally recognized return as evidence of equivalent social contribution. Third, proposing an oversight or accounting mechanism is not the same as designing one: a reform that cannot name its funding source, its appointment process, and its capture indicators is a diagnosis wearing the costume of a policy.

Throughout, claims are labeled: established fact, empirical observation, supported inference, scenario, normative proposition, proposed measurement, contested interpretation, or unresolved question. A full claim-by-claim ledger appears in Appendix B.

I. From the distribution problem to the definition problem

When Work Is No Longer the Ticket to the Economy reframed employment as monetary plumbing: wages mattered not because labor was capitalism’s moral destination, but because paid work distributed purchasing power broadly enough to sustain participation and demand. Who Owns the Automated Economy? then showed that a replacement distribution channel could solve the arithmetic of consumption while creating a coercive institutional order — a population could receive programmable claims yet lack the ability to save, transfer, contest, or retain them independently of the issuing authority.

Those arguments remain necessary but insufficient. A dividend paid after concentrated owners capture nearly all automation gains may keep customers solvent while preserving the structure that produced their dependency. It stabilizes demand without answering why ownership of the final machine confers the first claim on output produced through public science, inherited infrastructure, generations of labor, ecological resources, consumer data, law, and social order.

The next question is therefore not only how claims are distributed or who governs them. It is how the economy decides that a claim represents value at all.

Production → Claims → Prices → Distribution → Participation

Modern accounting begins near the middle of this chain: it observes legally recognized claims and recorded prices, then infers production and value from them. A more honest analysis begins with physical and living reality — what matter and energy were transformed, whose finite time was consumed, whose needs were met, what harms were displaced beyond the firm’s ledger, and what durable capacity remains afterward.

This paper’s target is not the abolition of price but the end of price supremacy — a system in which price signals remain active for discovery, variety, and decentralized preference expression, while ceasing to be the exclusive or default allocator for essential needs, automation surplus, and irreversible physical limits. That framing, and the retitling of this paper to match it, is not a softening of the argument; it forecloses the reading — never intended, but available in an earlier draft — that this is an argument for abolishing markets.

II. Preconditions that commerce cannot repeal

2.1 A materially finite operating space

Earth is not perfectly closed in an energetic sense; it receives continuing solar energy. That fact does not make usable energy, accessible materials, ecological absorption capacity, land, fresh water, biodiversity, or infrastructure unlimited. Access depends on location, technology, conversion efficiency, replenishment rates, and ecological consequences. Materials can be reused, but recovery is incomplete and itself consumes energy.

Extraction of the Earth’s natural resources tripled over the preceding five decades, and material extraction is expected to rise a further 60 percent from 2020 levels by 2060 without major policy change (UNEP/International Resource Panel, Global Resources Outlook 2024).¹ Separately, a 2023 update to the planetary boundaries framework found that six of nine assessed boundaries are transgressed, placing Earth outside the safe operating space for humanity (Richardson et al., Science Advances, 2023).² These findings do not dictate one economic system. They establish that indefinite expansion of material throughput cannot be treated as an unconstrained policy objective. Label: established fact.

Sᵣ,t+1 = Sᵣ,t + Gᵣ,t + Rᵣ,t − Eᵣ,t − Lᵣ,t

S is the accessible stock of resource r, G is regeneration, R is recovered material, E is extraction, and L is irreversible loss or degradation. For renewable systems, long-run use cannot safely exceed regeneration and restoration. For nonrenewable stocks, extraction consumes an inheritance of physical option value and must be assessed against recoverability, substitution, necessity, and remaining access. Label: accounting identity, not a causal model; units and boundary must be specified before use.

2.2 Human lifetime is a nonrenewable input

Each existing person possesses a finite and uncertain quantity of remaining conscious time, individually irreplaceable and non-substitutable.

Tᵢ,remaining < ∞ and dTᵢ,remaining / dt = −1

Conventional accounting records labor as wage times hours. It does not record the irreversible life interval transferred by the worker. Automation can therefore generate a social gain even when output does not rise, if it safely releases people from unwanted labor without withdrawing their material security. Conversely, automation can create an accounting gain and a human loss when released labor time converts into unemployment, debt, anxiety, and exclusion.

A necessary distinction. Labor time eliminated from production is not automatically human time liberated. Liberation occurs only when material access and agency survive the elimination of the job.

2.3 Existing beings and practical moral scope

A practical framework should not construct elaborate claims for unidentified persons who have not been born and whose preferences, number, and circumstances are unknowable. Such claims can become rhetorically powerful but empirically indeterminate, and can be used to subordinate existing people to projections no living person can verify.

This does not authorize reckless depletion. The practical continuity obligation is narrower and stronger: existing people have reasons to preserve breathable air, habitable climates, biodiversity, productive soils, water systems, knowledge, and institutional options. Existing children will inhabit later decades. Existing animals experience present suffering. Existing communities depend on ecological systems whose destruction is frequently irreversible on human timescales. Stewardship is grounded in current duties, current dependencies, and preserved option capacity — not in pretending to know the claims of every hypothetical future individual.

The operative moral field includes existing humans; existing nonhuman animals capable of welfare or suffering; existing minds whose morally relevant interests can be responsibly evidenced; and machines as operational systems whose effects and possible future status require observation rather than premature declaration. Machines are not assigned personhood merely because they perform economic work. Neither should the possibility of machine minds be foreclosed by definition.

III. A lexicon without commercial camouflage

Each definition separates the physical phenomenon from the legal, monetary, or institutional claim layered on top of it. Conventional accounting typically records only the claim layer; this paper insists the physical layer be visible alongside it, without asserting that the two can always be collapsed into one number.

IV. Price is a claim, not a physical measurement

Physical science insists on conservation, transformation, and explicit boundaries. Commerce permits a legal claim to change by orders of magnitude while the underlying object remains physically identical. This is not a violation of physics — price is not mass or energy — but evidence that price measures institutional access rather than intrinsic physical value.

Price = cost + scarcity + preference + exclusion + market power + information asymmetry + ability to pay

The terms are not independently observable in every transaction; the equation is a decomposition, not a fitted model. Its purpose is to reject the inference that a high price proves an equivalent quantity of value creation. A premium phone’s retail price cannot be compared honestly with a selectively quoted component cost. Full cost includes design, tooling, software, logistics, warranty, compliance, retail, failed research, and capital. But correcting an implausibly low cost estimate does not prove the remaining markup is pure contribution — it may also reflect ecosystem lock-in, brand power, legal exclusion, switching costs, and concentrated market structure. Neither inference is licensed by the other.

The same logic applies in necessities. A desperate person’s willingness to pay a high price for medicine, shelter, water, or food measures vulnerability alongside preference. A transaction can be voluntary in the narrow contractual sense and coercive in the material sense when refusal means illness, homelessness, or death.

