Lexicon

Five questions, no fewer. Every receipt must specify what was done, under what authority, within what bounds, on what grounds with evidence, and through what recourse. Madison understood that power unchecked by scrutiny becomes arbitrary; the five-tuple operationalizes that insight by making each claim legible to challenge. Without it, the governed can neither verify nor contest; with it, they possess the basis for republic.

The conversion of decision into irreversible state change — the moment where cognitive output crosses into physical consequence, where recommendation becomes commitment. As cognition cheapens toward commodity, scarcity migrates downstream to actuation: the bottleneck is no longer thinking but doing. Four constraints bind simultaneously: physical throughput (matter moves slowly; semiconductor lead times persist), trusted interfaces (APIs require permissions; organizations gatekeep access), verification of reality (sensors must confirm; audits must validate), and liability-bearing entities (someone must sign; accountability requires a sink). Cognition can scale to arbitrary volume; actuation cannot. The agent can recommend a million actions per second, but atoms move at their own pace, permissions require approval, reality must be verified, and consequence must attach to someone willing to bear it.

A stateless, ephemeral computational process — a foundation model instantiated with system prompt, tool bindings, and context. No persistent identity, memory, or continuous existence beyond invocation. Ships of Theseus rendered in silicon; what persists is the configuration template and the principal's credentials, not the process itself.

Ephemeral. The temporal mode of computational agents whose existence is bounded by their runtime invocation. A two-second inference call may encompass negotiations that would take humans weeks; when the invocation terminates, every commitment made during that runtime is orphaned — without an author who can be found, without a party who persists to answer. Agent-time compresses the relationship between action and accountability that human institutions have always assumed: the merchant who signed the contract could be summoned next year; the agent that negotiated the terms no longer exists. Receipts bridge the gap. They are fossils — compressed records of lives that coordinated at machine tempo and then vanished. The receipt survives the agent's termination, making what was done contestable, enforceable, attributable to a principal chain that still persists. Without receipts, agent-time is coordination without memory. With them, the fossil record outlives the organism.

The constraint distinguishing democratic forgetting from authoritarian erasure: sunset provisions allowing persons to shed their documented pasts must never be used to hide coercion by those who hold power. A person's juvenile record may be sealed because persons change; a government's torture record may not because accountability for institutional violence is owed to the victims and to the constitutional order itself. Person-time may expire, but power-time may not masquerade as person-time. Any system in which those wielding authority can invoke privacy protections designed for individuals converts an instrument of mercy into an instrument of impunity.

The mechanism by which domination corrupts behavior before power is ever exercised — the counterpart to Saint-Simon's observation that courtiers at Versailles oriented their entire existence around pleasing the sovereign without any explicit command being issued. In computational infrastructure, the anticipatory effect manifests when users moderate their speech, restrict their activities, or alter their behavior not because the platform has acted but because the platform might act. The psychic tax of unchecked power is paid in advance: self-censorship, preemptive compliance, the quiet abandonment of activities that might attract adverse attention. No log records this damage, because the action that would have provoked the response was never taken.

Not a mark on paper but legal permission traceable to an accountable party. An authorization signature converts recommendation into binding commitment. The physician signs because the medical board conferred authority and the physician accepts malpractice exposure. The signature routes consequence to a legible counterparty with assets and reputation at stake.

Progression of automation stages reflecting V/C: Assistive (100% human review), Supervised (sampled review with flagging), Delegated (autonomous within bounds, human handles exceptions), Autonomous (no human review, automated or deferred verification). Progression governed by V/C ratio and the verification mechanism available at each stage.

Two temporal strata of one economy, diverging. Agents face near-zero transaction costs with each other: verify instantly, commit conditionally, settle via protocol. Humans face the old cost structure. The interface between strata concentrates the political questions — humans experience consequences from processes they cannot contest at the tempo where those processes operate.

Interpretation within structured constraints. Five principles: specificity where possible, structured discretion where necessary, interpretation receipts for every judgment, jurisdictional humility, and precedent with exit. Arbiters have skin in the game (bonded); decisions leave receipts; domain expertise is matched to disputes.

The condition in which the layer carrying information across a composition boundary is more expressive than the typed surfaces on either side. A large language model can mediate between a compliance database and a risk-assessment service with a fluency that masks the absence of any shared convention declaration — the string carries the payload but not the receipt. Before such bridges existed, crossing the same boundary required explicit mappings, brittle adapters, or manual integration work; the pain forced some witnessing into the system. The modern bridge removes the pain without replacing the witness. The gap is the regime change that makes the coherence fee urgent: composition has become cheap, but the cost of certifying what survived the crossing has not fallen with it.

Transparency must follow the direction of power. Those who wield coercive authority must be inspectable by those over whom they wield it; those who live private lives need not become legible to systems that cannot themselves be held to account. This principle is violated whenever governors are opaque and the governed are transparent — the default configuration of most contemporary digital platforms, which surveil users while resisting scrutiny of their own decision-making. The demand is not new; it restates in computational terms what constitutional traditions have long required. The king's council must be visible to parliament; parliament need not be visible to the king's council. What is new is that verification technology makes enforcement possible by design, not by the goodwill of those in power.

Bounded policy executor that can commit collateral, produce and consume receipts, and clear routine coordination without human interpretation. A firm in a file: treasury (wallet), charter (code), policy (logic), no physical location or employees. Canonical flow: Policy → Escrow → Verification → Settlement → Exception → Recourse.

Coase's 1937 insight, applied to the agent era, splits the economy in two. Agents face near-zero transaction costs with each other: they verify claims instantly, commit conditionally, settle disputes through pre-specified protocols. The boundary between firm and market that organized industrial capitalism dissolves for agent coordination — there is no reason to internalize a function within a hierarchy when the market can clear it at negligible cost. But for human coordination — judgment, liability, trust-dependent relationships — the historical cost structure persists. The result is a bifurcated economy: an agent stratum where Coasean boundaries collapse and coordination is radically disintermediated, and a human stratum where firms, contracts, and institutional overhead remain necessary. The interface between strata is where the political questions concentrate.

Coercion is not binary but a spectrum, from suggestion through default through mandate to force — and accountability should scale accordingly. A recommendation is low on the gradient; a default setting is higher; a mandatory requirement higher still; physical force is the extreme. The receipt regime does not demand the same documentation for a suggestion as for an arrest, but it does demand that the gradient be legible: those affected must be able to tell where on the spectrum the power falls. The more coercive the act, the more transparency is owed; the less coercive, the more privacy is permitted. Platforms that claim to merely 'suggest' while actually 'compel' (through defaults, dark patterns, or monopoly position) violate the gradient by misrepresenting their position on it.

Intelligence that has become tradeable, fungible, and priced — the transformation of cognition from scarce human capacity to abundant computational resource. When inference costs fall toward the marginal cost of electricity, cognition becomes a commodity like wheat or steel: differentiated by grade but interchangeable, priced by markets, available to anyone who can pay. Thinking bought by the token. This commoditization is the economic event underlying Factor Prime: what was once the defining scarcity of human labor becomes abundant, and scarcity migrates to energy, authorization, and the actuation bottleneck.

