TriadicFrameworks
Overview

Minimal Observer API

The Triadic Observer Layer is intentionally lightweight. Its API is designed to be easy to emit, hard to misuse, and domain‑agnostic by default. Existing systems should be able to participate without architectural change or political risk.

The observer consumes observations. It does not request them.

Design Goals#

  • Minimal surface area.
  • Explicit phase declaration.
  • First‑class time and source.
  • No implied authority.
  • Safe under partial adoption.

The API favors clarity over completeness.

Core Observation Schema#

Every observation submitted to the observer must conform to the following structure:

{
  "domain": "string",
  "entity_id": "string",
  "phase": "string",
  "metric": "string",
  "value": "number|string",
  "unit": "string",
  "source": "string",
  "timestamp": "ISO-8601",
  "confidence": "string",
  "notes": "string (optional)"
}

Field Semantics#

  • domain — The operational context (e.g., elections, supply_chain, science).
  • entity_id — The smallest independently observable unit.
  • phase — Declared lifecycle stage of the observation.
  • metric — What is being measured.
  • value — The reported measurement.
  • unit — Measurement unit or classification.
  • source — System or process emitting the observation.
  • timestamp — When this observation existed in its reported form.
  • confidence — Declarative certainty level (e.g., provisional, final).
  • notes — Optional human context; never required.

No field implies correctness. All fields imply context.

Phase Declaration Rules#

  • Phase must be explicit.
  • Phase must not be inferred by the observer.
  • Multiple phases may exist simultaneously for the same entity and metric.
  • Phase transitions are observed, not enforced.

Phase disagreement is preserved as signal.

Time Semantics#

The timestamp represents existence, not ingestion.

  • Late submissions are valid.
  • Out‑of‑order events are valid.
  • Corrections are new observations, not overwrites.

Temporal inconsistency is informational.

Source Semantics#

Sources are identifiers, not authorities.

  • Multiple sources may emit the same metric.
  • Conflicting sources are expected.
  • Source trust is external to the observer.

The observer never ranks sources.

Confidence Field#

Confidence is declarative, not probabilistic.

Examples:

  • provisional
  • estimated
  • audited
  • certified
  • archived

Confidence does not override phase or source.

Domain Specialization#

Domains may extend the schema with additional fields, but must not:

  • Remove core fields.
  • Collapse axes.
  • Imply authority.

Extensions must remain observational.

Example: Elections#

{
  "domain": "elections",
  "entity_id": "MI-Wayne-P042",
  "phase": "counted",
  "metric": "ballots_cast",
  "value": 1832,
  "unit": "count",
  "source": "county_tabulator_v3",
  "timestamp": "2026-11-03T21:14:00Z",
  "confidence": "provisional",
  "notes": "late upload due to network outage"
}

Example: Supply Chain#

{
  "domain": "supply_chain",
  "entity_id": "DC-ATL-07",
  "phase": "in_transit",
  "metric": "units_shipped",
  "value": 4200,
  "unit": "items",
  "source": "logistics_system_A",
  "timestamp": "2026-04-12T09:30:00Z",
  "confidence": "reported"
}

Observer Guarantees#

The observer guarantees that:

  • All observations are preserved.
  • No observation is modified.
  • No synthesis implies correctness.
  • All outputs remain replayable.

The observer makes structure visible, not truth final.

This minimal API is sufficient to support triadic coherence analysis across domains without introducing control, prediction, or enforcement.

The simplicity is intentional. Complexity belongs in interpretation, not emission.

Powered by Docsbook