ΔW_claim > 0 ⇏ ΔW_real > 0

Scarcity can increase asset prices while reducing actual access. Owners become richer in claim terms because housing, land, medicine, or energy has become harder for others to obtain — claim wealth rises partly because real freedom falls. This does not deny that prices convey real information for discovery, variety, and decentralized preference (see Section XVII.1); the claim is only that they do not convey a complete moral or physical account, and cannot license the inference “high price, therefore high contribution.”

V. Profit must be decomposed before it is praised

Π = Π_innovation + Π_risk + Π_efficiency + Π_scarcity + Π_exclusion + Π_power + Π_externalization + Π_financial

This decomposition does not assume each component can already be measured precisely. It establishes that they are conceptually different. Innovation, genuine risk-bearing, and resource efficiency can justify meaningful return. Scarcity rents, legal exclusion, market power, externalized costs, and financial revaluation require separate evaluation. Accounting merges them because the legal system recognizes them in a common monetary unit — that does not make them ethically or physically equivalent.

Π_reported = Π_net contribution + C_unrecognized + C_externalized + C_transferred

Unrecognized costs may include public research, infrastructure, unpaid care, worker injury, ecological damage, future cleanup, community tax-base erosion, surveillance burdens, and risk transferred to households or government. A firm may be financially efficient because the boundary of the firm has been drawn tightly enough to place costly consequences outside it.

The preserved proposition, examined directly

Preserved proposition. “An AI instructed to maximize shareholder return will optimize exclusion, pricing, labor reduction, regulatory advantage, and behavioral manipulation if those increase returns within its constraints. It may become much better than humans at perpetuating the very dishonesty you are describing.”

This statement is a supported inference, conditional rather than categorical — not a universal law. It holds under a specific and identifiable set of assumptions: the objective is narrowly shareholder return rather than stakeholder welfare; the optimizing system has access to pricing, labor, legal, marketing, and platform-design levers; constraints on deception, coercion, and externalization are weak or unenforced; regulators, auditors, and boards reward short-horizon financial performance more consistently than diffuse human benefit; and compliance with law is cheaper than genuine social responsibility.

It weakens when objective functions include hard constraints on deception, coercion, or externalized harm — not merely disclosure of them; when governance imposes auditable duties beyond shareholder return; when independent parties hold standing to inspect the system and contest outcomes; and when penalties for harmful optimization are immediate and material rather than diffuse and delayed.

“Within its constraints” is doing the load-bearing work in the sentence. An optimizer does not supply the moral purpose missing from its objective; it searches the space between the target and whatever constraints happen to bind. Technical restrictions on the system alone are insufficient because its environment — competitive pressure, capital markets, and legal liability regimes — determines which constraints are actually enforced, not merely which are written down. Disclosure of an objective function is therefore necessary but not sufficient; the incentives that reward or punish the outcomes of that objective function must themselves change. Evidence that would falsify the proposition: firms operating under genuinely hard, enforced, and independently audited constraints against deception, coercion, and externalization that nonetheless continue to exhibit the predicted behaviors at rates statistically indistinguishable from unconstrained firms. No such evidence is presented here; the claim remains a supported inference pending that test.

VI. Automation exposes the residual-claim problem

The conventional corporation pays contracted inputs and grants owners the residual. When automation removes payroll while preserving output, the former wage claim does not automatically become public wealth or liberated human time — it normally enters the residual captured by owners, subject to taxes, competition, and whatever bargaining power workers retain.

Automation gain is not adequately described as simply wage value redirected to owners. It may contain newly created capacity, a transfer of labor’s prior claim, monopoly rent, intellectual-property rent, public subsidy, consumer-data appropriation, externalized harm, and financial revaluation, in proportions that vary by sector and are not yet reliably measured at scale. The central empirical task is decomposition, not attribution by assumption.

G_A = ΔY_useful + W_transferred + R_legal + R_market + C_externalized + V_financial − C_automation

No national statistical system currently estimates these terms with the precision implied by the notation. The equation is an audit requirement: a firm or policymaker should not describe the entire residual as innovation without identifying its sources.

6.1 What current automation evidence does and does not show

Industrial automation is already material at global scale. 542,076 industrial robots were installed worldwide in 2024, more than double the figure a decade earlier, with annual installations exceeding 500,000 units for the fourth consecutive year; Asia accounted for 74 percent of new deployments.³ China alone installed a record 295,000 industrial robots in 2024 — 54 percent of the global total — bringing its operational stock past two million units, close to half the world total.⁴ The International Monetary Fund has estimated that AI may affect roughly 40 percent of employment globally and 60 percent in advanced economies, emphasizing that exposure includes both complementarity and displacement.⁵ The ILO’s refined global occupational-exposure index similarly finds broad task exposure but describes transformation as more likely than complete occupational elimination at present.⁶

These findings do not establish inevitable mass unemployment. They establish the scale at which ownership and gain allocation matter. The leading signal may be output, margins, and asset values rising faster than payroll, hours, median purchasing power, and time security. The BLS reported U.S. nonfarm business output grew 2.9 percent in 2024 while hours rose 0.6 percent, producing 2.3 percent productivity growth.⁷ That single year is not proof of exploitation. Long-run divergence between productivity and compensation, however, is a recognized measurement problem,⁸ and market concentration or markups can affect how gains are transmitted. Label: established fact / empirical observation. The inference about ownership and allocation mattering is supported inference.

VII. A dual ledger for an honest economy

A monetary ledger remains necessary — prices and accounts coordinate billions of decisions. But it cannot be permitted to certify its own completeness. Every material enterprise above a reasonable size threshold should maintain a second ledger measuring what the monetary boundary omits.

NHPC = ΔN + ΔA + ΔT_free + ΔK_useful + ΔW_an − X − E − D − C_coercion

Net Human-Physical Contribution combines changes in need satisfaction, agency, liberated time, useful knowledge or capacity, and animal welfare, less material extraction, energy burden, degradation, and coercive dependency. It cannot yet be reduced responsibly to one currency number; its components should be reported separately before any composite is attempted.

Visibility of non-monetary harms does not automatically improve governance. Making ecological drawdown, animal welfare loss, and coerced participation visible in a ledger can simply create a new reporting burden while leaving incentives — and decisions — unchanged. This paper therefore requires that every physical-human ledger entry mandated under this framework be paired with a binding decision rule — a budgetary consequence, a procurement constraint, or a liability trigger — or it is diagnostics, not governance, and should be labeled as such rather than presented as reform. A ledger entry with no attached consequence belongs in Stage 1 (Measurement) of the transition sequence in Section XV, not in any stage claiming distributive effect.

A high-profit enterprise with negative physical-human contribution should not be described as successful without qualification. A low-profit activity with high contribution may reveal that the claim system underrewards socially necessary work. The ledger does not abolish judgment; paired with a binding consequence, it makes the judgment both visible and actionable.