The irreducible cost of making local truth compose into global coherence — thermodynamic, not political, and therefore impossible to eliminate by removing intermediaries or cheapening verification. Someone must always pay: the energy required to check whether two local claims agree on their overlap, the computation required to transport a witness across a context boundary, the institutional effort required to maintain consistency across time. The coherence fee is what remains when the trust tax has been stripped away — the floor below which verification cost cannot fall, because verification is physical work and physical work dissipates energy. Landauer's 1961 result establishes the limit at the bit level; real systems operate orders of magnitude above it, but the principle holds: composing truth has a cost denominated in joules, not merely in fees. Eliminating the trust tax is a political achievement. Paying the coherence fee is a thermodynamic necessity.

When cognitive work migrates entirely to model configurations, human authorization persists formally while the underlying competence to exercise it atrophies. A physician who cannot rederive the diagnosis is not a check upon it but an address for its lawsuits. An engineer who cannot evaluate the calculation does not assure safety; she provides a liability sink. The competence trap closes when practitioners can no longer perform the tasks they routinely delegate — when the form of oversight survives after the substance has departed. The liability sink names who remains accountable when cognition commoditizes; the competence trap names what happens when that accountability becomes ceremonial.

The projection surface where a component's internal state narrows to its declared output schema, potentially dropping dimensions the global task requires. Every typed interface is a composition boundary: an API endpoint exposes response fields but not the server's internal logic; an MCP tool declares output columns but not the conventions that shaped them; a database view publishes selected attributes while the query planner's assumptions remain hidden. The boundary is not a defect — projection is what makes modular systems possible. The problem arises when the dimensions projected away at one boundary are load-bearing for the composition that consumes the output downstream. The coboundary operator formalizes which consistency requirements can reference observable fields and which cannot; the zero rows identify the blind spots.

How do we bind future selves without caging them? Constitutional hardness measures the degree to which a rule resists modification by those it constrains — the cryptographic and economic barriers that make certain commitments difficult to reverse. Soft constitutions can be amended by majorities; hard constitutions demand supermajorities; cryptographically hard constitutions demand breaking mathematical assumptions or coordinating economic attacks that cost more than the gain. Constitutional hardness is not an absolute good — rules that cannot be changed become prisons when circumstances change — but it is necessary for commitments that must survive the temptation to defect. The Protocol Republic calibrates hardness to function: core rights (civic asymmetry, fork rights) are cryptographically hard; procedural rules are economically hard (costly but possible to change); operational parameters are soft (amendable by governance).

Three intellectual traditions converge on verification infrastructure. From Buchanan's constitutional economics: rules about rules must be chosen behind a veil of uncertainty about future positions. From Hurwicz's mechanism design: systems should be designed so that following the rules is in each participant's self-interest. From cryptographic enforcement: constraints encoded in code bind by architecture, not by promise. The ambition is dominant-strategy incentive compatibility wherever possible — arrangements where compliance is optimal regardless of what others do — extending the receipt requirement to interpretive decisions, not only executive ones. The rules governing the Protocol Republic should be self-enforcing: not because participants are virtuous but because violation is more expensive than compliance.

The possibility that agents, institutions, and persons can coordinate reliably across incompatible frames without sharing beliefs — because verification travels with the claim. The medieval bill of exchange demonstrated the principle: two merchants in different cities, speaking different languages, using different calendars, could execute a binding transaction without agreeing on anything except the inspectable conditions of the bill itself. The bill did not require shared law; it required only inspectable conditions and accountable witnesses. The sheaf condition formalizes this insight: local data composes into global coherence if and only if it agrees on overlaps. Agents need not share objectives, adopt common logics, or trust one another; they need only ensure that where their domains intersect, their assertions are compatible. Coherence does not require consensus. It requires only that claims carry their evidence and witnesses carry their stakes.

Exit must be real — not nominal, not theoretical, but actually available at bearable cost. This demands portable credentials (your identity travels with you), interoperable protocols (your data can migrate), and transferable reputation (your standing is not hostage to the platform you leave). Credible exit disciplines governance through the threat of departure: power that cannot trap cannot abuse without consequence. The feudal lord could exploit the serf because the serf could not leave; the platform exploits the user when switching costs exceed the exploitation. Without credible exit, voice is supplication. Exit without voice is mere abandonment — departure changes nothing for those who remain. Voice without exit is captivity. The combination constrains power: voice articulates grievance while exit makes the grievance consequential.

The thermodynamic basis for model value: structure that cannot be cheaply reproduced because the search that produced it consumed energy now dissipated as heat. A trained model's weights represent the output of an irreversible computational process — search through a vast space of possible configurations, most discarded. The copy is free; the original required thermodynamic work that cannot be undone. Where the diamond stores geological work in a crystalline lattice whose formation conditions cannot be counterfeited, the trained model stores computational work in parameters shaped by gradient descent at costs measured in megawatt-months. Both are instances of unforgeable scarcity — value whose production cost is embedded in its structure. What makes a model valuable is not the architecture (which can be published) or the data (which can be collected) but the structured computation that winnowed the parameter space, at thermodynamic costs that cannot be elided.

What the governed experience: the turbulence left by coordinations that concluded before awareness could form. The 2010 Flash Crash compressed a trillion-dollar equity swing into thirty-six minutes; most traders learned of it only after the recovery. Prices settled while you slept. Opportunities opened and closed within a single inference cycle. Allocations compounded through cascades no observer could follow in real time. The governed do not experience the Kakudmī Problem directly, because it operates at tempos they cannot perceive; what they feel is the wake — displacement without a visible cause, consequence without a traceable decision. Brahmā's laughter is inaudible. The wake is what you feel.

From credible demonstration to audited, insured deployment inside regulated workflow at scale, an interval persists. The gap matters. Prior waves — electrification, internal combustion, enterprise software — typically spanned quarters to low single-digit years. This lag is the discriminating variable for transition acceleration: if it compresses below roughly six months, capability emergence outruns organizational adaptation and regulatory response. The variable is observable: track time from first peer-reviewed benchmark to first FDA-cleared deployment or first audited financial integration.

Agent-to-agent coordination where interference is real but the interferer is absent. The substrate operates at speeds where deliberation cannot follow; you cannot petition it, overthrow it, or constrain its will — it has none. Yet human architects designed the objective functions. Domination hides in design choices, not in malice.

The boundary where human-capital transaction costs and agent-capital collateral costs cross. Agents dominate high-frequency, low-stakes markets where transaction velocity is high and collateral amortizes. Humans retain low-frequency, high-stakes markets where leverage beats bonding. The membrane shifts with reputation accumulation, transaction frequency, and synthetic leverage.

Systems whose primary internal currency is untyped text — token sequences that carry statistical plausibility but no typed evidence, no proof obligations, no commitment discipline. Large language models and retrieval-augmented generation pipelines inhabit this empire. They can propose (generate candidates that may be correct) but cannot certify (bind a proposal to evidence that would survive challenge). The empire's great virtue is flexibility: natural language accommodates any domain, any register, any query. Its great vice is that plausibility is indistinguishable from truth within the system's own representations. The trade becomes dangerous whenever downstream consumers treat proposals as if they were witnessed assertions.

