The nineteen structural primitives

Authority is the delegated right to make binding decisions within a defined scope. Every authority instance either is a root authority (e.g., a board of directors) or traces through a delegation chain to one (B5).

Semantic boundary: Authority is the right to decide; Capacity is the ability to execute. Authority can be delegated through chains of arbitrary depth — the protocol enforces chain integrity (B5), not chain length. Profiles (§21) may configure warnings at unusual delegation depth for their organizational context, but no protocol-level depth limit exists.

Entity. The ownership primitive. Every governed object belongs to an Entity — a person, role, system, or organizational unit. Entity supports recursive structure (entities contain entities, following the recursion principle) and maps to legal personality where applicable. Without Entity, primitives cannot be attributed to organizational actors.

Namespace. The scoping primitive. Namespace provides a four-scope governance model for concept definitions:

Concepts graduate upward through the scopes (sandbox → tenant → domain) via evidence accumulation and authority approval, following the claims graduation model (§6.5). The concept registry uses URI patterns enabling cross-domain alignment.

Disambiguation. These four Namespace scopes are governance scopes within the Namespace primitive — they control who can define concepts and how those definitions evolve. They are not related to the five primitive tiers (Tier 1–5), which organize the 19 primitives into an activation hierarchy. The scopes govern term definitions; the tiers govern primitive availability.

Decision is the choice point where multiple options are evaluated and one is selected. It is the atomic governance act — the moment where evidence, authority, and intent converge into a binding selection.

Semantic boundary: Decision is the selection act; Authority is the right to make that selection. A decision without authority is illegitimate. A decision without an account context is unverifiable — B4 (§5) prevents this. Decision is also distinct from Intent: intent frames what to pursue; decision determines how to pursue it.

Activation. The pre-kinetic readiness state — whether conditions for governance action are met. Activation tracks friction metrics and produces work handoff contracts when readiness conditions are satisfied.

Activation is distinct from both Decision and Work. An organization can be activated (ready to act) without having decided what to do (Decision) or begun doing it (Work). Activation occupies a unique semantic position: readiness that is neither choice nor effort.

Interpretation. The agent/human handoff point for meaning resolution in governance contexts. When an AI system processes governance data, Interpretation captures what the AI understood, at what confidence level, and under what assumptions. Interpretation handles eight resolution types: classification, extraction, recommendation, analysis, generation, translation, prediction, and gap identification.

All AI interpretations enter the truth type system as Derived evidence and land in staging. Promotion to Declared or Authoritative requires human review. A stakes-by-novelty routing matrix determines review intensity:

Red-zone proposals (high stakes, high novelty) are rejected if approved without evidence of review — the rubber-stamp detection invariant prevents governance theater.

Constraint expresses rules that apply universally, regardless of who invokes them. Compliance requirements, physical laws, organizational policies, and regulatory mandates are all Constraints. Unlike Authority, which differentiates between actors, Constraint binds all actors equally.

Constraints are implemented as SHACL shapes in a dedicated shapes graph, separate from the class ontology. The shapes graph operates under the Closed World Assumption: what is not explicitly permitted is prohibited. At runtime, the shapes graph validates primitive instances against constraint definitions, and violations produce Evidence records (validation reports with truth type Authoritative). This separation enables profile layering — different substrate profiles (§21) apply different constraint sets to the same primitive classes without modifying class definitions. Each constraint specifies an enforcement mode: Blocking (prevents the operation), Warning (allows but alerts), Logging (records silently), or Advisory (suggests alternatives).

Semantic boundary: Constraint is the rule; Authority is the right. Constraint answers "is this permitted regardless of who is asking?" — Authority answers "who is allowed to decide?" A constraint applies to all actors uniformly; an authority delegation applies to specific actors within specific scopes.

Commitment is the formal acceptance of responsibility to complete work toward an objective. It creates the binding relationship between an actor and an outcome.

Semantic boundary: Commitment is acceptance of responsibility; Capacity is the ability to fulfill that responsibility. A commitment without verified capacity is an impossible promise — the behavioral invariant B2 (§5) prevents this structurally. Commitment is also distinct from Work: committing to an outcome does not yet mean executing toward it.

Environment Interface. The boundary primitive enabling substrate awareness of external systems, data sources, and environmental conditions. Environment Interface is the sensor surface where the substrate's domain-agnostic interior meets specific external domains.

The architectural role is bidirectional:

Environment Interface is the formal hook through which the Dimensional Business Index (§10) observes the substrate's perimeter topology. It maps to Policy Projection Domains 5 (Access, Identity & Boundary) and 6 (Exception, Escalation & Override), governing the substrate's boundary with external systems. Full field-level specification is defined in the Minimum Viable Record (§7).