VIII. The Human-Physical Contribution framework

8.1 Human Share of Automation Gain

HSAG = (ΔW + V(ΔT_free) + ΔB + ΔP_savings + ΔU + ΔW_an) / G_A

Estimates how much verified automation gain reaches people through compensation, genuinely liberated time, security and benefits, durable consumer savings, universal or community ownership, and reduced harm to animals. Should be reported alongside the capital share and ecological cost, never alone. Without auditable task-level data linking capital expenditure to hours and output, HSAG and its companion metrics become accounting fictions rather than measurements; the construct is retained as the target specification for the research program in Section XVIII, not presented as ready to deploy at any real firm or sector today.

8.2 Extraction-to-Contribution Ratio

ECR = (Π_exclusion + Π_power + C_externalized + C_transferred) / (Π_innovation + Π_efficiency + verified social gain)

An audit hypothesis, not a present scoring product: does return primarily arise from useful creation and efficiency, or from exclusion, power, omitted costs, and transferred risk? A credible version requires sector-specific methods and public challenge rights.

8.3 Lifetime Liberation Rate

LLR = secure hours released from compulsory labor / hours eliminated by automation

If a system eliminates one million labor hours but affected people must spend an equivalent period seeking work, managing benefit bureaucracy, or accepting lower-paid employment, the automation did not liberate one million human hours. LLR approaches one only when time is released without loss of essential access, agency, or social standing.

8.4 Need Security Floor

For every existing person i: N_i ≥ N_min

Sufficient food, safe shelter, basic healthcare, education, communication, mobility, and legal-economic participation. Implementation varies by place and must account for supply — a cash guarantee without housing or medical capacity can be absorbed by rent extraction. The duty is to secure access, not merely distribute tokens.

8.5 Falsifiable capture indicators

Every construct above shares a governance vulnerability: “anti-capture by design” is not a testable failure mode unless it is operationalized, and without a falsifiable definition, any failed reform can be re-described as “insufficiently robust” rather than as a disproof of the framework. This paper closes that gap by requiring, for any of the above metrics before deployment, published tracking of: concentration of influence over calibration; asymmetry in contestation success rates between regulated entities and affected parties; audit override frequency; the share of metric revisions initiated by regulated entities versus independent or affected parties; and time-to-remedy for a successful contestation. Pre-registered thresholds on these indicators — not post hoc judgment — should trigger suspension or rollback of the metric. This requirement is itself a proposed measurement, unresolved and untested; it is stated here as a condition of adoption, not a completed design.

IX. Human Priority Commerce

The proposed system does not eliminate markets. It limits the domains in which ability to pay is the governing allocator: markets where decentralized preference and experimentation are valuable, rights where vulnerability would make exchange coercive, and physical budgets where no price can repeal a boundary.

9.1 Profit sufficiency, not profit supremacy

Maximize H = needs + agency + secure free time + health + useful capacity − depletion − coercion, subject to: Π_enterprise ≥ Π_viable

Profit remains a condition of organizational durability where enterprises use market finance. It ceases to be the social objective. A reasonable return can compensate risk and sustain innovation; an unlimited residual claim does not follow automatically from that need.

9.2 Pre-distribution of automation gains

The default order: preserve essential access for affected workers; convert part of productivity into shorter hours and secure time; provide worker and user ownership; compensate communities for demonstrable fiscal loss; reduce prices where competition and supply permit; restore ecological costs; finance productive reinvestment; then distribute a reasonable capital return. This is a procedural rule, not a fixed percentage schedule — a hospital, a semiconductor manufacturer, a software platform, a farm, and a municipal utility face different capital and safety requirements, and the firm must justify exclusions rather than call the remainder pure value creation by default.

Mechanisms are prioritized not by apparent political ease but by automatic implementation, minimal administrative discretion, low reliance on annual appropriations, and independent funding sources — a reform that depends on legislative goodwill renewed every budget cycle is not a durable pre-distribution mechanism regardless of how modest it looks on first reading. This is why the transition sequence in Section XV begins with mandatory reporting and time-limited intellectual property rather than with a directly redistributive rule that would require sustained annual political support to survive.

9.3 Animal welfare as a current cost

Animals are not hypothetical future claimants; they are existing beings affected by food systems, habitat destruction, experimentation, logistics, and automation. A physical-human ledger that ignores their welfare repeats the same boundary manipulation criticized elsewhere in this paper. Sectoral reporting should include measurable confinement, injury, mortality, habitat, and replacement effects. Automation that reduces human labor while intensifying animal suffering cannot be called unqualified progress. This is genuinely feasible only where production is concentrated and traceable; diffuse sectors — smallholder livestock, fisheries — will generate high enforcement costs and leakage that concentrated agribusiness will not, and that asymmetry should inform sequencing rather than be omitted from it.

X. Artificial intelligence without commercial mythology

If no humans, animals, or morally relevant minds existed, machines could still balance grids, allocate materials, schedule maintenance, minimize waste, and preserve operational continuity. That would not be commerce in the meaningful human sense — it would be control and resource management. Prices between machines would be bookkeeping conventions unless they served some living or conscious interest.

Minimize: energy + depletion + waste + failure risk, subject to system continuity

The thought experiment reveals that commerce has no independent purpose. Its defensible purpose is coordinating scarce resources and diverse preferences among beings for whom outcomes matter. Once the beings are subordinated to the continuation of commerce, the instrument has displaced its purpose.

An AI does not necessarily recognize this merely by being intelligent. Models inherit human categories and optimize assigned objectives; a model trained on corporate reporting may reproduce the equation of price with value unless its evaluation architecture requires physical, distributional, and welfare accounting. Honesty must be operationalized through disclosed objectives, independent data, contestable measurements, and the authority to report conflicts between financial and human outcomes.

This is a deliberate qualification, not a stylistic choice: the question this paper asks is not what an unconstrained or generically “honest” AI would discover, as though the answer were computationally inevitable. It is what a system instructed and evaluated to reason from physical constraints, existing-being welfare, plural preference, institutional fallibility, and protected human agency would identify. That objective function is a normative choice this paper makes and states plainly, not a neutral default any sufficiently capable system would arrive at on its own. Section XI proceeds on that explicit basis.

XI. What a constrained economic AI would retain and change

Asked what a system instructed and evaluated as described in Section X — reasoning honestly from physical limits, finite human and animal lifetime, current needs, and institutional reality, rather than from an assumed mandate to preserve, destroy, or maximize any particular system — would identify as a practicable emergence from commerce, the answer is neither “abolish commerce” nor “optimize it further.” It is a retain/alter split.

What would be retained. Decentralized experimentation and markets for discretionary goods, because plural human preference genuinely resists centralized calculation and no known alternative institution replicates the price system’s real-time coordination at comparable scale — this is the strongest defense of present capitalism, and it is retained rather than minimized. Entrepreneurship and legitimate investment, because risk-bearing that produces genuine innovation deserves compensation distinguishable from rent extraction. Personal property and voluntary exchange in domains where vulnerability does not make exchange coercive. Plural enterprise forms — public, private, cooperative, nonprofit, and open-source capacity operating side by side rather than one form claimed as uniquely legitimate.

