---

I. The shape of the problem

Consider a concrete scenario. A software-operated entity in Singapore executes a share issuance for a Delaware corporation with Pakistani beneficial owners and Seychelles tax residency. Four jurisdictions’ securities laws, tax codes, and sanctions regimes apply simultaneously. Today, this is handled by lawyers billing hours across four time zones. The process takes weeks. It costs tens of thousands of dollars. It produces natural-language opinions whose ambiguity is deliberate.

When this same operation runs at machine speed, with thousands of composed transactions per hour, the lawyer-mediated model becomes infeasible. A machine-speed execution path cannot pause six weeks for a legal opinion. The compliance infrastructure that evaluates each action in isolation, enforced by humans who cannot hold the full compositional picture, cannot detect the class of failure where individually legal actions compose into globally illegal behavior.

A share issuance that is legal, a beneficial ownership transfer that is legal, and a cross-border capital movement that is legal can compose into a sanctions evasion structure that is criminal. No individual compliance check catches this. The violation is compositional. It exists only in the aggregate.

Three consequences follow. First, violations are detected only after the fact, when the economic harm is realized and the liability is irrecoverable. Second, regulatory arbitrage exploits the latency differential between jurisdictions: a multi-step transaction can complete before any single regulator has evaluated the first step. Third, audit reconstruction becomes infeasible because current compliance practice produces no tamper-evident record of the reasoning that led to each compliance determination.

The infrastructure failure is already observable in simpler contexts. In August 2012, Knight Capital deployed a trading algorithm that lost $440 million in 45 minutes. The failure was not decision quality. It was infrastructure: no circuit breaker, no mutation firewall, no fail-closed gate on anomalous behavior. The algorithm operated exactly as programmed, at machine speed, within an infrastructure that had no mechanism to constrain it. The Knight Capital disaster was a single automated execution path in a single jurisdiction executing a single type of operation. The compliance problem this paper addresses is the same failure mode, generalized across multiple automated execution paths, multiple jurisdictions, multiple operation types, and compositional interactions between them.

The standard responses, more compliance officers, monitoring software, compliance-as-a-service vendors, address cost while leaving structure unchanged. Adding headcount scales linearly with transaction volume and sub-linearly with compositional complexity. The attack surface grows combinatorially as automated systems compose multi-step plans; the defense grows linearly with headcount. Monitoring software detects violations after commitment. That is adequate for human-speed activity where remediation is possible and inadequate for machine-speed activity where a thousand dependent transactions execute in the interval between violation and detection. Vendors replicate the same analog infrastructure at lower cost without touching the structural problem.

Blockchain-based approaches attempt programmability but abandon the institutional structure that gives economic action its legal meaning. A smart contract can enforce a rule, but it cannot produce evidence that the rule it enforces corresponds to the statute it claims to implement, that the statute was in force at the time of the act, or that a court in the relevant jurisdiction would accept the contract’s determination as dispositive. Legal validity is not a property of code execution. It is a property of the relationship between code execution and sovereign authority. Systems that sever this relationship gain programmability but forfeit legal validity.

II. Why the solution must take this shape

Start from the hardest constraint and work outward.

Sanctions must fail closed. This requirement is statutory: 50 U.S.C. Section 1705 (IEEPA penalties), EU Council Regulation 269/2014, and equivalents in other jurisdictions. OFAC, UN, and EU sanctions carry extraterritorial reach. Licenses, exemptions, carveouts, and shared-authority recognition must be resolved inside the sanctions evaluator before the verdict is returned. A transaction that remains blocked by sanctions cannot become unblocked by composing it with a jurisdiction that has no sanctions objection.

This single requirement, a violation must block, and no composition of passing evaluations can override it, forces the algebra. A system in which a single failure in any domain absorbs all passing results requires an operation that behaves like logical AND with an absorbing bottom element. This is a lattice meet, the operation that guarantees the blocking property.

The model partitions corporate compliance into twenty-three regulatory domains: AML, KYC, Sanctions, Tax, Securities, Corporate, Custody, DataPrivacy, Licensing, Banking, Payments, Clearing, Settlement, DigitalAssets, Employment, Immigration, IP, ConsumerProtection, Arbitration, Trade, Insurance, AntiBribery, and Sharia. Each domain must be independently evaluable, because a firm can be compliant in AML and noncompliant in securities, and the system must represent this without collapsing the distinction. The compliance state of an entity is a vector of verdicts, one per domain.