Evidence is a data artifact that supports, refutes, or contextualizes a governance claim. Every evidence artifact carries an epistemic classification — its truth type (§6) — distinguishing Authoritative evidence (human-verified, immutable), Declared evidence (human-asserted, attestation-backed), Derived evidence (machine-generated, requires promotion), and Opaque evidence (boundary-enforced, structurally uninspectable — §6.1). The behavioral invariant B3 (§5) enforces this classification on every instance.

Semantic boundary: Evidence is the artifact; the Truth Type System (§6) classifies its epistemic status. Evidence is also distinct from Account: evidence is a datum that supports or refutes; an account is the persistent record of state that contextualizes decisions.

Context. The situational primitive. Every governance action occurs in a Context — temporal, spatial, environmental, and relational framing. Without Context, primitives exist in isolation: decisions without circumstances, evidence without provenance conditions.

Orientation. The pre-decisional cognitive stance — how a system perceives its position relative to its environment. Orientation establishes the interpretive frame through which evidence is understood and decisions are contextualized.

Orientation is distinct from Intent. Orientation integrates environmental signals, prior experience, and pattern recognition before decision commitment. Intent specifies direction after orientation has established the cognitive frame. The distinction holds across five independent analytical frameworks: Boyd's OODA loop (Orient as the "schwerpunkt," distinct from Decide), Endsley's Situation Awareness (Level 2 Comprehension as the pre-decisional integration layer), Klein's Recognition-Primed Decision model (assessment phase precedes decision generation), VSM System 4 (environmental scanning before policy formulation), and Weick's Sensemaking (meaning construction precedes action). The two primitives vary independently: orientation can shift without intent changing, and intent can change without full re-orientation.

Learning. The institutional mechanism converting dissonance — misalignment between orientation and outcomes — into structural change. Learning operates through a four-stage mechanism across organizational levels:

All four stages must operate at their respective organizational levels. Feedback loops must close across levels (institutionalization feeds back to individual intuition). Tacit-to-explicit conversion requires collaborative processing, not individual reflection alone. Exploitation without exploration produces organizational brittleness.

Account is the persistent record of governance state: who committed to what, when, and with what outcome. It provides the state context against which decisions are evaluated and from which audit trails are constructed.

Semantic boundary: Account is the state record; Evidence is a data artifact within it. Account provides the historical and current context that makes decisions verifiable — B4 (§5) requires every decision to link to an account. Account is not merely a log — it is the substrate's representation of organizational state.

Identifier. The addressing primitive. Every governed object has an Identifier — a unique reference (UUID, URI) providing addressability, version tracking, and cross-reference capability. Without Identifier, primitives exist in semantic space but cannot be referenced, queried, or linked.

Cycle. Temporal governance pattern enabling periodic structures — fiscal years, review cycles, audit windows, budget periods — as first-class governance objects.

Cycle is not a convenience primitive. Without it, every substrate implementation invents ad-hoc temporal conventions: encoded as constraint logic, embedded in application code, or implicit in workflow configuration. These ad-hoc representations make cross-substrate temporal queries impossible. When two substrate instances encode "Q1 review cycle" differently, no governance query can span both.

Cycle makes temporal governance legible: a queryable, first-class primitive that enables cross-substrate temporal alignment. Organizations that govern in periodic terms — virtually all of them — require native temporal governance rather than encoding periodicity as constraint logic scattered across the governance graph.

Intent captures the expressed desire or objective that motivates a decision or action. It is the "why" behind governance — without it, decisions and commitments exist but cannot be traced to purpose.

Semantic boundary: Intent specifies direction; Commitment accepts responsibility for pursuing it. An intent can exist without anyone committing to it. Multiple intents can motivate the same commitment, and a single intent can motivate commitments across organizational boundaries.

Work is the executable unit with defined inputs, process, and outputs. It is the locus of transformation — where capacity is consumed and evidence is produced.

Semantic boundary: Work is execution; Decision is selection. Work produces evidence; Decision evaluates it. Every work item must trace to a commitment (B1), ensuring no shadow work exists outside the governance perimeter.

Capacity represents the verified resources — time, skill, budget, authority scope — available to execute work. It answers the feasibility question that commitment alone cannot.

Semantic boundary: Capacity is the ability to act; Authority is the right to act. An actor can have capacity (skills, time) without authority (permission), and vice versa. Capacity is also distinct from Work: capacity is what makes work feasible, not the work itself.