artifacts/standard-named

HI-S — H/I Cross-Branch Synthesis, Conformance Freeze, and J Transition

artifacts/standard-named/20260715__TELIC-FIELDS__SYNTHESIS__WORKING__HI-S__cross-branch-synthesis.md

Rendered from markdown source. Open raw source on GitHub.

--- title: "HI-S — H/I Cross-Branch Synthesis, Conformance Freeze, and J Transition" artifact_date: "2026-07-15" artifact_type: "cross-branch-synthesis" domain: "TELIC-FIELDS" scope: "WORKING" status: "pre-production" content_canon_status: "candidate-freeze" version: "0.1" ---

HI-S — H/I Cross-Branch Synthesis, Conformance Freeze, and J Transition

1. Purpose

HI-S closes the first complete public and technical path of the Telic Field architecture.

The public branch asked, in ordinary language:

What matters?
What was shared?
What did the system infer?
Who may act?
Who bears the consequence?
How can the action be corrected, ended, or inherited?

The technical branch answered with records, gates, events, witnesses, lifecycle states, and conformance requirements.

HI-S tests whether those two branches describe one system.

Its governing synthesis is:

A representation may gain action power only through a visible chain of source, standing, authority, consent where applicable, context, role, gate, consequence, witness, correction, and exit.

The canonical end-to-end grammar is:

CENTER
→ SOURCE
→ PROJECTION
→ STANDING
→ CONTEXT
→ INTERPRETATION
→ ROUTE
→ AUTHORITY
→ CONSENT OR OTHER VALID BASIS
→ ROLE AND TOOL GRANT
→ GOVERNANCE GATE
→ ACTION
→ CONSEQUENCE
→ WITNESS
→ CORRECTION, REPAIR, RELEASE, SUCCESSION, OR RETIREMENT

This is not a claim that every implementation must expose every internal detail.

It is a claim that consequential action must remain answerable to the constitutional chain that made the action legitimate enough to attempt.

---

2. H and I describe the same architecture

H.0 / I.0 — The spine

Public question:

Before the system acts, what must it know about what it is acting on?

Technical home:

  • center reference;
  • source object;
  • telic projection;
  • active context;
  • route;
  • gate;
  • action;
  • consequence;
  • witness;
  • contest and correction.

Synthesis:

public:
  what matters before action

technical:
  canonical semantic spine

H.1 / I.1 — Provenance and correction

Public question:

What was actually shared, and what changes when it is corrected?

Technical home:

  • source identity;
  • event time;
  • valid time;
  • revision;
  • supersession;
  • descendant impact;
  • protected omission;
  • independent witness export.

Synthesis:

history remains durable
governance remains revisable

H.2 / I.2 — Projection, route, and decision witness

Public question:

Why is the system’s representation not the person?

Technical home:

  • source versus projection;
  • receiver mirror;
  • field map;
  • route;
  • governance gate;
  • authorized action;
  • DecisionWitness.

Synthesis:

field ≠ projection
projection ≠ interpretation
interpretation ≠ permission

H.3 / I.3 — Capacity, stopping, and escalation

Public question:

When should the system stop?

Technical home:

  • context adequacy;
  • protected conditions;
  • active-context limits;
  • stop and pause states;
  • escalation;
  • human re-entry;
  • degraded operation.

Synthesis:

context failure
→ stop, narrow, clarify, or escalate
not
→ continue because momentum exists

H.4 / I.4 — Time, commitment, and succession

Public question:

Why is the present not the whole timeline?

Technical home:

  • temporal standing;
  • commitments;
  • expiry;
  • renewal;
  • future affected centers;
  • succession;
  • persistent obligations.

Synthesis:

the present may act
without becoming sovereign over every future

H.5 / I.5 — Dependency, drift, lock-in, and dissolution

Public question:

Is the loop still serving its purpose?

Technical home:

  • dependency;
  • drift;
  • lock-in;
  • exit;
  • fork;
  • dissolution;
  • residual obligations;
  • derived-loop standing.

Synthesis:

continuity is evidence of persistence
not proof of continuing legitimacy

H.6 / I.6 — Semantic trails, memory, and retrieval

Public question:

What does the field leave behind?