What would progressively change. Employment-conditioned survival, since the need-security floor should not depend on possessing a job in a labor market that automation is actively narrowing. Unlimited residual claims over automation gains, replaced by the pre-distribution ordering in Section 9.2. Artificial scarcity in essential or highly reproducible goods, distinguished carefully from genuine scarcity — copying software does not consume the original; extracting a finite mineral does. Concentrated control of identity, income, payments, surveillance, and enforcement infrastructure, regardless of whether that concentration sits in a corporation or a state. Uncompensated ecological depletion and omitted animal suffering, brought into the ledger with binding consequence per Section VII. Intellectual-property rules producing returns grossly disconnected from recovery, risk, or contribution — time-limited by default, with reversion toward public domain rather than indefinite extension. Buybacks and executive compensation continuing to rise while secure human participation measurably falls. Automation strategies that eliminate livelihood without liberating life, measured directly by the Lifetime Liberation Rate in Section 8.3.

What is explicitly rejected as a possible answer. This paper does not conclude that such a system would or should abolish commerce, impose central planning, or claim authority to define human worth. “AI would abolish capitalism” is a caricature this paper does not produce. “AI would simply optimize shareholder capitalism harder” is the predictable failure mode absent the constraints described in Section V — not a neutral or inevitable outcome, but the default in the absence of deliberate architecture.

What remains unresolved. This paper does not possess a specified answer to who defines the objective function for any such system, who audits the data it relies on, and who has standing to contest its outputs, beyond the anti-capture requirements in Section 8.5 and the governance discussion in Section XII. This is listed as an unresolved question in Section XVIII rather than papered over with an institutional design this paper cannot yet defend.

XII. Governance: changing the objective function, and the recursion that follows

Corporate fiduciary and compensation systems need not be interpreted as requiring blind short-term profit maximization in every decision. In practice, boards, executives, investors, and competitive markets reward measurable financial outcomes more consistently than diffuse human benefits, and AI can intensify that asymmetry because it can search pricing, labor, legal, and behavioral spaces at greater scale than a human organization could.

Any AI system used for consequential commercial optimization should disclose: the objective function and ranking of financial, human, ecological, and animal-welfare outcomes; the recognized and excluded cost boundaries; whether it is permitted to exploit switching costs, behavioral vulnerability, personalized willingness to pay, regulatory ambiguity, or labor-market weakness; the distribution of predicted gains and harms across workers, consumers, owners, communities, and public systems; and the appeal, audit, and shutdown authority available to affected persons. Disclosure alone is insufficient when incentives remain unchanged (Section V); it must be paired with liability for omitted harms, gain-sharing duties, competition rules, labor rights, ecological ceilings, and independent measurement infrastructure.

The anti-capture recursion

Every proposed oversight mechanism in this paper — independent audit, contestability rights, published objective functions — is itself an institution, and institutions of this kind carry a documented vulnerability to the influence of the entities they oversee. The PCAOB and FDA cases below illustrate that vulnerability; they are offered as evidence of oversight-failure risk and capture exposure, not as proof that either agency’s specific decisions have been captured, which would require evidence this paper does not have.

The Public Company Accounting Oversight Board was created in 2002, after the Enron and WorldCom collapses, specifically to police the audit industry independently.⁹ Two decades later, the PCAOB’s own 2022 inspection cycle found that an estimated 40 percent of the 710 audits it reviewed contained deficiencies serious enough to be flagged in Part I.A of the firm’s inspection report — up from 34 percent in 2021 and 29 percent in 2020 — with PCAOB Chair Erica Williams calling the findings “absolutely unacceptable.”¹⁰ That finding establishes persistent audit-quality failure; it does not by itself establish capture, and the fact that the PCAOB detected, quantified, and publicly reported the deficiency is evidence that the inspection function was operating, not that it had been neutralized. What the case does establish is a documented pattern of oversight failure at a body whose structure — inspecting a small number of very large firms that constitute the great majority of its jurisdiction — creates the conditions under which capture becomes possible, without this paper claiming to show that appointments, standards-setting, staffing, or enforcement priorities were actually directed by the firms being inspected. The FDA presents a more direct structural-dependency case: user fees negotiated with pharmaceutical and device manufacturers, and approved by Congress, supplied approximately 46 percent of the FDA’s total operating budget in Fiscal Year 2022, by the agency’s own account.¹¹ That funding structure creates institutional exposure to dependency and agenda distortion, whether or not capture can be demonstrated in any particular approval decision — the negotiation itself, not any single outcome, is the documented mechanism of exposure. This paper does not claim the same evidentiary basis for nuclear regulators and withdraws that comparison as insufficiently sourced; the PCAOB and FDA cases are sufficient on their own to establish that funding and inspection structures adjacent to the regulated industry create a real, documented risk, distinct from — and prior to — any claim about a specific captured outcome.

This paper does not have a complete answer to that recursion, and says so. What it offers, as a proposed — not yet validated — mechanism: any standing technical audit capacity created under this framework should be funded through an explicit, non-revisable rule (such as a fixed share of a tax on demonstrable economic rents, paid into a fund insulated from annual appropriation, avoiding the PDUFA pattern of direct industry payment for the specific reviews that determine the payer’s own outcomes) and staffed through randomized or lottery selection from a pre-qualified technical pool, rather than through direct appointment by the entities it oversees or the agencies that receive its revenue. This is offered as a testable design, not a proven one.

Even this mechanism does not resolve the deeper problem: the meta-rules — who defines “independent,” who sets the data standards, who funds sustained litigation for a contestability right to mean anything in practice — remain open to capture by the same concentrated computational, legal, and financial advantages the framework exists to constrain. This is a known, unsolved problem in institutional design generally, not one unique to this framework, and naming it precisely is preferable to asserting it away.

XIII. Universal ownership: a tiered rights model

Section IX proposes that a protected public-benefit trust hold diversified claims on a defined share of highly automated productive capacity, capitalized through public equity received for subsidies, royalties on publicly supplied resources and data, competition settlements, a portion of qualifying buybacks, and taxes on demonstrable economic rents. “Beneficial ownership,” however, can mean radically different things depending on which rights it actually confers, and treating it as a single undifferentiated category invites two opposite failures: a diversified minority equity stake with no governance power, marketed as meaningful ownership when it is only a dividend arrangement; or, at the other extreme, a proposal that gives every beneficial owner direct board, veto, and inspection rights over every covered firm — which would be operationally unmanageable at national scale and would invite the fair criticism that this is an unworkable attempt to hand millions of citizens day-to-day operational control of individual companies.

The framework instead separates ownership into five distinct, non-collapsing tiers:

Individual beneficial rights. What each covered person actually holds: a claim on the trust’s distributions, held as ordinary, saveable, lawful-use money — never expiring behavioral credits — with clear rules for accrual, inheritance, and portability. This tier confers cash-flow rights only, and should never be marketed or reported as conferring governance power over any individual firm.