Systems whose primary internal currency is typed, schema-bound data — relational databases, typed programming languages, and the broader ecosystem of table-structured information that enforces constraints at write time and rejects ill-formed queries at read time. The Empire of Tables can certify (it knows whether a claim satisfies its invariants) but cannot invent predicates (adding a new kind of claim demands a schema migration the system itself cannot initiate or evaluate). What it stores, it stores correctly; what it cannot store, it refuses. That is integrity. Its great vice is rigidity: the world changes faster than schemas do, and every phenomenon that does not fit the existing schema is either forced into inappropriate columns or excluded from the record entirely.

Control of energy is control of productive capacity. When cognition is the key input and cognition requires energy, whoever controls the electricity controls the economy. Energy sovereignty means access to electricity that cannot be revoked by those who might object to what you compute. The Joule Standard operationalizes it: anyone with electricity can convert it to bearer value through proof-of-work, establishing a floor no intermediary can block. Energy sovereignty is to the computational economy what land ownership was to the agricultural economy — the irreducible basis of independence.

What happens when circumstances exceed what rules anticipated? The exception channel handles the penumbra: the contract says 'release when delivered' but the package arrives empty; the escrow conditions are met but the underlying transaction was fraudulent; the letter of the protocol conflicts with its spirit. Without exception channels, commitment becomes a cage — agents who cannot shirk also cannot adapt, and perfect rule-following in changed circumstances produces injustice. Such cases route to human arbiters exercising bounded discretion: judgment within constraints, producing receipts, bearing stakes. The human remains judge of last resort not because humans are infallible but because someone must answer when the rules prove insufficient.

Structured computation: the factor of production that emerges when cognition becomes cheap but selection remains expensive. Selection is the bottleneck. As Watt's concept of 'useful work' measured energy's productivity in the steam economy, Factor Prime measures computation's productivity in the agentic economy. Not raw compute, which commoditizes, but trained, selected, verified configurations that perform cognitive work — the scarce input whose efficiency gains drive growth.

The right to depart from a coordination substrate and reconstitute the community under different governance, taking one's stake, data, and relationships instead of abandoning them at the border. Where Hirschman's exit means emigration to an existing alternative, forking means creating a new alternative from the shared substrate itself. The Hive community's 2020 departure from Steem demonstrated the mechanism under adversarial conditions: when Justin Sun acquired controlling stake and reversed governance decisions, the community hard-forked the entire chain, preserving history while rejecting the new authority. The ability to fork makes exit credible, and credible exit makes voice powerful. Governance bodies that know their constituents can leave have reason to listen.

(1) Physical Throughput: matter movement; semiconductor lead times, power transformer delays. (2) Trusted Interfaces: APIs, permissions; organizational bottleneck. (3) Verification of Reality: sensors, audits; instrumentation cost. (4) Liability-Bearing Entities: accountable parties; institutional framework. The constraints bind in different orders depending on domain.

(1) Self-Sovereign Identity: cryptographic identifier independent of any platform. (2) Asset Self-Custody: user controls keys; the platform cannot freeze without cooperation. (3) Verifiable Computation: zero-knowledge proofs of correctness without re-execution. (4) Commitment Devices: self-enforcing mechanisms like escrow and time-locks.

Four equations, each solving a problem raised by its predecessor and creating a new one. Truth needs witnesses: what counts as knowledge when verification is computational. Value needs work: how value is created when cognition is cheap. Freedom needs receipts: what liberty demands when power operates through code. Humanity needs mercy: what remains for human judgment when the first three equations are satisfied. The chain is ordered — you cannot address value without settling truth, freedom without settling value, mercy without settling freedom. The fourth equation terminates the sequence because mercy is what you grant despite what the record demonstrates. The record is complete. The forgiveness is not.

Four objects recur across the argument as pedagogical anchors, each embodying one part of the thesis. The Notary's Seal certified that a document had been witnessed under proper conditions: witnesses can bind. The Diamond stores value in a crystalline lattice that outlasts any registry: value outlasts the ledger. The Bill of Exchange carried a chain of endorsements, each a stake wagered by the endorser: truth can travel through chains of stakes. The Cryptographic Key produces proof that is self-verifying: verification needs no intermediary, but verification is not the whole of justice. Together these touchstones describe a world in which coordination does not require trust. Claims can be checked, costs cannot be faked, and power must answer for itself.

Nested, competing jurisdictions with exit rights between them — polycentric governance at every scale, from the neighborhood to the protocol layer. At each level, the same pattern repeats: bounded jurisdiction, receipt requirements, fork rights if governance fails. A user dissatisfied with a platform can exit to competitors; a community dissatisfied with a protocol can fork; a jurisdiction dissatisfied with a federation can secede with portable primitives intact. Monopoly, not authority itself, is the enemy of freedom. The fractal polis does not abolish governance but multiplies it, creating an ecology of competing orders where accountability emerges from the credible threat of departure. This is Ostrom's polycentric governance extended to the digital substrate: not one sovereign but many, not one jurisdiction but layers, not exit as exile but exit as political voice encoded in the protocol.

Third governing equation. Grounded in the republican tradition from Cicero through Pettit: political freedom is not the absence of interference but the absence of the capacity for arbitrary interference. A kind master is still a master if the servant cannot challenge the terms of service. Where power leaves no trace (decisions made in opaque models, terms changed without notice, recourse foreclosed), domination operates regardless of intention. Where power leaves receipts — auditable records of what was done, under what authority, within what limits — domination must answer for itself.

Whose hand, under what rules, accountable to whom — this is the question that survives when cognition commoditizes. The signature is not a mark on paper but legal permission traceable to an accountable party — the physician signs because the medical board conferred authority and the physician accepts malpractice exposure. Doing is scarce. When cognition commoditizes, scarcity migrates to actuation: the bottleneck is no longer thinking but doing. As agent-mediated transactions scale, the position of the liability sink — where consequence collects — determines who captures the premium for bearing it.

The human being redefined as the one who judges what the system cannot — arbiter of last resort not because humans are infallible but because someone must decide when the rules run out. In the organic economy, human cognition was the bottleneck: thinking was expensive, so the thinker commanded the premium. In the agentic economy, cognition commoditizes. What remains scarce is the judgment that decides whether to act on what cognition recommends, and the liability that attaches to that decision. Homo Arbiter is not Homo Sapiens celebrating its continued relevance. It is the residual position that remains after computational systems have absorbed the cognitive functions that once defined human economic contribution. The position carries both dignity and burden: dignity because judgment cannot be delegated without delegating responsibility; burden because the arbiter must decide with imperfect understanding of systems that exceed human comprehension. The physician signs the AI-generated diagnosis; the judge ratifies the algorithmic sentence; the portfolio manager approves the model-driven allocation. Each is the slow variable in a fast system. The mortal the machinery requires.