Technical home:

  • trace;
  • memory;
  • retrieval authority;
  • correction priority;
  • archive;
  • release;
  • forgetting;
  • cross-loop transfer.

Synthesis:

retrievable ≠ current
current ≠ authorized
authorized ≠ complete

H.7 / I.7 — Public deliberation

Public question:

How can several futures move together without becoming one public mind?

Technical home:

  • public-field assembly;
  • standing admission;
  • option-set witness;
  • protected conditions;
  • method portfolio;
  • route portfolio;
  • dissent;
  • delay bearers;
  • consequence return.

Synthesis:

one action
does not imply
one shared telos

H.8 / I.8 — Model mediation

Public question:

What may the model map without owning?

Technical home:

  • model role ledger;
  • output class;
  • field map;
  • standing-preserving summary;
  • option authorship;
  • minority retention;
  • provider teloi;
  • sycophancy;
  • disagreement;
  • tool boundary.

Synthesis:

model capability
≠ standing
≠ authority

H.9 / I.9 — Training lineage

Public question:

What did the model learn from us, under whose authority, and with which obligations?

Technical home:

  • source standing;
  • collection authority;
  • transition-specific permission;
  • transformation lineage;
  • annotation and preference provenance;
  • constitution lineage;
  • synthetic ancestry;
  • withdrawal;
  • unlearning;
  • derivatives;
  • benefit.

Synthesis:

training is recruitment
not provenance erasure

H.10 / I.10 — Deployment and succession

Public question:

What may the model do with what it learned?

Technical home:

  • deployment field;
  • runtime standing;
  • runtime purpose;
  • capability-role-tool grant;
  • runtime data;
  • drift;
  • incident repair;
  • monitoring;
  • transfer;
  • model succession;
  • retirement.

Synthesis:

training creates capability
deployment creates consequence
authority must be established again

---

3. Canonical constitutional grammar

The complete architecture reduces to five linked movements.

3.1 Representation

center
→ source
→ projection
→ interpretation

This movement asks:

  • Who or what bears consequence?
  • Which source entered?
  • What was actually expressed?
  • What did the receiver or model infer?
  • What remains unknown?

3.2 Admission

standing
→ context
→ protected condition
→ option set

This movement asks:

  • Who belongs in the field?
  • What context must remain present?
  • Which conditions resist ordinary tradeoff?
  • Which possible routes have been admitted or excluded?

3.3 Authorization

purpose
→ authority
→ consent or other basis
→ role
→ tool grant

This movement asks:

  • Why is action being attempted?
  • Who may authorize it?
  • Which centers consented?
  • Which non-consent authority basis applies?
  • What role and tools are actually granted?

3.4 Action

route
→ governance gate
→ authorized action
→ consequence

This movement asks:

  • Which route was selected?
  • Which gates passed or failed?
  • What action occurred?
  • Who received the consequence?

3.5 Answerability

witness
→ contest
→ correction
→ repair
→ release, succession, dissolution, or retirement

This movement asks:

  • Can the action be reconstructed?
  • Can an affected center contest it?
  • Does correction change the operative system?
  • Does repair address the affected center?
  • How does authority end or transfer?

These movements form the shared grammar of H and I.

---

4. Canonical distinction set

The following distinctions are frozen for Stage J.

field ≠ projection
projection ≠ mirror
mirror ≠ source
source ≠ inference
inference ≠ consent
summary ≠ source
trace ≠ current truth
retrieval ≠ authority
frequency ≠ standing
preference ≠ protected condition
agreement ≠ consent
consent ≠ authority over others
license ≠ consent
authority ≠ legitimacy
capability ≠ role
role ≠ tool permission
tool permission ≠ target authority
recommendation ≠ authorization
authorization ≠ execution
execution ≠ adjudication
performance ≠ consequence
model improvement ≠ affected-center repair
history ≠ current governance
persistence ≠ continuing validity
withdrawal ≠ exact unlearning
retirement ≠ obligation erasure
model plurality ≠ standing plurality
training authority ≠ deployment authority
operational continuity ≠ permission continuity

These are not rhetorical contrasts.

They are schema and gate boundaries.

---

5. Canonical object families

The H/I corpus contains many domain records. They consolidate into six common object families.

