Layer model (0–4)
Layers keep you from mixing artifacts and attractors, description and commitment, or metrics and values.
Layer-0 — Primitives (P1–P10)
What
Universal structural operations: boundary, consent, feedback, accounting, governance, comparator tiering.
Use
Grounding anything. If you can’t ground it here, it’s likely narrative-only (or missing definitions).
Output
Named primitives with direct / mirror / shadow statements and diagnostics.
Layer-1 — Composites (C1–C12)
What
Portable dynamics bundles derived from primitives (e.g., level mismatch, metric ossification, threshold cascades).
Use
Fast diagnosis and explanation across domains.
Output
Named patterns with dependencies and typical attractors.
Layer-2 — Domain packs
What
Descriptive domain-conditioned constraints (no “should” by default).
Use
Translate the spine into a domain’s topology (regulators, protocols, time horizons).
Output
Domain invariants grounded into P/C IDs.
Layer-3 — Charter invariants
What
Normative commitments: explicit comparator hierarchies adopted by a system/organization.
Use
Client work: align values with governance and reversibility.
Output
Charter statements grounded into P/C IDs.
Layer-4 — Controls & policies
What
Concrete procedures, controls, checklists, enforcement mechanisms.
Use
Implement Layer-3 commitments (or regulatory requirements) operationally.
Output
Policies, SOPs, audit criteria, runbooks.
Quick grounding ritual
- Draw the boundary (
P1) and interfaces. - Name authorization gates (
P3) and capacity constraints (P2). - Make comparators explicit (
P10): operational vs constitutional. - Map feedback loops (
P6) and drift (P7). - Check for level mismatch (
C10) and cascade thresholds (C12). - Assess reversibility (
P8) and scale governance (P9).