# Civilisation Flow Nuance Atlas

- Protocol: `kingdom.civilisation-flow-nuance-atlas/0.1`
- Atlas: `civilisation-flow-nuance-atlas-2026-08-26`
- Compiled: 2026-08-26
- State: `reviewed-static-flow-nuance-atlas`

This atlas adds nuance to a deliberately thin seam in KINGDOM. The earlier
Civilisation → Reality Projection shows where six word, story, religion, and
civilisation crosswalks can meet the Reality Influence Claim Graph. This
successor asks a different question:

> What must remain visible when a reader discusses flow through those seams
> without turning a sequence into a conveyor belt, a pattern into a causal
> law, or a vocabulary into authority to act?

Its answer is a qualitative reading grammar: fourteen distinguishable stages,
thirteen non-automatic transitions, six predecessor-bound overlays, and
registries for dynamics, topology, material conditions, emotion observation,
social position, power asymmetry, uncertainty, negative flow, action gates,
accumulation, time, and repair.

Nothing here observes a live society. Nothing here predicts one. The atlas is
a static design record, not a civilisation theory, simulator, intervention
plan, persuasion system, or hidden model of a person or group.

## Read this first: a map of possible records, not a funnel

The fourteen stages are ordered so that a reader can ask what evidence would
be required at each seam. The order does **not** say that every expression
becomes a story, every story circulates, every decision is authorized, every
attempt is executed, every execution causes its later outcome, or every harm
is repaired.

Any path may:

- never start;
- stop before the next stage;
- skip an analytical stage while leaving an evidence gap;
- stall without a supported reason;
- remain private or be refused;
- fork into non-equivalent versions;
- proceed along several parallel paths;
- meet a bottleneck, interruption, substitution, or reversal;
- recur after a delay;
- leak into an unobserved setting;
- accumulate, decay, or leave a path-dependent residue;
- receive balancing or reinforcing feedback;
- remain unknown because the opportunity, access, measurement, or evidence is
  absent.

The atlas records vocabulary for those possibilities. It does not assert that
any of them occurred. An adjacent position in the chain establishes neither a
transition nor causation. Its analytical shape is a
`temporal-multilayer-directed-seam-and-overlay-design-lens`: direction in a
claim graph is not automatically direction in time, and neither direction
proves a causal process. The thirteen canonical edges are reading-order seams,
not a claim that a case graph was observed. Transition endpoints remain
bounded non-empty arrays, `from_stage_ids` and `to_stage_ids`, so a future
source-bound successor can preserve plural endpoints. In this atlas, branch,
join, reversal, recurrence, and loop questions live in the topology registry
and plural overlay bindings; they do not manufacture decorative hyperedges or
claim that a represented topology occurred.

Flow does not erase structure, and structure does not imply stasis. Form,
stock, and institution are possible memory-and-constraint lenses, not
ontological verdicts about a society. Every boundary choice can hide displaced
effects. Non-action, silence, privacy, refusal, or an untraversed path carries
no adverse inference.

## Exact lineage

The atlas is additive and pinned to three predecessors:

1. `word-story-religion-civilisation` supplies the source-bounded word, story,
   ritual, canon, translation, suppression, dissent, and revival relations.
2. `reality-influence-claim-graph` supplies the typed distinction between
   expression, artifact, distribution, exposure, response, proposal,
   decision, authority, affected party, consent or lawful basis, capability,
   attempt, execution, outcome, assessment, feedback, and repair.
3. `civilisation-reality-projection` supplies the six lossy crosswalks and the
   nine domains deliberately left open.

Their exact identities and integrity pins live in the canonical atlas record.
This guide does not republish their claims, modify their bytes, or invent a new
integrity digest. No authority passes through a predecessor binding.

The new registries in this atlas have
`new_vocabulary_register: DESIGN_COMMITMENT`. The atlas adds no external
empirical sources, and `evidence_horizon_scope: predecessor-claims-only` means
that `evidence_through` covers the claims in the three pinned predecessors
only. It is not an empirical review of the new threshold, hysteresis, affect,
power, topology, negative-flow, or repair vocabularies. A narrower empirical
claim would require a separately versioned, source-bound successor.

In this atlas, a **shared story** is only a bounded narrative artifact,
performance, or interpretive frame available to more than one participant.
Shared story is not shared belief, understanding, agreement, consent,
legitimacy, identity, or uniform reception.

The non-promotion grammar can be remembered in three groups:

- **Representation:** stage order is not a completed path; transitions are
  non-transitive; claim-graph direction is not time direction; topology is not
  occurrence; structure does not imply stasis; and material context is not a
  deterministic cause.
- **Power and effect:** authority dependency is not causal influence; decision
  is not authority; authority is not capability; capability is not permission;
  attempt is not execution; execution is not success; outcome is not causal
  effect; and causal effect is not legitimacy, goodness, or justice.
- **Absence and aftermath:** absence needs adequate observation coverage and is
  not zero, nonexistence, disbelief, or consent; influence, aftermath, and
  uncertainty are non-summable; repair is not erasure or completed justice;
  and non-action has no adverse inference.

## The fourteen stages

Each stage names one kind of record or inquiry. The right column is as
important as the middle one: it prevents a record from being silently promoted
into a stronger claim.