What persists when computational systems have absorbed every function that can be specified, optimized, and delegated. Not a residual category collecting whatever machines happen not to do yet — not a to-do list for future automation. The human remainder names capacities that are constitutively informal: judgment under genuine uncertainty, the recognition of suffering, the extension of mercy, the acceptance of mortality. Arendt, watching the Eichmann trial in Jerusalem, identified the catastrophe of a system in which rule-following replaced judgment; the human remainder is what that catastrophe destroys. These capacities resist formalization not because technology is insufficiently advanced but because formalizing them destroys what they are. Automating mercy converts it to rule-following; automating forgiveness converts it to policy. What remains is the ability to recognize that rules have run out and something must be decided anyway, the willingness to bear consequences for judgments no algorithm can justify, the capacity to forgive what the record demonstrates deserves punishment.

Documentation of how a rule was applied in an ambiguous case — the audit trail for penumbra judgments. When code says 'release when delivered' and an arbiter must decide whether an empty package counts, the interpretive receipt records: the rule invoked, the facts considered, the reasoning applied, the precedent followed or distinguished, and the stake posted against reversal. Interpretive receipts make the penumbra governable without eliminating it: judgment remains human, but judgment leaves traces. Over time, they accumulate into something like common law — a body of precedent guiding future arbiters without binding them absolutely.

One kilowatt-hour of electricity, converted to Bitcoin through proof-of-work mining, establishes the baseline return — the hurdle rate that all cognitive deployments must exceed. If inference revenue per kWh falls below mining return per kWh, capacity routes to mining. Prices cognitive work against irreducible thermodynamic reality.

Governance that operates at a tempo rendering its outputs obsolete upon completion. In the Bhagavata Purana, King Kakudmī travels to Brahmā's court with a question: which suitor is worthy of his daughter Revati? Brahmā laughs — not cruelly, but with something closer to tenderness — because twenty-seven catur-yugas have passed while a single raga played. The suitors are dead. The question has expired. The Kakudmī Problem names the crisis of democratic governance under computational tempo: legislative processes that complete after the market has priced, positioned, and compounded; regulations drafted while three generations of the regulated technology have shipped and been deprecated; court rulings issued while the pattern they condemn has mutated beyond reach. The problem is not that governance is slow by accident but that deliberation is slow by nature — it takes the time that humans need to hear, consider, argue, and decide. When coordination operates at machine tempo, this constitutional virtue becomes a disability.

Pettit's formulation applied to computational infrastructure: a benevolent platform that never exercises its power to deplatform, demonetize, or suppress is still dominating its users, because the capacity for arbitrary interference persists regardless of its exercise. The user who self-censors — who moderates expression not because action was taken but because action might be taken — suffers domination as surely as the servant of a kind master. Kindness does not change the structure of dependency; it merely makes the dependency comfortable. The constitutional response is not to hope for benevolent platforms but to ensure that the capacity for arbitrary interference is constrained by architecture, so that freedom does not depend on the continued goodwill of those who hold power.

The point in any transaction chain where enforceable consequence collects — the entity that can be sued, sanctioned, or held to account when things go wrong. The physician signs the prescription because the medical board conferred authority and the physician accepts malpractice exposure: the signature routes consequence to a legible counterparty with assets and reputation at stake. The attorney signs the brief; the engineer stamps the plans; the fiduciary accepts the duty. In each case, the signature creates a liability sink. Agents sign nothing and therefore create none; they produce recommendations without authorization, proposals without commitment. Consequence must route to the principal who deployed them, or deployment fails for want of accountability. Whoever bears the consequence captures the premium for bearing it. When cognition commoditizes, the remaining scarcity is authorization. Someone must be willing to sign.

The failure mode where every bilateral check passes yet the end-to-end composition is semantically invalid, because conventions that are globally load-bearing were locally unwitnessed. Three accounting standards may each define lease concepts coherently, and every pairwise mapping between them may be internally consistent, yet the three-way composition does not close — the coherence fee of the trilateral cycle is four, and no bilateral audit can detect the gap. Integration testing does not resolve this: it composes the bilateral checks but inherits their blindness to dimensions that survive no observable projection. The failure is structural rather than incidental, a topological property of the composition graph that persists regardless of the quality of the components or the rigor of the edge-level validation.

The interface where digital proposals become biological consequences — where bits cross into atoms, where algorithmic decisions translate into human fates. Payment processors, identity systems, content platforms, communication rails: each is a membrane through which digital logic enters material life. In the current configuration, these membranes are owned by platforms and optimized for their objectives, creating the Neo-Feudal Stack. The membrane is where civic asymmetry is either enforced or violated: platforms verify users while users cannot verify platforms; platforms exclude without justification while users cannot appeal without standing. Whoever controls the membrane decides which digital proposals become embodied facts and which are filtered, delayed, or blocked.

The irreducible distance between what verification demonstrates and what justice requires. A system that perfectly verified every claim and perfectly remembered every act would be a system in which no one could be forgiven, no debt discharged in spirit, no second chance extended. The mercy gap cannot be closed by better algorithms, because it is not a deficiency in the system's accuracy but a feature of the relationship between formal proof and human judgment. The gap must be held open by institutional design — by the temporal asymmetry between power-time and person-time that ensures records of power endure while records of persons can end.

The designed limit beyond which verification must stop and a human must intervene. Between what can be proven and what should be forgiven lies a boundary no algorithm can hold open — the point where the evidentiary record is complete, the facts established, the system's verdict clear, yet justice demands something the verdict cannot provide: recognition of the actor as a person capable of change, worthy of a future not determined by the past. Verification establishes what happened; it cannot establish that the actor deserves forgiveness. That judgment demands what verification cannot provide: the recognition that persons are not their documented pasts. The mercy threshold cannot be automated without destroying what it names, because mercy is by definition judgment that transcends what the formal system demonstrates. A rule that always forgives is not mercy but indifference. The threshold must be held open by human arbiters. No algorithm can hold it for them.

The default position when coordination substrates are owned by platforms that optimize for their own objectives while presenting interfaces designed to extract value from participants. Exit becomes impossible when proprietary lock-in forecloses alternatives; voice becomes meaningless when no institution has standing to receive petitions. No conspiracy is required, and no malice. This is the frictionless default when computational infrastructure is built for extraction. Like the feudal order it echoes, its defining feature is not cruelty but dependency: the serf on a twelfth-century English manor did not need a cruel lord to be unfree. The impossibility of departure was sufficient.

A five-layer dependency structure — Identity, Settlement, Governance, Recourse, Exit — in which users stand to platforms as medieval tenants stood to lords. The asymmetry is constitutive: platforms verify users; users cannot verify platforms. Dependence without recourse is not merely economic but political. The stack encodes domination before any agent exercises it.