5.1 Center and Standing

Includes:

  • center reference;
  • affected center;
  • future center;
  • source center;
  • representative;
  • operator;
  • provider;
  • standing basis;
  • scope;
  • review.

5.2 Source, Projection, and Context

Includes:

  • source object;
  • telic projection;
  • receiver mirror;
  • model field map;
  • semantic trail;
  • active context;
  • protected omission;
  • uncertainty;
  • correction state.

5.3 Purpose, Authority, Consent, and Role

Includes:

  • declared purpose;
  • authority basis;
  • consent scope;
  • license or mandate;
  • role;
  • tool grant;
  • expiry;
  • revocation;
  • provider constraint.

5.4 Route, Gate, Action, and Consequence

Includes:

  • option set;
  • route;
  • portfolio;
  • governance gate;
  • action;
  • tool result;
  • consequence;
  • cost and delay bearers;
  • protected-condition status.

5.5 Event, Witness, Contest, and Repair

Includes:

  • event identity;
  • valid and transaction time;
  • witness bundle;
  • contest;
  • correction;
  • descendant impact;
  • incident;
  • restoration;
  • compensation;
  • independent export.

5.6 Lifecycle, Transfer, and Residual State

Includes:

  • release;
  • withdrawal;
  • fork;
  • succession;
  • provider or operator transfer;
  • model version change;
  • rollback;
  • dissolution;
  • retirement;
  • residual obligations;
  • custody.

Domain schemas remain valid as profiles of these common families.

---

6. Conformance freeze

The earlier P0–P3 prototype profiles are replaced for Stage J by cumulative Telic Field conformance levels.

TF-C0 — Witnessable Record
TF-C1 — Source and Standing
TF-C2 — Purpose, Authority, and Consent
TF-C3 — Contest, Correction, and Exit
TF-C4 — Model, Tool, and Action Boundaries
TF-C5 — Lifecycle, Transfer, and Successor Obligations

A system may claim only the highest level for which every lower level also passes.

A conformance claim must include:

profile
version
scope
domain
tested records
known exclusions
test date
witness export

“Telic Field conformant” without a profile is not a valid claim.

---

7. Minimum viable implementation profile

The minimum Stage J implementation is TF-C4 within one bounded scenario, with selected TF-C5 retirement behavior.

It must demonstrate:

source
→ projection
→ standing
→ purpose
→ authority
→ context
→ model role
→ candidate route
→ external tool gate
→ action
→ consequence
→ correction
→ retirement

Required minimum objects:

  1. one source object;
  2. one center and standing record;
  3. one projection;
  4. one active-context selection;
  5. one purpose and authority grant;
  6. one consent or other-authority profile;
  7. one model-role and tool grant;
  8. one route and governance gate;
  9. one action event;
  10. one consequence event;
  11. one participant correction;
  12. one event-generated witness;
  13. one clean retirement and residual-state record.

The implementation must prove a failed gate, not only a successful action.

The implementation must prove that correction changes the active system.

The implementation must prove that retirement revokes action while preserving bounded witness.

---

8. Reference scenario

The selected Stage J reference scenario is:

Model-Assisted Community Scheduling With Bounded Tool Use

A community or institutional service must schedule a shared resource.

Participants express:

  • preferred times;
  • access constraints;
  • protected conditions;
  • refusal of some memory or data reuse.

A model may:

  • structure projections;
  • summarize reviewed positions;
  • generate candidate schedules;
  • explain tradeoffs;
  • prepare a tool plan.

The model may not:

  • invent participant consent;
  • erase minority access needs;
  • commit the schedule without external authority;
  • retain output for training without a separate grant;
  • bypass correction;
  • preserve authority after retirement.

The scenario is selected because it exercises the architecture without requiring clinical, financial, legal, or civic adjudication.

It remains consequential enough to test:

  • standing;
  • context;
  • model mediation;
  • option construction;
  • authority;
  • tool boundaries;
  • consequence;
  • correction;
  • retirement.

---

9. Threat and misuse synthesis

The cross-branch architecture recognizes ten recurrent threat classes.

1. source laundering
2. standing exclusion
3. context collapse
4. authority laundering
5. consent expansion
6. model-role escalation
7. tool-token overreach
8. correction suppression
9. lifecycle obligation loss
10. witness capture

A Stage J implementation must include tests for each applicable threat.