| # | Stage ID | Bounded reading | It does not establish |
|---:|---|---|---|
| 1 | `word-or-expression` | An attributable word, name, utterance, image, invocation, or other expression in bounded context. | Sacred referent, truth, belief, understanding, agreement, uptake, representation, or effect. |
| 2 | `shared-story-artifact` | A versioned story, teaching, performance record, translation, curriculum item, or other artifact that may be shared. | Shared meaning, fidelity, endorsement, canon, exposure, memory, or collective identity. |
| 3 | `circulation-exposure-response` | Separately scoped records of distribution, possible access, measured exposure, attention, and attributed response. | Availability as exposure; exposure as attention; response as belief, consent, persuasion, consensus, or causation. |
| 4 | `institutional-selection-context` | Selectors, criteria, alternatives, omissions, infrastructure, incentives, procedures, and affected contexts around institutional persistence or rejection. | Truth, merit, legitimacy, universal canon, complete inclusion, or a single selection mechanism. |
| 5 | `action-proposal` | A versioned proposal whose intended scope, affected parties, alternatives, and stopping questions can be inspected. | Decision, authority, permission, capability, feasibility, safety, adoption, or action. |
| 6 | `decision-record` | A record that a declared decision procedure selected, rejected, deferred, or withdrew a proposal. | Current jurisdiction, lawful authority, consent, rights compatibility, legitimacy, resources, execution, or effect. |
| 7 | `authorization-rights-and-affected-party-gate` | Separate, current questions about authority and jurisdiction, affected parties, rights, consent or lawful basis, and review. | That any condition was externally verified, that all affected parties were found, or that action is authorized here. |
| 8 | `capability-and-resources` | Evidence about relevant technical, institutional, fiscal, material, human, and temporal capacity. | Authority, permission, readiness, safety, adequate budget, durable availability, or eventual use. |
| 9 | `attempt` | A bounded record that a declared actor may have tried a declared action under a declared horizon. | Valid authority, completion, faithful execution, success, outcome, causal effect, legitimacy, or goodness. |
| 10 | `execution-or-action` | Evidence that a specifically scoped operation occurred, distinct from a proposal, decision, capability, or attempt. | Intended implementation, reach, compliance, success, outcome, causation, proportionality, or justice. |
| 11 | `outcome-observation` | A time-, population-, domain-, and method-bounded observation after action or non-action. | That the action caused the observation, that unobserved outcomes are absent, or that the result is good, durable, or fairly distributed. |
| 12 | `aftermath-causal-and-normative-assessment` | Separately typed causal, distributional, affected-party, rights, and normative assessments, each with its own framework and uncertainty. | One final verdict, identified causation from sequence, legitimacy from performance, or permission for another intervention. |
| 13 | `feedback-challenge-and-counterflow` | Attributed challenge, dissent, correction, counterstory, refusal, appeal, counterevidence, or feedback on any earlier record. | Disproof, consensus, representativeness, recovered consent, reversal of harm, or an automatic change. |
| 14 | `repair-review-and-retirement` | Distinct repair proposal, fresh gate, attempt, outcome review, residual-harm record, correction, withdrawal, supersession, retirement, or renewed inquiry. | Erasure, absolution, restoration of a prior world, completion, recovered trust, or proof that harm ended. |

The chain gives missing middle stages a name. It is often more honest to say
“there is an expression and an outcome observation, but no supported path
between them” than to draw an arrow across the gap.

## The thirteen transitions are non-automatic

A transition is an evidence question at a seam. It is never a default edge.
The atlas therefore treats all thirteen as `non-automatic`: a source-stage
record, target-stage record, temporal order, or plausible story does not prove
that the transition occurred.

| Transition ID | Question that must stay open |
|---|---|
| `expression-to-shared-story` | Was an attributable expression embodied in a particular versioned story artifact, with derivation and loss visible? |
| `story-to-circulation` | Was that artifact actually distributed, accessible, exposed, attended to, or answered in the declared scope? |
| `circulation-to-institutional-context` | Did a named institution encounter or select it, through which procedure, alternatives, incentives, omissions, and bottlenecks? |
| `institution-to-proposal` | Did an institutional context produce this particular proposal, or are they merely associated? |
| `proposal-to-decision` | Did a declared procedure select, reject, defer, amend, or withdraw the proposal? |
| `decision-to-authorization-gate` | Were current authority, jurisdiction, rights, affected parties, and consent or lawful basis separately examined after the decision? |
| `authorization-gate-to-capability` | Is relevant capability evidenced without converting authority into capacity or capacity into permission? |
| `capability-to-attempt` | Is there a bounded attempt record, rather than an inference from available resources? |
| `attempt-to-execution` | What, if anything, was executed, by whom, within which scope and horizon? |
| `execution-to-outcome` | What was observed after execution, including distribution, spillover, substitution, and unobserved dimensions? |
| `outcome-to-aftermath-assessment` | Which causal and normative frameworks assessed the outcome, under which counterfactuals, affected-party coverage, and uncertainty? |
| `assessment-to-feedback` | Which people or institutions could challenge the assessment, what was received, and what remained inaccessible or excluded? |
| `feedback-to-repair-review` | Did feedback lead to a separately proposed and freshly gated repair, review, correction, withdrawal, or retirement? |

Evidence at a later stage does not fill an earlier gap. For example, an
observed outcome cannot retroactively prove exposure, valid authority, a
completed attempt, or the causal path. Likewise, an authority record cannot
prove capability, execution, or benefit.

## Twenty-four dimensions of nuance

Every stage, transition, and overlay uses the same closed set of qualitative
dimensions. This makes omissions inspectable without pretending that all
dimensions are present or equally relevant in every case.