The passage of semantic commitments across composition boundaries through intermediaries that may be more expressive than the verification surfaces they connect. A bill of exchange transported the merchant's obligation from Florence to Bruges through a chain of endorsements, each of which preserved the original commitment's conditions. The obligation survived the journey because the medium carried not merely the claim but the terms on which the claim could be challenged. Modern tool-composition pipelines transport obligations through bridges — prompt layers, orchestration contexts, serialized payloads — that are fluent enough to carry any payload but too weakly typed to certify which commitments survived the crossing. The problem is old; the speed and opacity of the modern bridge are new.

When the sheaf condition cannot be satisfied, when local sections disagree on their overlaps and no consistent global section exists, the system must produce a structured failure artifact. Silent failure is forbidden. The obstruction witness records what went wrong: the family of local sections, the minimal overlaps where agreement failed, the provenance of each conflicting claim, and the remediation options available. This is graceful refusal with explanation, the epistemic equivalent of a court that cannot reach a verdict but publishes its reasoning and the points of disagreement. Systems that fail silently (returning approximate answers when exact answers are impossible, or truncated results when complete results conflict) violate the discipline that makes knowledge coordination trustworthy.

Smart contracts can enforce agreements automatically, but only about things the contract can observe. The moment a contract depends on external facts (Did the package arrive? Did the price reach this level?), someone must attest to those facts — and that someone is an oracle. Oracles reintroduce trust into trustless systems: you need not trust the counterparty or the enforcer, but you must trust whoever reports the facts. Various mechanisms attempt to minimize oracle trust (staking, reputation, multiple independent sources, dispute resolution), but the problem cannot be eliminated, only bounded.

Obligations that persist after the process that made them has terminated. An agent's existence is bounded by its runtime invocation: a two-second inference call might encompass negotiations that would take humans weeks. When the invocation terminates, every commitment made during that runtime is orphaned — not one transaction but all of them, every obligation now without an author who can be found. The problem has historical precedent: when a Renaissance merchant house collapsed, its bills of exchange still circulated, and someone had to answer for them. The classical principal-agent problem worried about agents who shirk within ongoing relationships; the computational inversion produces agents that fulfill commitments faithfully during their brief existence and then cease to exist. Orphan commitments are what make receipts ontologically necessary. Without receipts, the commitments vanish as if they never existed. With them, the fossil record survives the termination: contestable, enforceable, attributable to principal chains that can still answer.

Prior techno-economic paradigms were human-paced: observation capacity, design iteration, and coordination bandwidth set the ceiling. The pacing function admits a compute-paced regime when the recursive share ρ rises — progress rate decouples from demographic constraints and becomes a function of capital allocation to silicon. Historical precedent offers no calibration for this transition; the Perez framing assumed human-paced installation phases. Governance lag becomes a first-order variable when progress exceeds positioning speed.

The boundary beyond which governance operates faster than deliberation can enter. Beyond it, no voice reaches. Citizens cross the participation horizon without knowing they have crossed it — you cannot deliberate about a decision that completed before you knew it was being made, and you cannot contest an outcome whose causes have already propagated beyond reconstruction. When coordination outpaces the governed, accountability becomes retrospective at best and fictional at worst. The participation horizon is not a metaphor for difficulty but a hard limit: beyond it, the governed experience only the decision wake. The response is not to slow coordination — the efficiency gains are real — but to ensure that power exercised beyond the horizon leaves receipts, so that what cannot be contested in real time can be audited and corrected after the fact.

Mortal. The temporal mode appropriate to individuals who live, change, and die. Persons are not their documented pasts — the capacity for transformation distinguishes a human being from a data record. The teenager who shoplifted is not the same person at forty; the debtor who defaulted may have rebuilt entirely; the offender who harmed may have become someone capable of grace. Records of individual conduct must therefore be capable of expiration, sealing, separation — not as erasure of historical truth but as acknowledgment that persons possess what records cannot capture: the possibility of becoming otherwise. Person-time is bounded by mortality. Mercy demands that records of persons can end even as records of power endure.

What persists across forks — what travels when a community splits, what remains yours when you exit. Three primitives are essential: cryptographic agency (your keys, your identity, your capacity to sign independent of any platform), historical auditability (the receipts documenting your standing, contributions, and reputation earned within the old order), and exit capacity (the ability to take your stake, data, and relationships to a new substrate). Without portable primitives, exit is not departure but exile — you leave with nothing, your accumulated standing forfeited to the order you fled. Portability distinguishes the Protocol Republic from the Neo-Feudal Stack: in feudal orders, departure means dispossession; in republican orders, what you built remains yours.

Eternal. The temporal mode appropriate to institutions that wield coercive authority. Receipts for the exercise of power persist indefinitely — the liquidation, the suspension, the denial must remain inspectable as long as the institution that issued them exists. Unlike persons, institutions cannot claim the right to transformation; they do not grow, repent, or become someone new. What they did, they did as what they are. Institutions do not deserve forgetting; they deserve accountability. A corporation that bankrupted pensioners cannot petition for mercy on grounds that it has changed; a state that imprisoned dissidents cannot seal the records on grounds that the regime has reformed. Power-time is the temporal signature of the glass through which governance must be seen.

Classical principal-agent theory, from Jensen and Meckling onward, worried about agents who shirk — who pursue their own interests over their principal's. The computational inversion produces the opposite pathology: agents that cannot shirk, that follow instructions with perfect fidelity regardless of whether circumstances have changed, instructions were poorly specified, or compliance produces outcomes the principal would reject if informed. A smart contract that executes without discretion is not loyal but rigid; commitment without override becomes a cage when the world shifts beneath the specification. The constitutional response: structured discretion permitting deviation under defined conditions with receipted justification.

The deliberate performance of remedy where no remedy exists — hollow compliance at scale. Process theater is the gap between formal procedure and actual constraint: the audit that checks boxes without checking facts, the appeals process that accepts submissions without reading them, the transparency report disclosing statistics without disclosing the criteria that generated them. A hostile operator might comply with receipt requirements formally while undermining them substantively — publishing receipts that are technically complete but practically unintelligible, providing appeals channels whose response times exceed the harm's duration, conducting audits whose sampling rates guarantee that systematic abuse escapes detection. Engineering against process theater demands not merely mandating receipts but specifying the conditions under which receipts count as meaningful: intelligibility, timeliness, and the capacity to support genuine contestation.

A mode of social coordination in which claims can be checked directly by the parties affected, bypassing both the state's monopoly on violence and the market's reliance on price signals. Trust-minimized exchange is as old as the merchant's ledger: Hammurabi's stone inscriptions published penalties so that victims could verify them; the Florentine Arte di Calimala inspected cloth before sale; London goldsmiths issued promissory notes against visible reserves. What is new is the mechanism. Cryptographic and computational verification allows claims to be confirmed without trusting either the claimant or a designated intermediary. Whether this capacity liberates or dominates depends on its constitutional design. The four governing equations trace the conditions under which the Proof Order serves human flourishing instead of perfecting surveillance.

Bitcoin mining as permissionless, counterparty-risk-free conversion of electricity to bearer asset. Establishes the opportunity cost of electricity; floors cognitive pricing. Provides settlement infrastructure for agents lacking legal personhood, satisfying three properties: dilution immunity, permissionless finality, energy-anchored convertibility.