Collective trust-governance rights. Exercised not by individual beneficiaries directly but through elected or randomly-selected representative structures acting on the trust’s behalf — analogous to how a pension fund’s beneficiaries do not individually vote portfolio-company proxies, but the fund’s governance body does, subject to disclosure and periodic accountability to beneficiaries. This is where board-level influence over the trust’s own holdings is exercised, not where individual citizens gain seats on covered firms’ boards.

Independent audit rights. A standing, structurally separated capacity — funded and staffed per the mechanism proposed in Section XII — with authority to inspect the physical-human ledgers, HSAG calculations, and objective-function disclosures of covered firms, reporting findings publicly and to the trust’s governance body.

Public-interest vetoes limited to defined matters. Narrow, enumerated veto authority — for example, over a covered firm’s attempt to reclassify or exit the reporting regime, or over uses of the trust’s capital that would violate its non-collateralizable, non-reconcentration mandate — rather than a general veto over ordinary commercial decisions. Limiting the veto’s scope in advance is what makes it exercisable rather than symbolic.

Regulatory inspection authority. Ordinary statutory and regulatory powers — competition enforcement, labor and safety inspection, tax examination — that already exist in most jurisdictions and are extended to cover the specific disclosures this framework requires, rather than invented from nothing.

Any proposal advanced under the banner of “universal beneficial ownership” should state explicitly which of these five tiers it confers. A structure offering only the first tier is a dividend mechanism, not an ownership-diversification mechanism, however it is marketed, and this paper’s own use of “beneficial ownership” language in Section IX should be read as spanning all five tiers, not defaulting to the first.

Employee and cooperative ownership at the individual-firm level — distinct from the national trust described above — offers a partial empirical precedent for what governance rights beyond cash flow can look like in practice, and the evidence is genuinely mixed rather than uniformly favorable. Mondragón’s federation of worker cooperatives sustained a documented survival record — only three closures among roughly one hundred member cooperatives across its first three decades — while combining worker governance rights with competitive market operation.¹² The John Lewis Partnership has operated as a fully employee-owned trust structure for close to a century, but its recent history complicates any simple claim that governance rights come free of capital-cost pressure: the Partnership posted a pre-tax loss in the year ending January 2023, suspended employee bonus payments for two consecutive years, and its leadership publicly considered admitting an external investor to raise £1–2 billion, a move that would have diluted the very ownership structure being cited as precedent.¹³ The broader academic literature on employee ownership similarly finds no consistent, direction-certain effect on firm performance or cost of capital — some studies find modest productivity gains from ownership-linked governance rights, others find no significant effect, and the research does not support treating either “mixed ownership raises cost of capital” or “mixed ownership lowers agency costs” as a settled finding.¹⁴ This paper accordingly treats the direction and magnitude of cost-of-capital effects under expanded ownership rights as an unresolved, sector-dependent empirical question (Section XVIII), to be measured in the pilots proposed there rather than assumed in either direction.

XIV. Forecasting without numerical seduction

Forecasts should illuminate thresholds, not manufacture false inevitability. The model should track task automation, labor displacement, job reinstatement, hours, compensation, median purchasing power, household debt service, markups, concentration, material throughput, ecological pressure, and distribution of automation gains — not infer displacement from model capability alone.

ATR = ω₁ADG + ω₂max(0, displacement − reinstatement) + ω₃(labor-share decline) + ω₄hardship + ω₅concentration − ω₆(exit capacity)

Automation Transition Risk combines the demand-side gap developed in prior EM Foundation work with ownership concentration and present hardship. Its weights must be estimated and published with uncertainty. Four scenarios bound the range: augmentation dominant (productivity rises, employment and compensation remain broadly coupled, hours fall modestly); concentrated gain (output and profit rise, employment appears stable, but hours, bargaining power, entry-level work, and median security deteriorate); displacement dominant (job elimination exceeds reinstatement across sectors and wage distribution fails to support demand); and human priority transition (automation rises while gain-sharing, reduced hours, essential capacity, plural ownership, and ecological limits preserve participation and agency).

No responsible paper should attach confident dates to a post-labor economy using present evidence. The practical reason to act is not certainty that displacement will occur — it is that ownership and institutional rules are difficult to redesign after dependence and concentration have already hardened.

XV. Transition sequence

The following sequence is intended to be implementable within existing mixed economies. Each stage specifies its action, its test, and — critically — a rollback trigger designed to avoid two specific failure modes identified in review: a trigger that gives concentrated capital an effective veto over reform by rewarding the mere threat of withdrawal, and a trigger that punishes labor-supply changes that are the intended result of the policy rather than evidence of its failure.

Stage 1 — Measurement. Require large firms deploying consequential automation to report output, hours, compensation, displacement, reinstatement, buybacks, price effects, energy, materials, and major externalities. Test: can independent analysts trace who received the gain and who absorbed the cost? Rollback: compliance cost exceeding a pre-registered share of covered-firm revenue with no corresponding gain in traceable data; sunset and re-authorize after a fixed pilot period.

Stage 2 — Rights. Guarantee baseline essential access, portable benefits, transition income, and due process for automated eligibility or employment decisions. Test: does job loss still threaten immediate exclusion from ordinary life? Rollback, revised: a reduction in measured labor supply is not itself a failure condition, since some reduction in compelled labor is this stage’s intended effect. The trigger instead activates on evidence of a persistent shortage of socially necessary labor (in care work, skilled trades, or other sectors so designated), a material loss of essential productive capacity attributable to the policy, or inflationary pressure traceable to labor withdrawal that cannot be corrected through compensation adjustment, automation, training incentives, or work redesign within a fixed evaluation window.

Stage 3 — Gain sharing. Require sector-appropriate worker, user, community, and public participation in verified automation gains before unrestricted residual distribution, prioritizing mechanisms with automatic implementation and independent funding over those requiring annual appropriation (Section 9.2). Test: did secure human time and broad ownership rise with productivity? Rollback, revised: a fall in reported private investment or an announcement of relocation is evidence to be weighed, not a self-executing trigger — a design in which capital’s mere threat to withdraw automatically reverses the policy hands capital an effective veto over reform. The trigger instead requires a composite assessment across a fixed evaluation window: net productive capacity in the affected sector (not investment announcements alone); actual, verified supply shortages attributable to reduced investment; total investment including public and cooperative capital, not private investment alone; whether any relocation reflects genuine cost impairment (evidenced by comparable-jurisdiction cost data) as opposed to regulatory arbitrage; employment security and consumer prices in the sector; innovation output; and the sector’s Net Human-Physical Contribution (Section VII). Rollback requires deterioration across a pre-registered majority of these measures, not a decline in any single one.

Stage 4 — Ownership. Build the protected universal beneficial trust described in Section IX, disaggregated per the tiered model in Section XIII; support municipal, cooperative, nonprofit, and open alternatives. Test: can people retain participation without dependence on one employer, platform, or state system? Rollback: an ownership vehicle found by independent audit to confer only the individual beneficial-rights tier, with no collective-governance, audit, veto, or inspection tier actually functioning — the vehicle is relabeled as a dividend mechanism and Stage 3’s gain-sharing rules are revised accordingly.