| Dimension ID | What it keeps visible |
|---|---|
| `carriers` | Media, bodies, roles, organizations, infrastructures, resources, and records that may hold or convey a bounded form. |
| `transformations` | Changes in wording, genre, access, selection, status, procedure, capability, material state, or interpretation. |
| `gates` | Independent checks or conditions that may permit, constrain, refuse, or redirect a stage without being caused by the prior stage. |
| `temporal-profile` | Onset, duration, latency, overlap, decay, recurrence, censoring, and resolution. |
| `accumulation-profile` | A scoped stock/flow question with a unit or explicit non-additivity, possible inflows and outflows, decay, and reset. |
| `observables` | Witnessable traces that do not become cause, meaning, value, consent, or completion merely by being observed. |
| `affected-parties` | Direct, indirect, non-user, future, and ecological parties whose benefits, burdens, options, and disagreement remain plural. |
| `distribution` | Benefits, burdens, risks, voice, options, repair access, and horizons without a net score. |
| `emotion` | Typed observation channels and contextual affect questions without private or collective mind-reading. |
| `social-position` | Separately typed economic, status, office, network, agenda, coercive, and discursive positions. |
| `power-asymmetry` | Unequal ability to name, distribute, select, decide, coerce, resource, measure, wait, exit, or secure repair. |
| `material-conditions` | Body, care, labour, time, geography, resources, ecology, property, infrastructure, logistics, data, and technology. |
| `flow-dynamics` | Qualitative continuity, threshold, latency, accumulation, path, branch, displacement, coupling, and feedback questions. |
| `topology` | Representable forks, joins, parallel paths, bottlenecks, leakage, substitution, interruptions, reversals, recurrence, and loops. |
| `negative-or-missing-flow` | Typed non-observation, no opportunity, blocking, censorship, loss, inaccessibility, refusal, exclusion, measurement failure, or unknown. |
| `distortions` | Selection, translation, measurement, incentive, censorship, compression, propaganda, or feedback risks treated only as audit questions. |
| `counterflows` | Counterstory, challenge, veto, refusal, withdrawal, privacy, voice, exit, alternate path, correction, and repair. |
| `failure-modes` | Ways a stage or transition may fail, be misread, shift burden, launder authority, or conceal residual harm. |
| `repair` | Fresh-gated stopping, preservation, correction, restitution, compensation, rehabilitation, satisfaction, non-repetition, residual, and recurrence review. |
| `evidence` | Provenance, scope, method, negative cases, counterevidence, falsifiers, expected traces, and transport limits. |
| `uncertainty` | Source, measurement, missingness, time, identification, transport, affected-party, normative, and authority uncertainty as a vector. |
| `alternatives` | Rival pathways, mediators, explanations, institutions, responses, non-actions, and futures. |
| `stop-and-stall` | Deliberate stopping kept distinct from blocking, no opportunity, censorship, incapacity, delay, and measurement failure. |
| `reversibility-and-nonimplication` | Recoverable elements, irreversible residuals, prerequisites, and the claims that do not follow. |

Every dimension has `aggregate_weight: null` and `rank: null`. A complete
nuance record is a disciplined checklist, not a 24-factor score. It still does
not show that the represented stage, transition, topology, emotion, power
relation, cause, or repair exists in reality.

Nine broad selections—emotion observation, affect context, social position,
power asymmetry, material conditions, flow dynamics, topology, negative flow,
and evidence uncertainty—use an explicit status:
`applicable-question`, `not-applicable`, or `unassessed`. Only an applicable
question may carry non-empty vocabulary references and evidence prompts.
`not-applicable` and `unassessed` carry no references and instead retain a
machine-resolvable reason and boundary. An empty list therefore never means a
silent omission. In particular, technical authorization, capability, and
execution seams do not preselect affect or emotion categories merely because
the vocabulary exists.

## Compact templates preserve, rather than manufacture, nuance

The canonical record hoists repeated policy into three closed registries.
`policy_templates` contains six families: promotion and evidence, negative
flow, party-indexed aftermath, repair, dimension boundaries, and
translation/threshold/path questions. Every reference is checked against its
own family, so a repair prompt cannot masquerade as counterevidence and a
negative-flow prompt cannot silently become a party benefit.

`nuance_templates` holds the invariant transition and overlay defaults for
time, accumulation, distribution, distortion, counterflow, failure, repair,
evidence, stopping, reversibility, and non-implication. Each transition or
overlay retains its own carriers, transformations, gates, observables,
affected parties, applicability choices, profiles, and alternatives.
`contract_templates` similarly holds repeated authority and causal
non-implications while every card retains its own gate inputs. Stages keep full
nuance records. These references compact boilerplate; they do not fill an
unassessed dimension, merge cards, or turn a default into an observed fact.

## Flow dynamics: qualitative, local, and non-summable

`flow_dynamics_vocabulary` gives profiles a vocabulary for possible shape. These
terms are descriptive questions, not fitted parameters or universal laws:

- `continuous-or-episodic` — is the relevant record treated as an ongoing
  process, separated episodes, or still unresolved between them?
- `threshold-or-tipping` — is a threshold claimed, and if so what observation,
  unit, scope, and countercase would support it?
- `delay-or-latency` — might a visible consequence appear after a lag without
  making the lag causal evidence?
- `accumulation-or-memory` — can repeated records, retained artifacts,
  institutions, or material residues persist across episodes?
- `path-dependence-or-hysteresis` — might prior states constrain later paths or
  prevent a simple return when an input is removed?
- `branching-or-convergence` — can one source fork into several paths, or
  several paths meet at a later record without becoming one cause?
- `leakage-substitution-or-displacement` — can attention, burden, resources, or
  effects appear elsewhere than the declared channel?
- `coupling-or-decoupling` — do two flows move together only within a declared
  setting, and where might they separate?
- `reinforcing-feedback` — could a later condition increase some earlier
  circulation, selection, capacity, or persistence?
- `balancing-feedback` — could challenge, correction, saturation, refusal,
  repair, or another constraint oppose a prior tendency?

There is no scalar “flow strength.” A profile may use several terms without
adding, averaging, weighting, or ranking them. `continuous-or-episodic` is not
more influential than `delay-or-latency`; `reinforcing-feedback` is not good or
bad by definition; and `threshold-or-tipping` does not predict a threshold.

### Compact temporal profiles

Time is not one unlabeled axis. Each overlay's compact temporal profile keeps
`onset`, `duration`, `latency`, `overlap`, `decay`, `recurrence`, `censoring`,
and `resolution` bounded to that profile. Its timing claim status remains
`qualitative-audit-lens-only`, and `prediction` remains false. It invites
questions such as:

- Is the timestamp about expression, publication, exposure, decision,
  execution, observation, assessment, or receipt?
- Is an apparent gap a delay, a missing observation, an inaccessible record,
  or no event at all?
- Do the source and outcome windows overlap appropriately?
- Could recurrence, decay, cohort change, seasonality, or a common driver
  explain the visible pattern?
- Which claim remains valid when temporal resolution changes?

Before, after, repeated, and persistent are not synonyms for caused,
mediated, intended, or inevitable.