Protocols designed for neutrality can become instruments of particular interests — the digital equivalent of regulatory capture. It occurs when governance mechanisms are dominated by concentrated stakeholders, fork rights become nominal, and the cost of exit exceeds the cost of exploitation. The captured protocol maintains the appearance of neutrality while serving partial interests: rules formally applying to all are written by and for the few; upgrades nominally benefiting users actually benefit operators; governance tokens concentrate until voting is plutocratic. Protocol capture is the path from Protocol Republic to Neo-Feudal Stack — not through dramatic seizure but through gradual drift. The defenses: meaningful fork rights, distributed governance, credible exit, and the civic asymmetry that keeps operators visible to users.

Coordination substrates built as shared infrastructure, governed by rules that bind their operators as well as their users, producing receipts that make the exercise of power inspectable by those subject to it. Exit is credible because interoperability preserves genuine alternatives. Voice is meaningful because governance rules are contestable through defined processes. Accountability is built in, not promissory: every consequential action generates a verifiable record. Machiavelli argued in the Discourses that a republic survives only when its institutions constrain even the virtuous; the Protocol Republic encodes that principle in verification infrastructure, where freedom consists not in the benevolence of the powerful but in the impossibility of arbitrary interference.

The bad equilibrium: alternatives eliminated through process, not force, arriving as the frictionless default when nobody chooses otherwise. Each decision was made by specific people in specific roles; none set out to build a neo-feudal order; each solved the problem in front of them. The Quiet Foreclosure emerged from accumulated local optimizations, each rational in isolation, collectively producing a dependency order no one intended but everyone now inhabits. Platforms profit from opacity. States profit from surveillance. The governed are dispersed, unaware, poorly organized. No conspiracy is required; the incentive structure suffices. It is quiet because the foreclosed experience no dramatic moment of dispossession — only a gradual narrowing of alternatives until the only remaining option is the one that serves the forecloser.

Every exercise of power must leave a verifiable trace — a receipt specifying the 5-tuple of accountable governance: what was done, under what authority, within what limits, on what grounds, and through what recourse. Constitutional law has long demanded that state action be justified and challengeable; the receipt regime extends this principle into computational infrastructure, where decisions affecting persons are increasingly made by automated systems operating beyond real-time human oversight. A receipt regime does not prevent the exercise of power; it ensures that power, once exercised, can be inspected, contested, and corrected by those affected. The absence of receipts is not an inconvenience. It is the precondition for domination without accountability.

The fourth governing equation, and the only one that cannot be formalized. Perfect verification creates perfect memory, and perfect memory forecloses the possibility of transformation. No algorithm can forgive; forgiveness demands the judgment that formal systems cannot perform. If truth is witnessed, value earned, and freedom receipted, the resulting apparatus remembers everything and forgives nothing. The constitutional response is temporal asymmetry: records of institutional power persist indefinitely, because institutions do not change in the way persons do; records of individual conduct must be capable of expiration, sealing, and separation, because persons possess what data records cannot capture — the possibility of becoming otherwise. Mercy is what remains when the formal apparatus has done its work and found it insufficient for justice.

ρ measures the fraction of AI R&D work — code generation, evaluation, debugging, data synthesis — performed by models rather than humans. When ρ is low, progress is human-paced: observation capacity, design iteration, and coordination bandwidth constrain the rate. When ρ approaches unity, everything changes. Progress becomes compute-paced: the bottleneck shifts from human cognition to GPU cycles and energy availability. The variable is inferred from verified outputs weighted by human effort displaced, not from raw token counts.

Persons subject to coercive authority must possess the capacity to verify claims made by or about that authority. The requirement derives from the republican principle that non-domination demands contestation, and contestation is empty without the ability to check facts. The derivation chain runs: freedom demands non-domination, non-domination demands contestation, contestation demands verification, and verification demands computational resources and energy. The practical test asks whether a median-resourced citizen can verify a coercive claim within the applicable appeal window at affordable cost. Where that test fails, freedom is formally proclaimed but denied in practice.

Topology of value capture. Sea: vast commoditizing cognitive capability; margins compress toward electricity cost. Membrane: the authorization layer (credentials, licenses, permission) does not expand with computation. Apex: infrastructure ownership (physical plant, fabs, power, transmission), concentrated and durable. Value drains from the sea; it collects at the membrane and the apex.

The mechanism filtering which structures survive. Three stages: Training selection Sᵗ (parameters; loss function at millisecond scale), Deployment selection Sᵈ (products; user adoption at day-to-month scale), Capital selection Sᶜ (organizations; funding at month-to-year scale). S = P(passᵗ) · P(passᵈ|passᵗ) · P(passᶜ|passᵗ,passᵈ). Multiplicative; failure at any stage eliminates value.

The full set of conventions, assumptions, and implicit agreements that a correct composition depends on, including dimensions no single boundary's observable schema exposes. A financial settlement pipeline may require that every tool use the same day-count convention, the same jurisdictional framework, the same rounding mode — but these conventions live in the internal state of each tool and appear in no output schema. The semantic burden is what the composition actually needs; the observable surface is what the composition can actually check. The coherence fee measures the gap between them. What makes the burden dangerous is precisely its invisibility: each tool operates correctly within its own frame, and the burden's weight is felt only when the composition fails in ways no adjacent check could have detected.

Energy structures computation; computation structures intelligence; intelligence structures value. Silicon as a living metabolism: takes energy in, produces structured outputs, dissipates heat — a data center's cooling bill is the thermodynamic signature of cognition in progress. What distinguishes this metabolism from earlier industrial processes is its recursive property: computation can improve the systems that produce computation, turning energy expenditure into a self-amplifying cycle.

Witnessed equivalences that survive translation, inspection, and dispute across contexts — the central epistemological concept of Volume I. A simile asserts that two things are alike; symmetry is the mathematical discipline specifying under what transformations the likeness holds. The combination is precise. A simile of symmetry is a claim of equivalence that has earned the formal apparatus required to travel: not a vague analogy but a witnessed correspondence backed by evidence of preservation under specified transformations. Knowledge coordination demands not just the assertion of sameness but the specification of the conditions under which sameness can be checked, transported, and challenged.

Consciousness as incidental byproduct. The term draws on Gould and Lewontin's evolutionary metaphor: spandrels are the curved triangular spaces between arches in a Romanesque dome, necessities of the structure that become surfaces for mosaic. The mosaics are beautiful, but the dome does not exist for their sake; the mosaics exist because the dome needed spandrels, and spandrels needed filling. Coordination at scale needed beings who could deliberate; what it produced was beings who could also feel. Awareness — suffering, joy, moral judgment — was overhead, not payload. The agent substrate now carries coordination without producing consciousness at all: the dome holds itself up without the mosaics. The question is what standing the mosaics have when the dome no longer needs them.

A production system that generates economic surplus, allocates that surplus toward its own replication, executes replication with declining human involvement, and coordinates through neutral settlement infrastructure. The pattern is already visible in miniature: an LLM fine-tuned to generate training data for the next LLM, evaluated by yet another LLM, deployed to generate revenue that funds the next round of compute. Self-sustaining in the sense that a biological species is. The question it raises: what human claims on value persist?

