artifacts/standard-named

Branch Recommendation

artifacts/standard-named/20260715__TELIC-FIELDS__BRANCH-RECOMMENDATION__WORKING__HI-0__controlled-branch-opening.md

Rendered from markdown source. Open raw source on GitHub.

Branch Recommendation

Status: bridge-pass recommendation Content canon status: unset

Decision

Begin H — Public Reader Path and I — Technical Specification Branch through a controlled parallel opening.

Do not begin them at equal depth.

The technical branch should lead by one small step.

The public branch should begin as reader architecture, not final publication.

The recommended next phase is:

HI-0 — Controlled Branch Opening

I.0 — Canonical Semantic Spine and Core Envelope Specification

Create the smallest implementable specification for:

  1. CenterReference;
  2. common Telic Record Envelope;
  3. Telic Projection Record;
  4. Semantic Trail / Provenance Event;
  5. Polytelometric Navigation Record;
  6. Model Role and Authority Envelope;
  7. Decision Witness;
  8. Contest and Correction Event.

Do not implement:

  • the complete training-lineage architecture;
  • universal ontologies;
  • automatic protected-condition classification;
  • universal scoring;
  • production model execution.

I.0 should prove type separation and lifecycle.

H.0 — Public Reader Architecture

In parallel, design the public path around ordinary questions:

  1. What matters?
  2. Who bears the consequence?
  3. What was actually said?
  4. What did the system infer?
  5. Who may act?
  6. What cannot be traded here?
  7. What happens if the context is incomplete?
  8. How can the decision be corrected or ended?
  9. What changes when AI participates?
  10. Where did the AI's own field come from?

H.0 should produce:

  • reader sequence;
  • concise glossary;
  • diagrams;
  • examples;
  • non-claims page;
  • relation among papers;
  • navigation between personal, relational, institutional, and AI scales.

It should not yet compress every F paper into final public prose.

---

Why the technical branch leads

The series now has strong conceptual coherence and extensive candidate schemas.

Its remaining danger is accidental ontology.

Terms such as:

  • standing;
  • protected condition;
  • authority;
  • consent;
  • semantic field;
  • model constitution;

can sound clearer than their implementation actually is.

A small technical spine forces the project to answer:

  • which distinctions are real types;
  • which are annotations;
  • which are policies;
  • which are events;
  • which are domain-specific;
  • which must remain unresolved.

The public path will be more trustworthy after those boundaries survive worked examples.

---

Why the public branch begins immediately

A specification developed without public-reader pressure will overgrow.

H.0 provides an independent test:

Can the framework explain a consequential decision without requiring the reader to learn the whole ontology?

The public branch should pressure I.0 toward:

  • simpler names;
  • fewer mandatory fields;
  • better examples;
  • visible non-claims;
  • progressive disclosure;
  • ordinary language.

The relationship is reciprocal.

I.0 prevents H.0 from making claims the architecture cannot support.

H.0 prevents I.0 from mistaking formal completeness for human usability.

---

Gate before deeper branching

Do not move to H.1 or I.1 until three worked examples pass the bridge.

Example A — Personal reversible decision

A person considers a job, move, treatment, or commitment.

Test:

  • minimum necessary projection;
  • internal plurality without clinical reification;
  • future standing;
  • protected conditions;
  • reversible route;
  • no-decision state.

Example B — Dyadic mediated disagreement

Two participants and a witness or model mediator.

Test:

  • private source channels;
  • shared projections;
  • receiver mirrors;
  • consent;
  • role symmetry and asymmetry;
  • dissent;
  • repair or clean exit.

Example C — Model-mediated institutional action

A model retrieves evidence, structures alternatives, recommends a route, and may call a tool.

Test:

  • source/model/provider separation;
  • constitutional retrieval;
  • role and credential envelope;
  • authority gate;
  • semantic trajectory;
  • human re-entry;
  • contest and recourse.

A training-lineage reference should be attached to Example C without implementing the full F.11 schema.

---

Gate criteria

The examples pass only if:

  1. source statements and model inferences remain distinguishable;
  2. missing standing can stop or narrow action;
  3. protected conditions cannot silently become low-weight preferences;
  4. consent and other authority bases remain distinct;
  5. the record supports correction without falsifying history;
  6. the human can understand and alter the route;
  7. private context can remain protected;
  8. model role cannot silently expand;
  9. no universal score is required;
  10. the record remains small enough for proportional use.

---

Recommended branch order after HI-0

I.1 — Provenance, Event, and Correction Specification

  • W3C PROV profile;
  • bitemporal event model;
  • correction propagation;
  • protected omission;
  • selective disclosure;
  • portable identifiers.

H.1 — Public Primer

Working title:

What Matters Before the System Acts

The primer should explain:

  • telic field;
  • projection;
  • standing;
  • consent and authority;
  • witness;
  • correction;
  • non-sovereign AI.

I.2 — Navigation and Gate Specification

  • field classes;
  • routes;
  • cost bearers;
  • gates;
  • no-decision;
  • analysis adapters.

H.2 — Relational and Institutional Reader Path

  • dyad;
  • loops;
  • context capacity;
  • time;
  • dependence;
  • dissolution.

I.3 — Model Session and Training-Lineage Profiles

  • model roles;
  • provider field;
  • retrieval;
  • semantic trajectory;
  • recourse;
  • training authority;
  • withdrawal;
  • succession.

H.3 — AI and Training Reader Path

  • what the model sees;
  • what it infers;
  • what it may do;
  • where its knowledge came from;
  • how to contest or leave.

---

What should not begin yet

Do not begin:

  • a universal Telic Field ontology;
  • a production platform;
  • legal compliance claims;
  • clinical deployment;
  • broad civic automation;
  • a certification badge for consentful training;
  • a patent or standards claim based solely on candidate terminology;
  • publication of the papers as settled science.

The next branch is specification and explanation.

It is not deployment.

---

Operational dependency

D and E remain unconfirmed in the current project record.

Repository installation and validation should be confirmed before the new H/I artifacts are treated as integrated wiki state.

This does not block HI-0 drafting.

It does block claims that the repository already contains the frozen bridge architecture.

---

Final recommendation

Proceed with controlled parallel work:

I.0 leads:
canonical semantic spine
core envelope
worked examples

H.0 follows in parallel:
reader map
glossary
examples
non-claims

The first deliverable should be one combined HI-0 package so the public and technical interpretations remain phase-locked.