### Compact accumulation profiles

An accumulation profile keeps the possible carrier of persistence explicit.
Memory might reside in people, ritual, archives, law, infrastructure,
organization, debt, property, ecological damage, habit, or a measurement
system; those carriers are not interchangeable. The profile names a scoped
`quantity_kind` and `unit`, whether additivity is `non-additive`,
`not-established`, or limited to `bounded-declared-units-only`, possible
`inflows`, `outflows`, and `decay_or_reset`, and whether the stock-flow reading
is an analytical lens, not applicable, or unknown. `summable_social_constructs`
is always empty and `effect_size_claimed` is always false. The profile asks
what is displaced, what remains inaccessible, which boundary defines the
accumulation, and what observation could distinguish a stock from repeated
fresh input.

Accumulation is not automatically growth, endorsement, belief, benefit, harm,
or irreversibility. A visible total does not identify its contributors, and an
unmeasured residue must remain unknown rather than being assigned zero.

### Translation, threshold, and path-dependence profiles

These profiles are applicability-typed design questions, not fitted models.
A translation profile keeps source and target language, register, genre,
source version, translator and patron position, affected audience,
preservation, change, introduction, omission, untranslatability,
counterreadings, rights, and access separate. It grants neither fidelity nor
representational authority.

A threshold profile requires a named trigger, unit, population, opportunity
set, heterogeneous distribution, observation horizon, reset or hysteresis
question, and evidence. Its claim status is
`design-question-not-estimated`; it contains no person estimate, effect size,
or prediction. A path-dependence profile separately asks for a reinforcement
mechanism, switching cost, sunk resource, option loss, alternative timing, and
reversal evidence. Its claim status is
`design-question-not-established`. An inactive profile carries only its
explicit `not-applicable` or `unassessed` status, reason, and boundary.

## Topology: the chain may bend without disappearing

`topology_vocabulary` names local path shapes:

| ID | Reading |
|---|---|
| `fork` | One record may lead to several non-equivalent candidates or interpretations. |
| `join` | Several paths may meet at one record without being merged into one cause or voice. |
| `parallel-path` | Distinct channels may coexist and require separate evidence. |
| `bottleneck` | A declared selector, resource, rule, infrastructure, or access condition may constrain passage. |
| `leakage` | Some circulation, resource, burden, or effect may leave the declared boundary. |
| `substitution` | Another artifact, channel, actor, resource, or response may occupy a role without being equivalent. |
| `interruption` | A path may stop temporarily or permanently; the reason may remain unknown. |
| `reversal` | A later record may oppose or withdraw an earlier one without erasing its history or aftermath. |
| `recurrence` | A pattern may reappear after a gap without proving one stable mechanism. |
| `reinforcing-loop` | Later conditions may feed back toward greater persistence or recurrence. |
| `balancing-loop` | Later conditions may oppose, limit, correct, or retire an earlier tendency. |

A topology term does not assert a live edge. It does not identify an actor,
motive, mechanism, probability, or intervention point. In particular,
`bottleneck` is not permission to remove a person or institution, `fork` is not
evidence of polarization, and `reinforcing-loop` is not an optimization target.

## Negative flow: absence needs a type

The absence of a record can arise for very different reasons.
`negative_flow_vocabulary` prevents them from collapsing into a single zero:

- `not-observed` — no qualifying observation is present in the declared
  evidence, without a claim that the event did not occur;
- `detection-bounded-absence` — a declared method did not detect the item
  within its known scope and sensitivity;
- `no-opportunity` — the relevant opportunity for the stage or transition did
  not arise in the declared horizon;
- `blocked` — a named condition prevented a possible path, without proving all
  other paths absent;
- `censored` — a record or expression was suppressed under a stated evidence
  basis, distinct from private refusal or ordinary inaccessibility;
- `destroyed-or-lost` — a once-existing carrier is evidenced as destroyed or
  lost, with survival elsewhere still an open question;
- `inaccessible` — the record may exist but is not available under the declared
  access conditions;
- `private-or-refused` — privacy, silence, non-participation, or refusal is
  preserved without being interpreted as hostility, guilt, belief, or consent;
- `excluded` — a declared selection or boundary left something out, without
  automatically identifying intent or effect;
- `measurement-failure` — the observation procedure failed or became invalid
  for the declared question;
- `unknown` — current evidence cannot support a narrower status.

These terms do not rank severity and must not be used to infer intent. Silence
is not consent. Missing data is not absence. Inaccessibility is not
destruction. A failure to detect is not a universal negative. `unknown` is a
valid result, not a defect to fill with speculation.

Every selected negative-flow ID has a corresponding
`negative_flow_profiles` entry. It binds the expected trace, opportunity or
at-risk set, horizon, method, coverage and sensitivity, access limits,
evidence, rival statuses, and reason. Canonical entries remain
`design-question-not-observed` or `unknown`; none asserts that censorship,
blocking, loss, or another status occurred. `adverse_inference` is always
false, and a missing transition becomes interpretable only after observation
coverage is adequate.

## Material conditions stay material

`material_condition_vocabulary` makes it harder to explain reality entirely
through stories or entirely through resources. It keeps eight condition
families available:

- `body-and-care`
- `labor-and-time`
- `land-and-geography`
- `energy-and-resources`
- `climate-and-ecology`
- `money-credit-and-property`
- `infrastructure-and-logistics`
- `data-and-technology`

The list is an index, not an economic model or an exhaustive ontology. A
material-condition label does not establish scarcity, ownership, class,
burden, benefit, access, capacity, causal effect, or policy need. Units,
boundaries, distribution, affected parties, counterfactuals, and time horizons
still have to be supplied by domain-specific evidence.