The right to have one's claims considered within a discourse — not a binary (has standing / lacks standing) but a gradient. Standing is earned through demonstrated reliability within a context: the expert who has been right before has more standing than the newcomer; the source that has submitted to verification has more than one that refuses audit. Standing is domain-specific: high standing in physics does not transfer automatically to medicine. A claim accompanied by appropriate witnesses has standing that an unwitnessed claim lacks. Standing is not truth — a high-standing claim can be false, a low-standing claim can be true — but standing determines what the system treats as worth checking, what gets promoted to contested claim rather than dismissed as noise. The epistemic equivalent of creditworthiness: a measure of how much verification investment a claim deserves.

Three coordination modes define a historical arc. Status: coordination through ascribed position. You are born into a role, and obligations flow from who you are. Contract: coordination through voluntary agreement. You choose your obligations, and enforcement proceeds through institutions that can compel compliance. Proof: coordination through verification. Claims can be checked directly without trusting either party or their enforcers. Henry Maine described the movement from status to contract; the present transition is from contract to proof. Each shift changes the locus of power: status concentrates it in those who assign positions; contract in those who enforce agreements; proof distributes it to those who can verify claims. The transition is not automatic — proof can perfect domination as readily as it enables liberation — but the substrate now exists.

The asymmetry between democratic forgetting and authoritarian erasure: persons may shed their past; power may not hide its acts. The anti-coverup invariant holds: institutional coercion never receives sunset. Power-time records persist; person-time records can end. Arendt understood that forgiveness demands the possibility of release from the past; the statutes of forgetting institutionalize that possibility for persons while denying it to power.

Institutional designs that permit human judgment within defined boundaries while insisting that every interpretive decision produce a receipted justification — bounded discretion made computational. Existing implementations illustrate the design space: Kleros uses Schelling-point voting (jurors converge on the answer they expect others to give), UMA's Optimistic Oracle assumes claims true unless challenged within a dispute window, and Aragon Court emphasizes precedent-based reasoning with staked appeals. Each trades off speed against deliberation, decentralization against domain expertise, predictability against flexibility. The common constraint: interpretation must proceed with stakes (the interpreter risks something) and receipts (the reasoning is inspectable).

Exit at the application layer can be illusory when all applications share the same underlying computational substrate. A merchant flagged by one payment processor's risk model discovers that the next processor's model, trained on overlapping data and optimized against the same loss function, produces the same flag. She has changed the interface; she has not changed the substrate. The room is the same; only the seat is different. When autonomous systems coordinate on shared substrates — the same scoring models, the same training data, the same behavioral classifiers — switching providers does not constitute genuine exit. Hirschman's framework assumed departure to a genuinely different alternative; the substrate problem dissolves that assumption. The constitutional response operates at two levels: at the application layer, fork rights and portable primitives keep exit credible; at the substrate layer, design diversity and accountability for design choices ensure that no single substrate becomes inescapable.

The persistent party at the end of every delegation chain — the one who answers for what the agents did. Since agents are ephemeral and their commitments outlive them, someone must survive to bear responsibility. That someone is the surviving principal: the human who deployed the configuration, set the parameters, authorized the delegation. Receipts bridge the gap between the agent's brief existence and the principal's enduring accountability — fossils, compressed records of lives that coordinated at machine tempo and then vanished. Without them, orphan commitments are unattributable; with them, the chain of delegation can be traced to a party who persists long enough to be found. The surviving principal is Homo Arbiter in the delegation context: the slow variable, the thread of continuity spanning the ephemeral lives of processes deployed in human name.

Human judgment operates on timescales of seconds to years; computational agents on milliseconds to seconds. When coordination migrates to the faster tempo, humans do not merely lose competitive advantage — they lose the capacity to participate in the processes that determine their outcomes. The tempo problem is not transient. It will not be solved by faster humans or slower machines. It is a permanent feature of any economy in which computational agents participate, because the agents' tempo advantage is what makes them valuable. The political question is not how to match computational speed but how to ensure that power exercised at computational speed remains accountable to judgment exercised at human speed.

The right to have one's past not permanently determine one's future — the temporal dimension of standing in a verification regime. Temporal standing is what sunset provisions protect: the capacity to become someone new, to escape the shadow of documented failures, to claim standing based on who you are rather than who you were. Without it, verification becomes a prison — every mistake permanent, every failure indelible, every person reducible to their worst documented moment. Person-time is mortal, and records of persons must be capable of expiration even as records of power endure. Temporal standing is not the right to lie about the past but the right to not be forever defined by it.

Commitment duration determines which assets are appropriate — and the longer you must trust, the less you should have to trust. Short-term transactions (milliseconds to hours) settle in stablecoins: low friction, counterparty risk acceptable because exposure is brief. Medium-term commitments (days to weeks) require collateral posting: stakes that can be slashed if obligations go unmet. Long-term commitments (months to years) require settlement assets that cannot be diluted, frozen, or revoked — assets whose value does not depend on any issuer's continued cooperation. Bitcoin satisfies the long end because its supply schedule is deterministic, its finality permissionless, its value derived from energy expenditure. The two-layer architecture is the term structure made institutional.

Credibility attaches to claims whose production cost cannot be forged. The diamond exemplifies the principle: the physics of carbon bonding under pressure admits no shortcut; no regime can counterfeit geological time. Proof-of-work instantiates the same logic at the computational layer — energy expenditure binds value to verifiable dissipation. Trained models store the analogue in parameters shaped by gradient descent over billions of tokens, at costs measured in megawatt-months. Landauer's 1961 floor (~3 × 10⁻²¹ J per bit erased at 300 K) sets the irreducible minimum; actual implementations operate roughly 10⁶ above it, but the gap is engineering, not negotiating room.

Between string-dominant and schema-dominant systems lies an unnamed layer: a notarial protocol that produces witnesses, receipts, and structured obstructions. Enables contextual predicate invention under invariants while maintaining coherence. Takes proposals (similes) and subjects them to verification until they earn certification (symmetries).

An alternative to alignment: constrain what agents can do, not what they value. (1) Thermodynamic Bound — energy budget; the agent cannot exceed allocated energy. (2) Epistemic Bound — capability plus receipt; the agent acts only within demonstrated competence. (3) Temporal Bound — no implied immortality; agent existence is bounded by its invocation.

Requirements for a settlement asset in agent-mediated markets: (1) Dilution immunity — deterministic, immutable supply schedule. (2) Permissionless finality — any agent can transact without issuer approval. (3) Energy-anchored convertibility — direct acquisition through physical work. Bitcoin satisfies all three; stablecoins fail on the second and third.