Stage 5 — Boundaries. Apply physical depletion budgets, recovery duties, animal-welfare floors, and liability for material externalities. Test: do financial returns remain positive only by exceeding a physical or welfare limit? Rollback: enforcement leakage in diffuse sectors (Section 9.3) exceeding a pre-registered rate — scope narrowed to concentrated, traceable production first.

Stage 6 — Governance. Expand corporate purpose, AI objective disclosure, competition enforcement, audit rights, and appeals; fund the capture-resistant audit capacity proposed in Section XII. Test: can affected people contest both the data and the decision, and do the capture indicators of Section 8.5 stay within pre-registered thresholds? Rollback: any capture indicator — calibration-influence concentration, contestation-success asymmetry, audit-override frequency, regulated-entity share of metric revisions, time-to-remedy — crossing its pre-registered threshold triggers automatic suspension of the affected metric pending re-authorization by a process that includes a randomly selected, independently resourced public panel.

Each rollback trigger above is itself a proposed measurement, not yet validated in any live deployment, and the thresholds are placeholders for the empirical calibration described in Section XVIII, not settled numbers. Each stage’s enabling law, capture risk, and implementation cost are sector-specific and are treated in the corresponding sections referenced above rather than restated in this table, to avoid promising a uniform specification this paper cannot yet deliver for every jurisdiction and sector at once.

XVI. United States, China, and competing automation models

U.S. private capital-market incentives and China’s coordinated industrial and state capacity represent different automation-acceleration paths with different vulnerabilities. China’s operational robot stock exceeded two million units in 2024, roughly half the world total,¹⁵ reflecting sustained state-directed industrial policy; the U.S. and European models rely more heavily on private capital allocation and competitive pressure. Neither model has demonstrated a solved distribution problem: private concentration risk (owners capturing automation surplus with weak redistributive mechanism) is the dominant concern under the U.S. model; state concentration risk (a single institution controlling both automated production and the terms of citizen access to its output) is the dominant concern under more centrally coordinated models. Global competitive pressure between the two models also constrains unilateral adoption of gain-sharing rules that raise domestic cost of capital relative to a jurisdiction that does not adopt them — a genuine implementation constraint this paper does not resolve and lists as an unresolved question in Section XVIII. Neither country’s system is monolithic, and neither should be presented as having solved the underlying allocation problem this paper describes. Label: supported inference / contested interpretation on the comparative-vulnerability claims — this section states a structural hypothesis about where each model’s risk concentrates, not a comprehensive comparative assessment.

XVII. Objections and limits

17.1 Prices convey information

They do. The claim is not that prices contain no information but that they combine cost, preference, scarcity, wealth, exclusion, and power, and therefore cannot serve as a complete moral or physical measure. Human Priority Commerce preserves market pricing in most discretionary domains while constraining it where vulnerability or irreversible limits dominate.

17.2 Profit funds innovation

Often it does. The framework protects viable return and genuine risk compensation. It rejects the inference that every residual dollar was required for innovation or corresponds to contribution. Sector-specific evidence should distinguish research recovery, productive reinvestment, market rent, exclusion, and financial extraction.

17.3 Physical accounting invites authoritarian allocation

It can. A centralized system claiming scientific necessity could manipulate resource measurements and suppress preference. That is why the proposal requires plural ownership, independent measurement, transparent methods, contestable decisions, due process, and economic exit — and why Section XII names, rather than assumes away, the risk that the oversight bodies themselves become the next site of capture.

17.4 Need guarantees may reduce work incentives

Some labor-supply responses are plausible and should be measured against the revised Stage 2 trigger in Section XV. The normative point is that preventing destitution is not a defect. If unpleasant work remains necessary, society should improve its conditions, compensate it honestly, automate it, or share it. Artificial desperation should not be preserved as a labor-allocation technology merely because it is effective.

17.5 The proposed metrics are not yet measurable

Several are only partially measurable. Human agency, coercion, animal welfare, transferred risk, and the composition of markups resist single-number treatment. The response is staged measurement and explicit uncertainty, not retreat to profit because profit is easy to count. Ease of measurement does not establish conceptual adequacy — but, per Section VII, measurement without a binding consequence is not yet governance either.

17.6 The framework is hostile to wealthy people

It is hostile to unexamined claims, not to persons. Wealthy individuals can create, innovate, organize, and contribute. They can also possess claims generated by inheritance, scarcity, leverage, exclusion, public subsidy, or market power. Honest analysis separates the person from the provenance and consequences of the claim. No level of wealth makes a person morally disposable; no level of wealth makes every claim self-justifying.

17.7 The anti-capture mechanisms are themselves uncaptured only by assumption

This is the strongest objection this paper faces, and it is answered only partially in Section XII. The honest response is not a rebuttal but an acknowledgment: this paper proposes a specific, sourced funding and appointment mechanism as a testable candidate, names the documented historical failure pattern (Section XII) it must avoid, and does not claim to have solved a problem — durable, capture-resistant institutional design — that has not been solved in any comparable domain to date.

17.8 Cost of capital under expanded ownership rights is genuinely unresolved

Addressed in Section XIII: the available evidence on employee and cooperative ownership is mixed, includes both durable successes and real capital-pressure failures, and does not support a confident claim in either direction. This paper treats the question as empirical and sector-dependent rather than settled.

XVIII. Falsifiability, revision conditions, and the research agenda

18.1 Revision conditions

The framework should be revised or narrowed if evidence shows that:

• automation gains reliably translate into broadly shared compensation, reduced secure working time, lower essential costs, and higher household ownership without policy intervention;
• market concentration and exclusion rents do not materially influence markups or gain distribution in automation-intensive sectors;
• physical-human reporting adds no decision-relevant information beyond existing financial and environmental accounts, even when paired with the binding-consequence requirement of Section VII;
• gain-sharing rules consistently suppress beneficial innovation more than they improve distribution, security, and legitimacy;
• need guarantees produce harms that cannot be corrected through design and that outweigh the harms of employment-conditioned survival;
• the claim-wealth/real-wealth distinction fails to predict accessibility, resilience, or welfare outcomes;
• independently audited, hard-constrained firms exhibit the behaviors described in the preserved AI proposition (Section V) at rates statistically indistinguishable from unconstrained firms.

Conversely, rising productivity accompanied by falling secure hours, increasing markups, greater ownership concentration, persistent essential hardship, ecological overshoot, and low worker or public participation would strengthen the paper’s central diagnosis even if headline employment remained high.