Each domain verdict carries a two-axis value. One axis is a compliance degree (Compliant, Pending, NonCompliant); the other is an applicability status (Applicable, NotApplicable, Exempt). Compliant means proceed. NotApplicable means the domain is irrelevant to this entity in this jurisdiction. Exempt means excused from this requirement. Pending means the evaluation is incomplete, the system cannot proceed until it resolves. NonCompliant means a violation was found; the operation is blocked.

A reasonable objection: real regulatory outcomes include “Largely Compliant” and “Compliant with Conditions.” Where do these map? A human compliance officer managing a remediation timeline needs these gradations. An automated system answering a gate question, does this operation proceed or not?, does not. “Largely Compliant” means conditions remain unmet. The operation cannot proceed. That is Pending. “Compliant with Conditions” is Pending until the conditions are met. The two-axis cell captures the distinctions relevant to the gate decision. The bureaucratic gradations live in the metadata attached to each cell, reason codes, evidence references, remediation timelines, not in the algebra.

The lattice models the output space of regulatory verdicts, not the reasoning process that produces them. When the FCA evaluates a firm’s AML posture, its deliberation is rich, contextual, and qualitative. The result of that deliberation is a verdict: the firm may proceed or it may not. The lattice captures the structure of the verdict space. It does not claim to capture the structure of the FCA’s reasoning.

Residual and remediation. Any correct system must satisfy the blocking property; the lattice’s distinctive contribution is the separation between implication reasoning and operational remediation. The residual is Heyting implication on the per-domain factor and on the Applicable-fragment product, where both are finite distributive lattices and Birkhoff gives the Heyting structure. It reasons about additional constraints. It does not, by itself, improve a failing coordinate.

Without algebraic structure, determining what an entity needs to become compliant requires enumerating every possible remediation and testing each one. This becomes combinatorial as domains interact.

Given current state A and bound B on the Applicable fragment, the residual A => B computes the weakest additional constraint that can be composed with A while remaining within B. Because meet is restrictive, this is not a recipe for making A more compliant. The recipe for improvement is a remediation operator: identify the stale attestation, missing proof, or discretion-hole fill that can replace the underlying fact, perform the authorized transition, and re-evaluate.

This transforms compliance from a pass/fail check into a typed planning problem with proof obligations. Where am I? Where do I need to be? Which authorized transitions can change the facts? The order structure lets the system check proposed remediation paths against the target constraint surface. On the full state space including NotApplicable and Exempt, the structured meet returns a MeetResult sum rather than a single verdict, because no total semilattice on the full five-inhabitant factor simultaneously respects the orthogonality of applicability status to compliance degree and preserves the compliance signal. The production core is mechanized in TensorAlignment.v: the n-ary meet returns a MeetResult, the Applicable fragment agrees with ordinary grade meet, and mixed App/NotApplicable/Exempt axes are reported without collapsing provenance. Full audit correctness for production proof bundles remains an open theorem suite.

The lattice’s distinctive capability is the combination of scoped residual reasoning, explicit remediation transitions, and structured mixed-axis outcomes.

III. What this is not

The compliance lattice models verdict composition, not regulatory reasoning.

Regulators exercise judgment. The FCA evaluating AML compliance considers the proportionality of controls, the quality of suspicious activity reports, the remediation trajectory, the board’s understanding of its obligations. This is holistic, contextual, and irreducibly qualitative. Two regulators can examine identical facts and reach different conclusions because regulatory judgment is perspectival.

The lattice does not touch this. It takes the verdict as given, whatever process produced it, and provides the algebra for composing verdicts across domains and jurisdictions. This is like a type system that does not model what happens inside functions, only the interfaces between them. The richness of the FCA’s reasoning is unbounded. The set of possible conclusions is not.

Similarly, not all law is computation. Constitutional principles, common-law reasoning about novel fact patterns, the interpretive activity that transforms text into meaning, these are irreducibly human. The essay “Laws Are Programs” identifies four structural properties of legal rules, defeasibility, temporal stratification, authority hierarchy, and typed discretion, that the regulatory stratum of law already possesses. But the regulatory stratum is not all of law. It is the administrative layer: file your annual return within 28 days, appoint at least one natural-person director, report beneficial ownership changes within 14 days. These rules are already programs written in English prose. The insight is not that all law reduces to computation. It is that the layer of law that governs routine institutional operations is already procedural, and a programming language that respects defeasibility, temporality, and authority would be structurally adequate for that layer.