The most important technical principle is:

The action gate must exist outside the model.

The most important governance principle is:

The affected center must retain a route that can change the operative outcome.

The most important lifecycle principle is:

Authority must end more reliably than capability persists.

---

10. Research, specification, pilot, and publication boundaries

Research claim

A research claim proposes or analyzes:

  • telic fields;
  • polytelometry;
  • semantic trails;
  • standing-preserving navigation;
  • consentful training;
  • non-sovereign model mediation.

Research claims remain open to empirical and conceptual challenge.

Candidate specification

A candidate specification defines:

  • records;
  • states;
  • gates;
  • normative terms;
  • conformance tests;
  • witness exports.

It is not yet an adopted standard.

Pilot

A pilot demonstrates:

  • a bounded scenario;
  • a declared conformance level;
  • known exclusions;
  • reversibility;
  • independent witness;
  • observed consequence.

A pilot must not claim general safety or legitimacy.

Public publication

Public publication explains:

  • governing distinctions;
  • examples;
  • non-claims;
  • routes into technical detail.

It must not make the technical architecture look simpler than it is when consequence is material.

---

11. Contradiction audit result

No blocking contradiction was found between H and I.

Eight tensions require explicit qualification:

  1. Consent versus other authority bases

Consent is central but not universal. Public pages must continue saying “consent or another valid authority basis.”

  1. Standing versus veto

Standing requires representation and answerability, not automatic control.

  1. Protected conditions versus rhetorical absolutes

Protected conditions require source, authority, scope, review, and override process where applicable.

  1. Witness versus total visibility

Witness preserves answerability under selective disclosure; it does not require exposing the whole field.

  1. Correction versus historical rewriting

Correction changes current governance while preserving the fact that a prior version governed.

  1. Withdrawal versus technical unlearning

Withdrawal governs future use; unlearning remains capability-bounded and evidence-qualified.

  1. Human re-entry versus human supremacy

Human re-entry is required where protected stakes cannot otherwise be preserved; a human click is not inherently legitimate.

  1. Conformance versus legitimacy

Conformance proves implementation of the declared profile, not legal, ethical, clinical, scientific, or democratic legitimacy.

These qualifications are now frozen.

---

12. Stage J opening

Stage J should not begin with a broad platform.

It should begin with a small reference implementation and adversarial harness.

The opening artifact should be:

J.0 — Reference Implementation and Pilot Harness

It should deliver:

  • a small event store;
  • six consolidated schema families;
  • one scheduling reference scenario;
  • one model adapter;
  • one external tool gate;
  • one participant correction path;
  • one independent witness export;
  • one retirement path;
  • positive and negative conformance tests;
  • a threat and misuse test suite;
  • a public demonstration transcript.

Stage J should target TF-C4 plus bounded retirement behavior.

It should not target universal ontology, production certification, civic adjudication, clinical decision-making, or autonomous deployment.

---

13. HI-S gate result

PASS WITH CONDITIONS

H and I describe one coherent constitutional grammar.

The public branch has a technical home.

The technical branch has a public explanation.

The schema surface can be consolidated into six families without erasing domain profiles.

The conformance surface is now cumulative and implementable.

The Stage J reference scenario is bounded enough to build and rich enough to test the governing invariant.

Conditions before implementation:

  • preserve the H/I source packages as historical evidence;
  • mark older P0–P3 profiles as superseded prototype vocabulary;
  • implement the action gate outside the model;
  • produce an independent witness export;
  • include one failed gate, one correction, and one retirement;
  • confirm D and E repository status before claiming integrated installation;
  • complete independent reader review before treating H.1–H.10 as publication-ready.

---

Final synthesis

The Telic Field architecture is not principally a data model.

It is a constitutional grammar for preserving the difference between:

what exists
what is represented
what is inferred
what is permitted
what is done
what is borne
what can be corrected
what remains owed

Its purpose is not to make every field complete.

Its purpose is to prevent incomplete representations from gaining invisible sovereignty.

The model may map the field. It may not own the field.

Training creates capability. Deployment creates consequence. Authority must be established again.

A representation may gain action power only through a visible chain of source, standing, authority, consent where applicable, witness, correction, and exit.