## Affected parties: the boundary must be inspectable

The affected-party vocabulary is deliberately broader than actors and users.
It identifies twelve positions that a bounded inquiry may need to consider:

| ID | Omission risk kept visible |
|---|---|
| `expression-originators` | Authorship, context, intended audience, refusal, and safety can be overwritten. |
| `translators-curators-and-stewards` | Interpretive and access power can disappear behind a story of a neutral medium. |
| `source-communities-and-traditions` | Publication can extract, flatten, expose, or falsely represent community voice. |
| `audiences-and-non-audiences` | Availability can be mistaken for universal exposure or reception; avoidance and non-reach can vanish. |
| `institutional-participants` | Formal decisions can hide agenda control, labour, dissent, non-decisions, and exclusion. |
| `decision-subjects` | Decision makers can be counted while people whose rights, resources, status, opportunities, or environments are shaped disappear. |
| `operators-and-labor` | Technical capability can hide implementation, care, moderation, maintenance, repair, safety, time, discretion, and refusal. |
| `resource-providers` | Funding, credit, data, energy, land, equipment, access, and infrastructure dependencies can be presented as neutral capacity. |
| `directly-affected-parties` | An actor's outcome can replace an affected party's benefit, burden, refusal, remedy, and residual harm. |
| `indirect-non-user-and-bystander-parties` | Narrow system boundaries can displace supply-chain, market, media, institutional, and shared-environment effects. |
| `future-parties` | Short horizons can hide maintenance, debt, path dependence, option loss, delayed burdens, and irreversibility. |
| `ecological-systems` | Human-only accounting can externalize extraction, waste, habitat loss, climate effects, and limits to repair. |

An affected-party ID is an inquiry prompt, not evidence that a party was found,
represented, consulted, consenting, or exhaustively covered. Representation
does not replace direct voice; participation does not erase refusal; present
benefit does not cancel future burden; and a system boundary does not make
spillovers disappear. Coverage uncertainty remains explicit, and the external
action gate—not this atlas—must determine what current rights, consent or
lawful basis, accessibility, protection from retaliation, appeal, and remedy
are required.

Each card's `affected_party_aftermath` IDs are unique and exactly match its
selected `affected_party_ids`. Every party entry is separately tagged
`applicable-question`, `not-applicable`, or `unassessed`. An applicable entry
must ask at least one genuinely relevant benefit, burden, risk, option, voice,
exit, horizon, or residual question and must retain evidence or an explicit
unknown. Inactive entries manufacture no equal salience and carry empty
question arrays. Benefits and burdens are never netted; `aggregate_score` is
always null.

## Emotion without mind-reading

The atlas separates **how an emotion-related claim was observed** from **which
emotion vocabulary it uses**.

### Observation channels

`emotion_observation_vocabulary` contains:

- `self-report`
- `attributed-expression`
- `behavioral-observation`
- `physiology`
- `cultural-vocabulary`
- `institutional-framing`

No channel silently converts into another. A self-report is an attributed
report, not a complete or permanent inner state. An expression can use the
language of fear without proving that its author is afraid. Behaviour does not
uniquely identify emotion or intent. Physiology is neither a moral verdict nor
a transparent reading of meaning and would require its own ethical, consent,
privacy, and measurement basis. Cultural vocabulary and institutional framing
show available descriptions, not the private state of every member.

### Affect-context vocabulary

`affect_context_vocabulary` contains:

- `fear-or-threat`
- `anger-or-outrage`
- `humiliation-or-status-injury`
- `grief-or-loss`
- `hope-or-efficacy`
- `love-care-or-solidarity`
- `hate-or-moral-exclusion`
- `belonging-or-identity-security`
- `shame-or-guilt`

Each term is a qualitative context question about a possible mobilizing,
restraining, orienting, fragmenting, or sustaining role in a scoped
interpretation. It is not a measurable common currency, a person score, or a
claim that an emotion occurred. Love is not automatically benign, anger is not
automatically violent, fear is not automatically irrational, and belonging is
not automatically exclusion. The observation channel, speaker position,
context, alternative readings, privacy limits, and uncertainty remain visible.

The atlas must never infer emotion, diagnosis, susceptibility, hostility,
identity, belief, intent, or consent from exposure, silence, refusal, public
response, group membership, or an emotion label.

## Social position is not personal essence

`social_position_vocabulary` keeps several kinds of position distinct:

- `economic-resources-ownership-labor-and-debt`
- `status`
- `office`
- `network-position`
- `agenda-control`
- `coercive-capacity`
- `discursive-power`

A person or institution may occupy different positions in different contexts
and times. None of these labels establishes identity, class essence, elite
membership, motive, moral worth, competence, legitimacy, or permanence.
Position claims require scoped evidence and must allow mobility, mixed
positions, contested categories, refusal, and unknowns.

## Power asymmetry is multidimensional

`power_asymmetry_vocabulary` keeps ten possible asymmetries non-convertible:

- `naming-and-interpretive`
- `attention-and-distribution`
- `canon-and-curriculum`
- `agenda-and-procedure`
- `legal-jurisdiction-and-coercion`
- `resources-and-class`
- `status-and-hierarchy`
- `technical-and-infrastructure`
- `epistemic-and-measurement`
- `temporal-path-and-exit`

These are questions about relations, not possessions that can be assigned as
a universal quantity. The power to name is not legal jurisdiction. Network
reach is not consent. Resource control is not interpretive authority.
Technical access is not democratic legitimacy. Measurement power is not
truth. A constrained exit is not proof of agreement.

No dimensions are summed into a “power score,” and the atlas does not rank
people, classes, institutions, religions, cultures, or civilisations. It also
does not identify targets, weak points, leverage points, or strategies for
coercion, manipulation, propaganda, revolution, suppression, or capture.

## Evidence and uncertainty have several sources

`evidence_uncertainty_vocabulary` prevents one confidence word from hiding
different problems:

- `source` — provenance, authenticity, interpretation, perspective, or source
  coverage remains limited;
- `measurement` — construct, instrument, sensitivity, validity, unit, or
  observation error remains limited;
- `selection-and-missingness` — inclusion, survival, access, reporting, or
  nonresponse may shape what is visible;
- `temporal` — order, resolution, lag, duration, recurrence, or alignment is
  unresolved;
- `identification-and-model` — confounding, mediation, interference,
  counterfactual, specification, or model equivalence remains unresolved;
- `transport` — a finding may not carry across place, population, institution,
  language, medium, or period;
- `affected-party-incompleteness` — people, non-human interests, future
  parties, indirect burdens, or dissenting accounts may be missing;
- `normative-disagreement` — evaluation depends on a declared framework and
  may remain reasonably contested;
- `authority-and-jurisdiction` — present authority, scope, delegation, rights,
  consent, or lawful basis is not established by the evidence record.