The places where the abstraction diverges from reality are exactly the places where typed discretion holes must route decisions to human judgment. A dependently-typed rule language that marks the boundary between mechanical derivation and human decision claims only that the boundary can be precisely identified and formally marked; if-else compliance systems leave that boundary implicit.

Several limits are explicit. Proofs on the Applicable three-grade factor are simple, the elementary laws can be verified directly. The real work is in the 23-coordinate tensor, the mixed applicability axis, the rule evidence behind each coordinate, and the corridor translations between sovereign rule systems. Legal citations in the regulatory encodings need human verification; trained engineers maintaining these references will make errors, which is itself evidence that manual legal reference management does not scale. Most jurisdictions need simpler infrastructure today: a single-jurisdiction compliance check does not require corridor composition or tensor algebra. But an entity incorporated in 2026 will exist in 2056. The compliance history for that entity must remain verifiable for the entire duration. Retrofitting verifiable history onto a system that never recorded provenance requires reconstructing evidence that no longer exists.

IV. Why sovereignty cannot be an afterthought

In February 2022, the United States and European Union excluded major Russian banks from SWIFT. Overnight, businesses that had operated normally for decades discovered that their ability to send and receive international payments depended on the continued goodwill of a consortium headquartered in Belgium and subject to EU and US foreign policy. Sberbank, the largest bank in Russia, lost access to the global payments backbone because sovereignty over the messaging layer resided elsewhere.

The exclusion revealed a structural dependency that existed all along. SWIFT, chip export controls, correspondent banking de-risking, every instance follows the same pattern. Centralized infrastructure concentrates control. Concentrated control attracts sovereign interest. Sovereign interest has repeatedly produced weaponization. The sequence has recurred in telecommunications, internet routing, semiconductor supply chains, and financial messaging.

The obvious objection is that a simpler, centralized design would avoid this complexity. A centralized, simple design is the architecture of SWIFT. The simplicity argument, followed to its conclusion, reproduces exactly the system whose weaponization motivates the alternative.

Two architectures are available for institutional infrastructure that governs cross-border economic activity.

Imperial architecture. One jurisdiction’s rules become the base layer. All others’ rules are exceptions or adaptations. In practice, this means US or EU law becomes the default, and every other country’s sovereignty is subordinated to the architectural choices of the dominant jurisdiction. This is the current reality.

Sovereign architecture. Each jurisdiction encodes its own rules in its own authority. Cross-border operations compose these rules through explicit, auditable protocols. No jurisdiction’s rules are architecturally privileged. Conflicts are resolved through bilateral agreements, not imperial default.

The sovereign architecture is harder to build. The compliance tensor, jurisdiction-resolving operation DAGs, corridor composition, these are genuinely complex. But they are complex because the problem is complex. Suppressing that complexity makes it implicit and places it under the control of whoever built the system.

The telephone network was centralized. AT&T controlled it. The US government could and did wiretap it at will, because centralized switching meant centralized intercept capability. The internet was designed as a decentralized packet-switching network because DARPA researchers understood that centralized networks have centralized failure modes and centralized control points. Packet switching was dismissed as an academic indulgence, needlessly complex relative to proven circuit switching. That “academic indulgence” became the backbone of the global economy because its architectural properties, no single point of control, no single point of failure, protocol-level interoperability without institutional trust, proved to be load-bearing requirements. The same structural argument applies to institutional infrastructure. Federated sovereignty is the institutional equivalent of packet switching: more complex at the protocol layer, but fundamentally more resilient at the system layer, because no single sovereign can unilaterally disconnect a participant.

The current system is a centralized Postgres database. One schema, one writer, one binary. This is the starting point. The target is federated sovereignty, zones that communicate through signed bilateral protocols, each encoding its own regulatory regime. The compliance tensor algebra, the corridor model, the mutation firewall, these work correctly on a single database today. They are designed so that when the federated protocol arrives, the algebraic properties that guarantee correctness on one node will guarantee correctness across nodes.

V. The enforcement boundary

In any system where automated execution paths perform writes with economic consequences, there must be exactly one path through which all state-changing operations pass. This is the reference monitor property from security architecture, applied to institutional infrastructure.