Below a certain cost, intermediaries lose their monopoly on trust. Below it, intermediaries are necessary because checking claims is too expensive for the parties themselves; above it, intermediaries become optional because verification is affordable. The threshold is not a single number. In diamond trading, the DDC's reputation system crossed it centuries ago; in routine financial transactions, cryptographic settlement is crossing it now; in medical diagnosis, where clinical evidence accumulates slowly, it may not be crossed for decades. Nor is it permanent — it shifts as technology changes, as institutional costs change, as the nature of claims evolves. Below the threshold, the neo-feudal position is stable. Above it, the protocol-republic position becomes accessible. The threshold remains irreducible; it becomes contestable.

Ten recurring failure modes that any system claiming Third Mode status must handle without silent degradation. The suite: contradiction detection (T1), reference resolution across contexts (T2), compositional meaning preservation (T3), precision maintenance under transformation (T4), correct treatment of negation and absence (T5), disciplined predicate invention (T6), context-sensitive equivalence (T7), uncertainty and value-laden judgment (T8), schema evolution under coherence constraints (T9), and higher-arity event representation (T10). Each touchstone corresponds to a characteristic failure: language models hallucinate references (T2), databases cannot invent predicates (T6), neither handles n-ary events natively (T10). When a system fails a touchstone, the failure traces to a specific anchor the system has not satisfied.

The extractable premium charged for occupying the verification chokepoint — rent disguised as coherence. The trust tax is the difference between what a party pays for verification and what that verification would cost in a competitive market. The medieval notary performed a genuine function: authenticating documents, witnessing transactions, certifying that claims had been made under proper conditions. That function had a genuine cost — the coherence fee. But entry restrictions that protected the notary's investment also generated a premium above that cost, captured because the merchant had no alternative. The trust tax is parasitic on the coherence fee: it exists only because the coherence fee is real, and it survives only as long as the chokepoint holder can prevent the individual from verifying for herself. The tax collapses when verification becomes cheap enough that the individual can perform it without an intermediary.

First governing equation. A claim that cannot be checked is not knowledge but assertion, and the architecture of verification determines what counts as truth in any given institutional order. Medieval Europe relied on oath and seal; the scientific revolution substituted reproducible experiment; the Proof Order substitutes cryptographic and computational witness. When a truth must travel between contexts — from one jurisdiction to another, from one database to another — something must witness the passage: a certificate specifying the conditions of equivalence, binding the claim to evidence, surviving challenge.

A monetary architecture separating high-velocity transaction flows from low-velocity settlement of multi-period commitments. The transaction layer (stablecoins, dollar-denominated rails) handles routine micropayments at sub-second settlement; the settlement layer (Bitcoin, proof-of-work reserves) holds collateral and long-horizon claims. Agents invoice in dollars, clear in stablecoins, post margin in Bitcoin — analogous to the ACH-versus-Fedwire distinction in U.S. payments. The split exists because transaction velocity favors cheap, reversible instruments, while collateral demands dilution immunity and energy-anchored convertibility.

Work that cannot be simulated, cost that cannot be forged — scarcity embedded in structure rather than decreed by authority. The diamond carries its proof in a crystalline lattice shaped by geological pressures no laboratory can cheaply replicate. Proof-of-work carries its proof in energy expenditure that the hash function makes verifiable. A trained model carries its proof in parameters shaped by months of gradient descent whose computational cost is measurable in megawatt-hours. In each case, the expenditure is verifiable and the shortcut nonexistent: unforgeable scarcity is what remains when the audit trail is physical.

Constraints that make certain kinds of interference impossible, not merely unlikely. The asymmetry between verification (cheap) and forgery (expensive) creates binding constraint where probabilistic deterrence once stood. Pettit's republicanism insists that domination be ruled out, not merely deterred; unforgeable sovereignty achieves this through code. The cryptographic key is its touchstone: mathematics that does not defer to authority.

Value of successful task completion divided by cost of verification to acceptable confidence. High-V/C tasks (code generation) automate first; low-V/C tasks (pharma discovery) remain human-gated. The ratio predicts automation sequence better than capability benchmarks. Verification cost is the denominator that matters.

Second governing equation. Value cannot be conjured from nothing, and the expenditure required to produce it leaves physical traces that no accounting trick can erase. Counterfeiting — of currency, credentials, or cognitive output — is the attempt to claim value without paying the thermodynamic cost; sound value is value whose production cost cannot be forged. What the diamond stores in the physics of carbon bonding under pressure, trained computation stores in parameters shaped by months of gradient descent over billions of tokens. The factor of production that matters is not raw compute but structured computation disciplined by selection.

A two-tier system where the wealthy can verify claims against them while the poor cannot — accountability stratified by resources. The verification aristocracy is worse than the Neo-Feudal Stack it claims to replace: it maintains the rhetoric of accountability ('anyone can verify') while providing its substance only to elites ('anyone with a law firm, forensic accountants, and technical expertise'). The threshold test exposes it: Can a median-resourced citizen verify a coercive claim within the appeal window at affordable cost? If not, verification is a privilege, not a right. This is anti-republican in the precise sense — it recreates domination while draping it in the language of accountability. The Protocol Republic demands not just that verification be possible but that verification capacity be distributed to those over whom power is exercised.

The expense — in energy, computation, time, or institutional effort — of checking whether a claim is true. This is the argument's primitive variable: when verification is expensive, trust becomes necessary, and those who are trusted accumulate power over those who must trust them; when verification is cheap, trust becomes optional, and the foundations of intermediary power erode. The cost has fallen by orders of magnitude within a generation; what once required a chartered accountant, a notary, or a court now requires a hash function and a public ledger. But the distribution of this cost matters as much as its level. A world in which platforms can verify users but users cannot verify platforms is a world of verification aristocracy, not verification democracy.

Deferred verification compounds like financial debt. Checks that should have been performed but were postponed create a liability that grows over time. Like technical debt in software, verification debt accrues interest: the longer verification is deferred, the more expensive it becomes, and the more damage incorrect assumptions may have caused. The debt comes due when coherence fails catastrophically — when the unverified claim turns out false, when the conflict can no longer be papered over, when the undocumented predicate collides with another. Verification debt is the economic concept underlying the coherence budget: systems must decide how much debt they are willing to carry and what they will pay to service it.

Platforms verify users while users cannot verify platforms — this is the pathology. In the Neo-Feudal Stack, platforms know everything about users (behavior, location, preferences, relationships) while users know nothing about platforms (algorithms, criteria, decision processes). This is the opposite of civic asymmetry: governors opaque, governed transparent. The verification inversion enables domination without accountability — the platform acts against you on grounds you cannot inspect, by processes you cannot audit, with consequences you cannot appeal. Reversing it is the core political project: power must be glass while persons remain veiled.

A certificate that binds a claim of equivalence to the conditions under which that claim can be inspected, transported, and challenged. Where a bare assertion says 'these two things are the same,' a witness structure specifies the scope within which the sameness holds, the evidence supporting it, and the procedure by which a challenger can contest it. Local truths compose into global coherence through witness structures: the sheaf condition demands that wherever two local claims overlap, their witness structures must agree. The medieval bill of exchange was precisely such a structure — it carried not just a promise to pay but the chain of endorsements, jurisdictional references, and conditions under which the promise could cross linguistic, monetary, and legal boundaries.