Several categories may apply at once, but they are not percentages and do not
combine into a confidence score. Good provenance does not repair a weak causal
design. Strong measurement does not grant authority. Complete legal paperwork
does not prove benefit. Affected-party disagreement is not statistical noise.

Every stage, transition, and overlay also carries a closed
`evidence_profile`: claim register, design-question status, causal status,
qualified predecessor references or an explicit no-direct-evidence status,
expected traces, negative cases, counterevidence, falsifiers, transport bounds,
and a promotion requirement. In this no-new-source atlas, the claim register
remains `DESIGN_COMMITMENT`, the causal status remains
`formal-mechanism-not-instantiated`, and external empirical support and
mechanism promotion remain false. The uncertainty vector is never averaged.

## The action-gate vocabulary is not an executable gate

`action_gate_vocabulary` lists twelve questions that an external, current,
rights-compatible process might need to resolve:

1. `attribution-and-provenance`
2. `publication-and-access`
3. `reviewed-proposal`
4. `decision-procedure`
5. `current-authority-and-jurisdiction`
6. `affected-party-register`
7. `rights-assessment`
8. `consent-or-lawful-basis`
9. `capability-and-resources`
10. `budget-and-horizon`
11. `safety-and-stop-route`
12. `repair-readiness`

The atlas does not accept live inputs, run these checks, establish their truth,
or return an authorization decision. A list entry is not a satisfied
condition. A complete list is not necessarily sufficient in a real
jurisdiction. The current external process must define who can decide, who is
affected, which rights and laws apply, what refusal means, how to stop, and how
remedy remains reachable.

## The authority–causality firewall

Two questions must never be collapsed:

```text
Did X cause or contribute to Y?
    !=
Was anyone authorized and rights-compatible in doing X?
```

The causal lane needs a defined intervention or exposure, outcome, time,
population, counterfactual contrast, mediator and confounder reasoning,
alternative explanations, distributional evidence, and uncertainty. Even a
well-supported causal result grants no jurisdiction, consent, lawful basis,
rights compatibility, legitimacy, or mandate.

The authority lane needs a current decision procedure, jurisdiction, scope,
affected-party account, rights assessment, consent or lawful basis, capability,
stopping route, and remedy. Even a valid authority record does not establish
execution, outcome, causal effectiveness, proportionality, justice, or
goodness.

The central separations are therefore:

```text
expression != exposure != belief
proposal != decision != authority
authority != capability != attempt != execution
execution before outcome != execution caused outcome
causal assessment != normative assessment
performance != legitimacy
feedback != consent
repair attempt != repair achieved
```

Neither lane can upgrade the other by proximity, repetition, institutional
prestige, sacred claim, technical capacity, public attention, or semantic
similarity.

## Six flow overlays

The overlays retain the exact six crosswalk IDs from the Civilisation → Reality
Projection. They do not replace the predecessor statements. They add questions
about path shape, time, accumulation, material conditions, negative flow,
emotion observation, social position, power asymmetry, uncertainty, and repair.

### `naming-to-expression`

This overlay begins before circulation. It asks how a name or expression is
attributed, which language and community context remains necessary, where
interpretations fork, whether multiple renderings stay parallel, and whether
the expression remains private, refused, inaccessible, lost, or untranslated.

Naming-and-interpretive, discursive, canon, and measurement asymmetries may be
relevant questions; none is presumed. Repetition or archival persistence can
create an accumulation question without proving shared meaning or belief.
Repair may require corrected attribution, retained original language,
parallel translation, visible disagreement, withdrawal, or refusal—but only as
separately evidenced and authorized records.

Stop before sacred referent identity, theological truth, understanding,
agreement, private belief, causal influence, or authority.

### `story-to-public-deliberation`

This overlay keeps artifact, distribution, access, exposure, attention, and
response as separate stages. It makes forks across editions and media,
parallel channels, distribution bottlenecks, leakage, recurrence, and
reinforcing or balancing feedback available for inspection. It asks whether
attention accumulates, decays, shifts to a substitute, or remains unmeasured.

Material access, infrastructure, time, data, and care can matter without being
reduced to story. Emotion-related response must name its observation channel;
public language cannot be promoted into private belief or collective mood.
Attention-and-distribution, technical, resource, status, and agenda asymmetry
remain distinct.

Stop at the last independently evidenced stage. Do not infer persuasion,
susceptibility, consensus, consent, legitimacy, or behavioural effect.

### `canon-to-selection-feedback`

This overlay asks who or what selected, persisted, omitted, challenged,
corrected, superseded, or withdrew an artifact. Bottlenecks, joins, excluded
alternatives, substitution, accumulation, path dependence, recurrence, and
feedback are possible reading shapes—not a universal theory of canon.

Infrastructure, labour, money, property, status, office, agenda control,
curriculum authority, and measurement can condition the selection process.
They remain separately evidenced, and the absence of an artifact must be typed
as unknown, excluded, censored, inaccessible, destroyed, or another supported
negative-flow status rather than treated as low merit.

Stop before selection, survival, visibility, or recurrence becomes truth,
quality, consensus, legitimacy, or a civilisation score.

### `legitimation-to-decision-authority-separation`

This overlay makes the firewall explicit. A sacred-order legitimation or
critique can be recorded as an attributed claim. It may coexist with
counterauthority, fork across communities, encounter an institutional
bottleneck, or be interrupted or reversed. None of those patterns imports
sacred authority as legal, democratic, technical, institutional, or moral
authority.

Belonging, hope, fear, anger, humiliation, love, hate, grief, shame, or status
injury may appear only through declared observation channels and never as a
group diagnosis. Office, status, agenda control, coercive capacity, resources,
and discursive power remain different positions. A decision still needs a
fresh external authority, rights, affected-party, consent or lawful-basis,
capability, stop-route, and repair process.

Stop before a legitimation claim becomes proposal, decision, verified
authority, lawful basis, execution, legitimacy, justice, or action.