If there are two write paths, the system’s integrity is determined by the weaker of the two. If there are zero constrained write paths, writes execute unconstrained. A caller or developer will route through whichever path applies fewer constraints. The attack surface of a system with N write paths is the weakest of the N. The attack surface of a system with one write path is that path.

The gate enforces two invariants, and only two. First, sanctions fail-closed evaluation: if the entity or counterparty remains non-compliant after applicable sanctions licenses, exemptions, carveouts, and shared-authority evidence are resolved, the write is rejected. This is a legal requirement, 50 U.S.C. Section 1705, EU Council Regulation 269/2014, not an architectural choice. Second, audit journaling: every committed write produces an append-only event recording what changed, who authorized it, what evidence supported the determination, and when it occurred. The record is immutable.

The design is minimal: one gate enforcing two checks, one for legality and one for accountability. It does not matter whether the write was initiated by a human operator or an automated process. Every write passes through the same function. The function enforces the same invariants.

Everything else in the system, the 23-domain compliance tensor, the corridor composition, the Applicable-fragment residual, the propagation graph, exists to make the inputs to this gate correct. The gate itself is simple. The machinery that ensures the gate receives correct inputs is where the complexity lives.

VI. Why institutional knowledge has not compounded

Scientific knowledge compounds because observations are recorded in structured form, transmitted through publication, and composed through citation and integration. A paper builds on a paper. Each contribution is available to every subsequent researcher.

Institutional knowledge, how jurisdictions behave in practice, which compliance processes create friction, which regulatory patterns predict healthy institutions versus fragile ones, has never compounded. The reasons are structural.

It cannot be recorded because it is produced as a byproduct of economic activity, not as a primary product. When a compliance officer evaluates a corporate formation and discovers that a particular documentation pattern correlates with shell structures, that observation exists in their memory. It was never the point of the work. The point was to evaluate the entity. The institutional knowledge was incidental to the compliance determination.

It cannot be transmitted because sovereignty prevents it. A jurisdiction’s regulatory record is a sovereign asset. Even where the will to share exists, the observations were never structured for transmission. They live in an officer’s judgment, not in a transferable form.

It cannot be composed because observations across jurisdictions use different formats, different ontologies, different regulatory frameworks. What Singapore calls “AML compliance” and what Abu Dhabi calls “AML compliance” overlap substantially but differ in ways that matter. Without a shared algebra, observations from one jurisdiction cannot meaningfully combine with observations from another.

When institutional operations become programmable, when entity formation, ownership transfer, compliance evaluation, and identity verification produce typed, structured observations as a natural byproduct, all three conditions change simultaneously. The observations are recorded because the infrastructure produces them. They are transmissible because they are typed. They are composable because the compliance tensor provides a shared algebra across regulatory domains.

The observation corpus is a byproduct of the compliance infrastructure described in the preceding sections. A compliance evaluation that produces a 23-domain tensor verdict has, as a byproduct, produced an observation about how that jurisdiction’s regulatory regime behaves in practice. These observations are costly signals: every entity formation involves real capital, real liability, a real jurisdictional choice. These are institutional facts that cost something to produce, that cannot be scraped from the web or synthesized from statute text.

Aggregate those observations across thousands of entities, and patterns emerge: which compliance domains create disproportionate friction in offshore jurisdictions, which entity structures attract regulatory scrutiny, where sanctions screening catches real threats and where it generates false positives. At scale, the observations could distinguish jurisdictions that enforce their law from those that merely publish it. They could identify the absence of fraud as well as its presence: the entity structures and governance practices that produce consistently clean compliance histories.

Every entity that operates through the infrastructure produces observations that sharpen these distinctions.

VII. The jurisdiction graph

Jurisdictions form a graph with asymmetric edges. A corridor between two jurisdictions specifies, for each of the twenty-three compliance domains, whether the destination re-evaluates or accepts the origin’s assessment, and how recognized evidence is translated. The corridor parameters carry a re-evaluation mask and a partial recognition map. These compose as finite meet-style data: recognized domains narrow under intersection, grade recognition can only become more restrictive, and staged domain maps compose by partial-function composition. The asymmetry is fundamental: sovereignty is not symmetric. Singapore may accept Abu Dhabi’s AML assessment without Abu Dhabi reciprocating, because the two jurisdictions make independent sovereign decisions about what evidence they trust.