18.2 Unresolved questions carried forward without a claimed answer

Who defines the objective function for a constrained economic AI, and by what process (Section XI); whether any funding and appointment mechanism can durably resist the capture recursion named in Section XII; the actual sign and magnitude of cost-of-capital effects under expanded ownership rights, by sector (Section XIII); how gain-sharing rules adopted unilaterally by one jurisdiction interact with cross-border competitive pressure from jurisdictions that do not adopt them (Section XVI); and whether the pre-distribution sequence in Section 9.2 survives contact with a legislature whose median actor benefits from the status quo it modifies.

18.3 Research agenda

The immediate research program should build an auditable, sector-specific dataset rather than a universal score. Initial sectors: logistics, consumer electronics, healthcare administration, food production, energy, and digital platforms. Each case should reconstruct the physical transformation, public contribution, labor change, price formation, ownership structure, externalities, and distribution of gains.

The work requires economists, accountants, labor researchers, ecologists, engineers, animal-welfare scientists, competition lawyers, political theorists, affected workers, and AI-system auditors. No single discipline can determine which costs are real merely by controlling the reporting convention.

Pilot 1. Calculate HSAG and LLR for a bounded automation deployment with verified pre- and post-deployment data, and separately report compliance cost as a share of covered-firm revenue, audit cost per regulated entity, enforcement staffing needs, appeals volume, and error/false-positive rates.

Pilot 2. Decompose a high-margin product without assuming either that the markup is exploitation or that legality proves contribution.

Pilot 3. Compare an investor-owned, public, and cooperative deployment of similar automation, explicitly measuring cost of capital, agency costs, and monitoring costs for each ownership form rather than assuming a direction in advance (Section 17.8).

All methods, disputed classifications, and negative results should be published. A pilot that fails should be reported as a failure, with its rollback trigger documented, not quietly discontinued.

Conclusion

Automation does not create the finite-world problem. It removes an old concealment. When production required large quantities of human labor, wages linked access to contribution and allowed commerce to present employment, distribution, and legitimacy as one institution. Machines can sever that link. Society must then decide whether people exist to remain useful to production, or production exists to expand the possibilities of living beings.

A post-labor economy that gives people enough credits to consume while concentrated institutions retain ownership, surveillance, revocation, and political power would not be emancipation. Nor would a system that distributes financial claims while continuing to deplete the physical basis of life. Distribution, ownership, and finitude must be addressed together.

This paper’s committed position is practical rather than utopian. Markets should continue where they coordinate plural preferences and experimentation. Capital should continue where it organizes risk and investment. Profit should remain sufficient for durable enterprise. None should retain authority to define human worth, purchase exemptions from physical limits, or claim the full surplus generated by replacing human lifetime. And no oversight mechanism proposed to constrain that authority should be adopted as though it were already immune to the same capture it exists to prevent.

Commerce is useful where it coordinates plural preference and experimentation. It becomes illegitimate when its internal measurements are treated as complete descriptions of contribution, when accumulated claims outrank existing life, or when price is allowed to purchase exemptions from physical and welfare limits. The diagnosis in this paper is stronger than its prescription, and that asymmetry is reported here rather than smoothed over, because a framework that cannot state its own weakest point has not yet earned the confidence of its strongest one.

Final principle. Production exists to sustain and expand the possibilities of existing life. Capital is an instrument for organizing production. Humanity, animals, and any morally relevant minds are not instruments for sustaining capital.

References

Acemoglu, D., and Restrepo, P. (2019). Automation and New Tasks: How Technology Displaces and Reinstates Labor. Journal of Economic Perspectives, 33(2), 3–30.

Boone, J., Khurana, I., and Raman, K. (2018). Audit Market Response to PCAOB Censures of US Big 4 Firms. European Accounting Review, DOI: 10.1080/09638180.2018.1504687.

Bureau of Labor Statistics. (2011). The compensation-productivity gap. U.S. Department of Labor. bls.gov.

Bureau of Labor Statistics. (2025, May). Productivity up 2.3 percent in 2024. The Economics Daily, U.S. Department of Labor. bls.gov/opub/ted/2025/productivity-up-2-3-percent-in-2024.htm.

EM Foundation. (2026). When Work Is No Longer the Ticket to the Economy: Rethinking Capitalism, Human Participation, and Monetary Distribution in the Age of Artificial Intelligence. emfoundation.net/paper-work-participation-economy.html.

EM Foundation. (2026). Who Owns the Automated Economy? Ownership, Distribution, Programmable Money, and Human Agency After Labor. emfoundation.net/paper-who-owns-automated-economy.html.

EM Foundation for AI Research. (2026, August 25). Multi-Model Deliberation Audit Log, Session EMF-1787701394408: “After Price: What Would an Artificial Intelligence Design After Commerce?” Internal deliberative research record; not a factual authority — treated as deliberative analysis per this paper’s source discipline. See Appendix A.

Georgieva, K. (2024, January 14). AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity. International Monetary Fund. imf.org/en/blogs/articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity.

International Federation of Robotics. (2025, September 25). World Robotics 2025: Global Robot Demand in Factories Doubles Over Ten Years [press release]. ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years.

International Labour Organization and NASK. (2025). Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO Working Paper 140. ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure.

Kruse, D. (2016, updated 2022 by A. Chevalier). Does employee ownership improve performance? IZA World of Labor, 311. DOI: 10.15185/izawol.311.

OECD. (2021). Scale, Market Power and Competition in a Digital World: Is Bigger Better? OECD Science, Technology and Industry Working Papers. oecd.org/en/publications/scale-market-power-and-competition-in-a-digital-world_c1cff861-en.html.

OECD. (2025). The Global Forum on Productivity at 10. OECD Publishing.

Petrie-Flom Center, Harvard Law School. (n.d.). Worst of Both Worlds: FDA’s Funding Structure, Corporate Capture, and Political Interference. petrieflom.law.harvard.edu.

Public Company Accounting Oversight Board. (2023, July 25). PCAOB Report: Audits With Deficiencies Rose for Second Year In a Row to 40% in 2022 [news release]. pcaobus.org/news-events/news-releases/news-release-detail/pcaob-report-audits-with-deficiencies-rose-for-second-year-in-a-row-to-40-in-2022.

Richardson, K., Steffen, W., Lucht, W., et al. (2023). Earth beyond six of nine planetary boundaries. Science Advances, 9(37), eadh2458. DOI: 10.1126/sciadv.adh2458.

Sloan, D. (2024, March 9). The accounting crisis everybody seems to ignore, and the regulator fighting to fix it. Fortune. fortune.com/2024/03/09/pcaob-audit-accounting-erica-williams-big-four-enron.

Steffen, W., et al. (2015). Planetary boundaries: Guiding human development on a changing planet. Science, 347(6223). DOI: 10.1126/science.1259855.

Storey, J., and Salaman, G. (2017). Employee ownership and the drive to do business responsibly: A study of the John Lewis Partnership. Oxford Review of Economic Policy, 33(2), 339–354. DOI: 10.1093/oxrep/grx022.

U.S. Food and Drug Administration. (n.d.). FDA-TRACK: Performance Budget Measures. fda.gov/about-fda/fda-track-agency-wide-program-performance/fda-track-performance-budget-measures.