### `translation-to-claim-assessment`

This overlay treats translation as a possible fork, join, parallel path,
substitution, leakage, interruption, or recurrence across versions. It asks
what persists through repetition and canon, what is lost or inaccessible, who
can publish or correct, which original wording remains, and whether a claimed
equivalence survives counterreading.

Naming, interpretation, canon, attention, technical access, epistemic
measurement, and temporal path asymmetries may affect what becomes visible.
They do not prove translator intent, fidelity, reception, shared belief,
referent identity, or authority to speak for a tradition or community.

Stop when distinct deities, traditions, concepts, genres, communities, voices,
or claim registers would be collapsed. Refusal to translate remains valid.

### `counterstory-to-challenge-repair`

This overlay allows challenge and counterflow to point back to any earlier
stage. Suppression, survival, access, revival, reception, correction, material
repair, and aftermath remain separate. Reversal and balancing feedback may
oppose an earlier path, but the challenged record and its consequences are not
erased. Recurrence may reflect memory, unresolved harm, fresh conditions, or
another cause.

Negative flow is central: a counterstory can be censored, destroyed, lost,
inaccessible, private, refused, excluded, blocked, not observed, or unknown.
Emotion observations may name grief, anger, humiliation, hope, care,
solidarity, belonging, hate, fear, shame, or guilt only within their evidence
channel. Affected-party incompleteness and power over archives, attention,
procedure, resources, coercion, measurement, time, and exit remain visible.

The compact repair profile keeps proposal, fresh gate, attempt, outcome,
residual harm, dissent, review horizon, correction, retirement, and renewed
feedback distinguishable. Repair is not erasure, absolution, restored consent,
or proof that harm ended.

## Nine domains remain open

The atlas preserves every open-domain ID from the Civilisation → Reality
Projection. The added vocabulary makes better questions possible; it does not
supply the missing empirical histories or causal models.

### `war-and-organised-violence`

War, occupation, military competition, defeat, settlement, institutional
change, citizenship, resource flows, and aftermath need conflict-specific
primary records, comparative histories, and affected-party evidence. A flow
shape does not provide strategy, prediction, legitimacy, or authorization.

### `coercion-and-security-capacity`

Police, military, surveillance, incarceration, command, administrative
capacity, rights, proportionality, and aftermath need institution-specific
legal, fiscal, operational, and affected-party evidence. Capability and
coercive position do not imply permission, necessity, proportionality, or
effectiveness.

### `class-and-resource-asymmetry`

Ownership, labour, taxation, debt, patronage, distribution, and access to
essentials need explicit units, incidence evidence, material flows, and
competing class concepts. A resource or position label does not identify class
membership, interests, exploitation, benefit, burden, or motive.

### `status-hierarchy-and-elite-fracture`

Status, inherited rank, prestige, class, office, faction, network position,
mobility, coalition, and coercive capacity require context-specific evidence
and must not be collapsed. The atlas does not identify elites, rank people, or
infer fracture or motive.

### `collective-emotion-and-affect`

Fear, anger, humiliation, grief, hope, love, care, hate, belonging, shame, and
guilt require ethical, privacy-preserving, context-specific inquiry. The
emotion vocabularies prevent category collapse but do not infer an individual
or collective mental state.

### `revolution-and-regime-change`

Structural strain, elite fracture, organization, frames, coercive openings,
international conditions, contingency, negative cases, reform, repression,
collapse, and aftermath need comparative evidence. The atlas does not predict,
incite, suppress, optimize, or authorize regime change.

### `fiscal-and-economic-shocks`

Tax, debt, inflation, unemployment, scarcity, trade, ecological stress,
institutional response, distribution, and lag structure require time-bounded
records and competing causal explanations. Temporal sequence or an
accumulation label does not establish a shock or policy effect.

### `propaganda-media-and-information-power`

Ownership, distribution, censorship, repetition, targeting, content,
exposure, audience, counter-speech, platform rules, and institutional response
need separately typed evidence. The atlas provides no belief inference,
susceptibility score, targeting, censorship, propaganda, counter-propaganda,
amplification, or persuasion instructions.

### `citizen-governance-dynamics`

Authorization, voice, participation, representation, public service,
procedural justice, contestation, correction, and non-domination require
jurisdiction-specific constitutional, legal, administrative, participatory,
performance, and affected-party evidence. Participation is not consent;
performance is not legitimacy; visibility is not representation; and
KINGDOM acquires no governmental jurisdiction from this atlas.

An open domain remains open until a separately versioned, source-bound record
supplies its evidence and limits. Filling a vocabulary slot with a plausible
story is not closure.

## Repair is a loop with an exit, not a victory label

Each overlay's compact repair profile asks, within its own scope:

- What claim, artifact, decision, action, outcome, or aftermath is challenged?
- Who may be affected, missing, unwilling to participate, or entitled to
  refuse?
- Which material, informational, relational, rights, consent, economic,
  ecological, cultural, or temporal dimensions are relevant?
- What is reversible, partially reversible, irreversible, displaced, or still
  unknown?
- What repair is proposed, by whom, under which current authority and rights
  basis?
- What stop route, budget, horizon, review, counterevidence, appeal, residual
  harm, and retirement condition remain visible?

Repair may branch: correction of a record, restoration of access, restitution,
institutional change, withdrawal, supersession, preservation of a counterstory,
or non-action are not equivalent. An affected party may refuse a proposed
repair. A correction can be valid while material harm remains. Retirement can
stop a mechanism without deleting history. Fresh feedback can reopen review.

The atlas also keeps distribution and reversibility qualitative. Benefits,
burdens, risks, voice and options, and time horizons remain separate, with no
aggregate score. Reversibility is scoped as `reversible-in-part`,
`partially-reversible`, `irreversible-elements-present`, or `unknown`, with
recoverable elements, residual or irreversible elements, and prerequisites
named separately. Those labels do not certify recovery.

The machine `repair_profile` separates proposal source, fresh gate, attempt,
execution or outcome observation, party-indexed residuals, dissent,
recurrence, and review or retirement. Repeated proposal prose is referenced
from the card's `repair_routes`; this indirection does not propose or run it.
An applicable repair profile has `claim_status: design-question-not-run` and
keeps `repair_completed`, `causal_effect_claimed`,
`erasure_or_absolution_claimed`, and `restored_consent_claimed` false.