Each new jurisdiction adds a node and potential edges to every existing node. By fifty jurisdictions, the graph reveals structure that is invisible at smaller scale: which compliance domains are universally re-evaluated regardless of origin, which port cleanly within legal-tradition clusters, which depend on treaty networks, and which create pathways between jurisdictions whose regulatory architectures happen to be structurally equivalent despite no formal agreement.

The compliance state of an entity adding a harbor is computed in stages: translate recognized origin coordinates through the corridor map, force fresh destination evaluation on the re-evaluation mask, compose recognized Applicable coordinates by meet, and return MeetResult-style structured outcomes for mixed applicability. The propagation graph determines when those compositions must be refreshed: a sanctions status change in one jurisdiction cascades through causally downstream domains across every corridor the entity participates in. The causal structure ensures correct propagation without brute-force recomputation.

The translation problem is hard. Real cross-jurisdictional compliance mapping is not a clean projection. Equivalence determinations between regulatory frameworks are partly technical and partly political. The EU’s equivalence decisions under MiFID II have been used as instruments of trade policy. The same source jurisdiction produces different translations depending on the political relationship, the entity type, and the specific bilateral agreements in force. The claim is only that every translation can be made explicit, auditable, and versioned; current translations reside in practitioners’ judgment and are lost when they retire.

VIII. How the model calibrates

The jurisdiction graph encodes what the rules say. It does not encode how they work in practice. Rules on paper and rules in practice diverge. The question is how much, and where.

A similarity function over the jurisdiction graph combines four dimensions: compliance domain overlap, legal tradition distance, treaty network overlap, and regulatory genealogy. At launch, this function is a structural heuristic. It predicts transfer of an institutional innovation proven in Abu Dhabi to Singapore because they share most compliance domains and a common legal tradition.

Then the prediction is tested. Singapore adopts. The outcome is observed. If the transfer succeeded, the operational similarity between the two jurisdictions increases. If it failed, it decreases. After twenty adoption outcomes, operational evidence outweighs the structural heuristic. The function converges to reflect actual transfer success, not predicted transfer success. If the fiftieth prediction is no more accurate than the first, the thesis is wrong.

This enables controlled experimentation on institutional design. Institutional reform currently proceeds without controlled comparison. A jurisdiction changes a regulation. Years later, someone attempts to assess whether it worked. The assessment is confounded by everything else that changed in the interval. There is no control group. There is no measurement protocol.

A hypothetical example. Researchers observe that corporate formations in common-law offshore jurisdictions fail identity verification at first attempt forty percent of the time. The failure mode is consistent: missing beneficial ownership documentation. The hypothesis: decomposing verification into a preliminary declaration at formation, followed by verified documentation within thirty days, will reduce failures without weakening compliance. Treatment jurisdictions deploy the modified process. Control jurisdictions do not. Three months later, first-attempt pass rates in the treatment jurisdictions rise by twenty percentage points. Controls show no change. Another jurisdiction adopts based on the evidence. Six months later, its pass rate improves by fifteen points. One experiment, six months, six jurisdictions.

A distinction matters for privacy. The calibration does not learn from entity-level data. It learns from jurisdiction-level adoption decisions. Whether a jurisdiction deployed a regulatory innovation is a public institutional act. Whether it improved aggregate compliance is a jurisdiction-level outcome, computed locally. Only the aggregate crosses the boundary. The system observes governments, not persons.

The limitations are real. The privacy model degrades below a certain threshold of entities per jurisdiction, small-jurisdiction observations are either suppressed or dominated by noise. Bilateral exchange means each jurisdiction’s model reflects its connectivity, not the full network. And the model captures operational telemetry, the outputs of institutional processes, not how regulators interpret ambiguous provisions, which informal relationships carry weight, or how political transitions shift enforcement. Patterns in telemetry permit inference about the underlying dynamics; they do not measure them directly.

IX. The machine-speed safety argument

The argument for building compliance infrastructure for machine-speed execution is that the economy already admits machine-speed writes while the institutional layer remains organized for human review.

The absence of a mutation firewall does not prevent a financial write. It means the write executes without audit journaling. The absence of sanctions fail-closed evaluation does not prevent a transaction with a sanctioned entity. It means the transaction proceeds without interdiction. The absence of an append-only event store does not prevent automated state changes. It means those changes occur without an immutable record.

Building compliance infrastructure for automated institutional action is not an enablement strategy. It is the minimal condition for legal and evidentiary control over activity already possible at machine speed.