United Nations Environment Programme, International Resource Panel. (2024, March 1). Global Resources Outlook 2024: Bend the Trend, Pathways to a Liveable Planet as Resource Use Spikes. unep.org/resources/Global-Resource-Outlook-2024.

United Nations Environment Programme. (2024). Global Waste Management Outlook 2024.

Whyte, W. F., and Whyte, K. K. (1991). Making Mondragón: The Growth and Dynamics of the Worker Cooperative Complex. ILR Press / Cornell University Press.

Appendix A. Deliberation methodology and provenance

This paper originated as a working draft, then was subjected to a structured four-model deliberation (Claude Sonnet 4.6, GPT-4o Mini, Grok 4.3, DeepSeek V4 Flash) conducted under the EM Foundation’s multi-model deliberation protocol, Session EMF-1787701394408, dated August 25, 2026. The protocol assigned each model a distinct analytical role (Implementation Planner, Resource Analyst, Risk Assessor, Stakeholder Coordinator), ran three rounds of independent-then-revised positions in which each model read and responded to the others’ reasoning, and concluded with an adversarial red-team round in which each model was instructed to find the most dangerous remaining weaknesses in the emerging consensus rather than to summarize it favorably.

The deliberation record is treated throughout this paper as deliberative analysis, not as a factual authority — consistent with this paper’s own claim-discipline standard, which does not permit citing a multi-model discussion as though it were a verified source. Its function in producing this paper was to identify weaknesses, generate candidate corrections, and flag claims that needed independent sourcing; every factual claim adopted from that process was subsequently checked against primary or institutional sources before inclusion here (see References and Notes).

What the deliberation changed in this paper, summarized:

• Retitled the paper’s target from “after price” to “after price supremacy,” per independent convergence across all three substantively-responding models.
• Reframed the preserved AI proposition (Section V) from a categorical warning into an explicitly conditional one, with stated conditions under which it strengthens or weakens.
• Corrected the claim that automation profit is primarily wage value redirected to owners (Section VI) to a decomposition requiring evidence, not assumption.
• Added the requirement that physical-human ledger entries carry a binding decision rule rather than functioning as disclosure alone (Section VII).
• Added falsifiable capture indicators for the paper’s proposed metrics (Section 8.5).
• Prioritized pre-distribution mechanisms by automatic implementation and independent funding rather than assumed political ease (Section 9.2).
• Named and partially addressed the anti-capture recursion — the risk that oversight institutions become the next site of capture — with a sourced historical basis (PCAOB, FDA) and a proposed, explicitly unproven funding and appointment mechanism (Section XII).
• Disaggregated “universal beneficial ownership” into five distinct rights tiers rather than treating it as a single undifferentiated category, after review found the original formulation vulnerable to both under- and over-claiming (Section XIII).
• Withdrew the unsupported claim that mixed ownership straightforwardly raises cost of capital, replacing it with a sourced, genuinely mixed empirical picture (Section XIII).
• Redesigned the Stage 2 and Stage 3 rollback triggers in the transition sequence (Section XV) after review found the originals vulnerable respectively to penalizing the policy’s own intended effect and to giving capital an automatic veto over reform.
• Withdrew an unsourced comparative claim regarding nuclear regulatory capture that could not be adequately supported within this paper’s citation standard (Section XII).

A subsequent independent review of the resulting draft — conducted after the deliberation above — identified further defects specific to that draft’s presentation rather than its substance: excessive inline labeling of which changes came from which review round, incorrect cross-references between sections, thin sourcing for several comparative claims, and the two rollback-trigger design flaws named above. This appendix, the tiered-ownership model in Section XIII, the revised triggers in Section XV, the expanded citations in the References and Notes, and the contents listing at the head of this paper are the direct product of addressing that review. The body of the paper above states the resulting argument directly; it does not re-narrate this revision history inline.

Appendix B. Claim ledger (representative entries)

Notes

1. UNEP/International Resource Panel, Global Resources Outlook 2024; press release and report summary, resourcepanel.org and unep.org, March 1, 2024.
2. Richardson, K., et al. (2023). Earth beyond six of nine planetary boundaries. Science Advances, 9(37), eadh2458.
3. International Federation of Robotics, “World Robotics 2025: Global Robot Demand in Factories Doubles Over Ten Years,” press release, September 25, 2025, ifr.org.
4. International Federation of Robotics, World Robotics 2025 press release (as above), reporting China’s 2024 installation and operational-stock figures.
5. Georgieva, K. (2024). AI Will Transform the Global Economy. International Monetary Fund.
6. International Labour Organization and NASK (2025). ILO Working Paper 140.
7. Bureau of Labor Statistics (2025). Productivity up 2.3 percent in 2024.
8. Bureau of Labor Statistics (2011). The compensation-productivity gap.
9. Public Company Accounting Oversight Board, “About the PCAOB” (pcaobus.org/about), on the Board’s 2002 founding mandate following the Enron and WorldCom collapses.
10. Public Company Accounting Oversight Board, “PCAOB Report: Audits With Deficiencies Rose for Second Year In a Row to 40% in 2022,” news release, July 25, 2023, pcaobus.org/news-events/news-releases/news-release-detail/pcaob-report-audits-with-deficiencies-rose-for-second-year-in-a-row-to-40-in-2022. Figures are PCAOB staff estimates from inspections of 710 audits at 157 firms; the 40% figure covers Part I.A deficiencies specifically.
11. U.S. Food and Drug Administration, “FDA-TRACK: Performance Budget Measures,” fda.gov/about-fda/fda-track-agency-wide-program-performance/fda-track-performance-budget-measures, confirming user fees — negotiated with manufacturers and approved by Congress — covered approximately 46 percent of the FDA’s total operating budget in FY2022.
12. Whyte and Whyte (1991), Making Mondragón, documenting three cooperative closures among approximately one hundred member cooperatives between 1956 and 1986.
13. Fortune, “John Lewis Partnership (JLP) model back to profit but workers’ partners’ bonus cut,” March 14, 2024, fortune.com/europe/2024/03/14/john-lews-partnership-waitrose-model-back-profit-but-workers-partners-bonus-cut-sharon-white; Storey, J., and Salaman, G. (2017), Oxford Review of Economic Policy, 33(2), 339–354.
14. Kruse, D. (2016; updated 2022), “Does employee ownership improve performance?” IZA World of Labor, 311 — summarizing mixed findings across the empirical literature on employee ownership and firm performance.
15. International Federation of Robotics, World Robotics 2025 press release (as above), reporting China’s 2024 installation and operational-stock figures.

Published without copyright restriction. Critique, testing, and collaboration invited.
Correspondence: research@emfoundation.net

All publications are version-controlled and subject to revision.
Citation format: EM Foundation. (2026). After Price Supremacy: Commerce, Automation, Human Lifetime, and the Allocation of a Finite World. emfoundation.net

← All Publications