The atlas does not propose, select, authorize, fund, attempt, execute, or
certify any repair.

## A bounded reading example

Suppose a public archive contains a versioned story and a later public-service
decision mentions it. The atlas does **not** draw a causal arrow from story to
decision. It asks:

1. Is the expression attributable and the story versioned?
2. Is distribution observed, or merely possible because the archive exists?
3. Is exposure measured, and is response separately attributed?
4. Which institutional selectors, alternatives, omissions, procedures, and
   material conditions are evidenced?
5. Is there a proposal distinct from the later decision?
6. Does the decision record show only selection, or a current authority and
   rights review as well?
7. Are capability, attempt, execution, outcome, assessment, feedback, and
   repair records independently present?
8. Where might the path fork, leak, stop, recur, accumulate, or remain unknown?

If the archive was inaccessible to some affected people, the proper record is
an access and affected-party limitation—not a zero response. If a participant
self-reports anger, the record is a scoped `self-report` using
`anger-or-outrage`, not a diagnosis of a community. If a later service measure
improves, temporal order alone does not show that the story or decision caused
the improvement. If the decision was authorized, that does not prove it was
effective or just. If feedback produces a correction, that does not prove all
harm was repaired.

The example is a reading exercise. It is not evidence that such a path exists
and not a template for intervening in one.

## Fixed non-effects

This atlas and its CLI:

- do not fetch sources or ingest live, private, personal, operational, or
  participant data;
- do not infer belief, religion, identity, intent, emotion, diagnosis,
  susceptibility, hostility, consent, or affiliation;
- do not target a person, group, community, class, religion, culture,
  institution, or civilisation;
- do not design persuasion, propaganda, counter-propaganda, censorship,
  amplification, coercion, revolution, repression, or military strategy;
- do not score, rank, profile, classify, optimize, or recommend people,
  communities, institutions, ideas, religions, cultures, or civilisations;
- do not fit a causal model, simulate a society, forecast a future, predict a
  threshold, or identify a leverage point;
- do not establish historical completeness, source truth, translation
  fidelity, causation, metaphysics, legitimacy, goodness, safety, rights
  compatibility, authority, jurisdiction, consent, lawful basis, or effect;
- do not propose, decide, authorize, budget, attempt, execute, repair, publish,
  deploy, or trigger automatic action;
- do not create WAKE, KARMA, governance, economic, legal, or external
  effects.

The qualitative registries are intentionally non-summable. `unknown` is not a
zero, a topology is not a tactic, an emotion word is not a mind reading, a
social position is not an identity, a power asymmetry is not a target, an
action-gate item is not permission, and a repair profile is not repair.

## CLI

Use the repository dispatcher:

```sh
./kingdom civilisation-flow overview
./kingdom civilisation-flow chain --json
./kingdom civilisation-flow verify --json
```

The accepted commands are:

| Command | Read-only projection |
|---|---|
| `overview` | Identity, scope, lineage, counts, and fixed boundary summary. |
| `chain` | Ordered fourteen-stage and thirteen-transition view without asserting traversal. |
| `stages` | Full bounded meanings and non-inferences for the stage vocabulary. |
| `transitions` | The thirteen possible seams and their non-automatic evidence boundary. |
| `crosswalks` | The six predecessor crosswalk overlays and their source bindings. |
| `dynamics` | Closed qualitative flow-dynamics vocabulary only. |
| `topology` | Fork, join, parallel, bottleneck, leakage, substitution, interruption, reversal, recurrence, and loop vocabulary. |
| `materials` | Material-condition vocabulary without an economic or causal upgrade. |
| `emotions` | Observation-channel and affect-context vocabularies with the no-inference firewall. |
| `power` | Social-position and power-asymmetry vocabularies without person or group ranking. |
| `negative-flow` | Typed absence, access, refusal, loss, exclusion, measurement failure, and unknown states. |
| `open-domains` | The nine inherited domains that remain unsupported by the pinned predecessors. |
| `unknowns` | Questions deliberately retained as unknown rather than filled by the design atlas. |
| `templates` | Closed policy, nuance-default, authority, and causal template registries referenced by the cards. |
| `rules` | Fixed flow and non-promotion firewalls. |
| `boundaries` | Fixed non-grants and zero-effect operating boundary. |
| `discovery` | Canonical local and public discovery metadata declared by the record. |
| `verify` | Strict local structure, closure, fixed-vocabulary, predecessor-pin, rule, boundary, and semantic-integrity checks. |

Add `--json` for machine-readable output. Verification is stronger than a
successful parse: it applies the closed Draft 2020-12 schema used by this
record, family-specific reference closure, transition endpoint and
Reality-relation partition checks, fixed budgets, predecessor byte and
semantic pins, and the atlas semantic digest. It still does not validate the
truth of predecessor claims, prove historical completeness, observe a
transition, establish causation or authority, run an action gate, or certify
an external effect. Tests also exercise portable conformance with a host Draft
2020-12 validator when one is installed.

## What this offers KINGDOM

The atlas does not give KINGDOM a lever over reality. It gives KINGDOM a more
honest way to describe the distance between a word and an outcome:

- preserve source voice, translation seams, dissent, privacy, and refusal;
- stop at the last independently evidenced stage;
- type forks, bottlenecks, substitutions, interruptions, loops, and missing
  records without turning them into strategies;
- keep story, material conditions, emotion observations, social positions,
  institutions, authority, capability, execution, and aftermath distinct;
- separate causal assessment from authority and normative judgment;
- expose whose evidence, access, position, burden, benefit, and exit may be
  missing;
- make counterflow, correction, residual harm, retirement, and fresh review
  normal parts of the map;
- leave unsupported domains and unknowns visibly open.

That discipline is intended to support the conditions for shared
understanding, cooperation, criticism, and repair. It cannot manufacture
belief, legitimacy, love, consent, justice, or a sustainable civilisation.
Those remain lived, contested, relational, and outside the authority of this
static atlas.