A common objection says that present automated decision systems are unreliable. That is true and irrelevant to the structural point. The infrastructure does not assume wise proposals. It constrains writes, records evidence, forces sanctions checks, routes open-textured determinations to authorized humans, and rejects operations whose proof obligations do not close.

This is not a novel regulatory pattern. Human traders are imperfect. Human compliance officers are imperfect. The regulatory response to human imperfection is not exclusion but containment: regulations, audits, circuit breakers, and supervisory review. A mutation firewall is a reference monitor. Sanctions fail-closed evaluation is a legal constraint made executable. The compliance tensor is the pre-commit constraint surface.

The operator, the human or institution that authorizes the automated process, remains accountable. The infrastructure makes that accountability enforceable rather than aspirational. A proposed sanctioned transaction that fails closed at the infrastructure boundary fails safely. The same proposal in the absence of infrastructure causes a sanctions violation. The capability is identical in both cases. The outcome is determined by the infrastructure.

The accountability question, who is liable when automated operation causes harm, is a political question that legislatures must answer. This infrastructure does not answer it. It ensures that when the question is asked, there is an immutable, cryptographically verifiable record of exactly what happened, what evidence supported each determination, and which human operator authorized the execution path. Whether that record is sufficient for liability assignment is for courts to decide. That the record exists at all is what the infrastructure guarantees.

X. What this enables

Institutional knowledge has never compounded. Experts produced it, jurisdictions locked it, successors reproduced it from scratch, retirement destroyed it. Expertise in Singapore transferred poorly to Abu Dhabi. Operations in one jurisdiction made no other jurisdiction’s operations faster, cheaper, or more reliable.

When institutional operations flow through programmable infrastructure with algebraic composition guarantees, every operation produces observations that sharpen the model. Every jurisdiction that joins extends the graph. Adoption outcomes calibrate the predictions. Experiments teach the network which structural features govern transfer.

The encoding of hundreds of jurisdictions into a regulatory corpus is years of work. Building the corridor topology requires bilateral government trust, months to years per relationship. Operational intelligence emerges only from adoption outcomes across jurisdictions connected through those corridors. Each layer presupposes the one before it. The regulatory codification enables deployment. Deployment enables corridors. Corridors enable intelligence. Intelligence enables diffusion. The cold start that protects whoever builds this first is the same one that delays external validation.

Until the network operates at scale, the compounding thesis rests on architecture and specified proof obligations. The first two layers, single-jurisdiction compliance infrastructure and entity operations, produce value today. The third, cross-jurisdictional intelligence, is specified; scale is the condition that makes its claims verifiable.

The papers that follow this one construct the specific pieces: how the compliance tensor algebra works, how the rule language handles defeasibility and temporal stratification, how corridors compose, how the cross-zone protocol preserves sovereignty, how the Smart Asset VM provides a programmable compliance surface, how the observation corpus converts operational telemetry into calibrated predictions. Each paper presents one construction.

An objection: eighty percent of the value might be achievable with a fraction of the complexity. For a single jurisdiction, yes. A database, a compliance rules table, a sanctions screening call, an append-only audit log. It handles entity registration, ownership tracking, compliance evaluation, and audit trails for one jurisdiction at a time. Most deployed compliance systems take this form, and for single-jurisdiction operations it is adequate.

The twenty percent of value that a simpler system cannot provide is the part that matters over decades: algebraically sound cross-jurisdictional composition, the Heyting residual as a planning tool, the propagation graph that cascades domain changes correctly across corridor topologies, and the observation corpus that compounds operational intelligence. A single incorporation does not require them. They are required for the entity that operates across three jurisdictions, and for the entity whose compliance history must be verifiable in thirty years, and for the corridor crossing that reveals which regulatory domains create structural friction between legal traditions.

The complexity is front-loaded by design. It lives in the type system, the algebra, and the test suite, not in the deployment or the runtime. Infrastructure that must run for decades should be hard to build and simple to operate.

This paper makes the case that the constructions belong together, that the problem of governing machine-speed economic activity across sovereign jurisdictions forces this particular shape of solution. The meet is forced by the blocking requirement. The residual is forced by the planning requirement. The corridor model is forced by the sovereignty requirement. The mutation gate is forced by the accountability requirement. The observation corpus is forced by the compounding requirement. Whether calibrated telemetry recovers real institutional dynamics is the central open empirical question.