#Part D - Multi-scale Ethics and Conflict Optimization

Preface node heading:part-d-multi-scale-ethics-and-conflict-optimization:62966

#What this page is

This is generated FPF reference text from the specification preface or supporting sections. It helps interpret FPF; it is not FPF Reference product documentation.

#Methodology

Use it to understand how the specification wants to be read, then return to a route, pattern, or work packet for active work. Cite generated IDs only when the wording changes the task decision.

#Content

#Ethical Value Plurality and FPF Boundary

Type: D-family ethical boundary pattern Status: Stable Pattern role: This compact pattern gives the stable entry boundary and conformance checks for value-plurality use; fuller ethical theory remains outside FPF unless a direct pattern names it.

Use this when. Use this pattern when an FPF claim, method, work plan, architecture move, policy, recommendation, model, or system change has ethical force, but the value theory or ethical concern behind the claim is not yet explicit.

Not this pattern when. If the current question is already a conflict across declared levels or scopes, use D.3. If the current question is how to mediate that conflict or use it in a decision, use D.4. If the current question is bias, fairness, human or group impact audit, causal-fairness audit consumption, or ethical assurance, use D.5.

What goes wrong if missed. FPF looks ethically neutral because it names evidence, method, architecture, or work but leaves the value frame and affected EntityOfConcern implicit.

What this buys. The ethical concern becomes a bounded FPF claim with value frame, affected EntityOfConcern, evidence, admissible use, and direct next owner.

#Problem Frame

FPF cannot prescribe one final ethics doctrine and still remain usable across engineering, research, organizational, public, and AI-enabled work. But FPF also cannot treat ethical neutrality as permission to omit ethics. Many working claims already carry values: who may be harmed, who benefits, which consequences count, which responsibilities are accepted, what evidence is enough, and which sacrifice is treated as admissible.

D.1 supplies the boundary rule. When the ethical claim matters, make the value concern explicit enough that neighboring FPF patterns can inspect it. Do not hide it inside words such as "responsible", "safe", "fair", "humane", "acceptable", or "aligned" without saying what is being valued, for whom, in which context, and with what evidence.

#Problem

Ethically loaded FPF claims often arrive as ordinary technical, architectural, method, evidence, or publication claims. The failure is not that the text lacks moral vocabulary; the failure is that the affected EntityOfConcern, value concern, evidence, admissible use, and stronger-source return condition are not explicit enough to inspect.

#Forces

#Solution

Recover an EthicalValueFrame@Context before treating the claim as ethically admissible:

EthicalValueFrame@Context:
  ethicalClaimRef
  affectedEntityOfConcernRef
  boundedContextRef
  valueConcernRefs
  ethicalTheoryOrTraditionRefs?
  affectedHolonRefs?
  affectedEpistemeRefs?
  roleAssignmentRefs?
  methodOrWorkRefs?
  transformationRefs?
  evidenceRefs
  uncertaintyOrCurrentnessCondition
  admissibleUse
  inadmissibleOverread
  strongerSourceReturnCondition

This frame does not settle the ethical question. It makes the value frame inspectable. A utilitarian consequence claim, a deontic constraint, a virtue or character claim, a care-ethics concern, a rights claim, a professional-duty claim, and a project-specific value trade-off may all be admissible starting points, but they must not be presented as the same claim merely because the same word "ethical" appears.

#Boundaries

D.1 keeps value plurality and FPF boundary discipline. It does not replace:

#Archetypal Grounding (Worked Slice)

A team says that a triage model is "ethical because it maximizes total benefit." D.1 does not accept the phrase as a finished ethical judgment. It records the affected patients and institutions, the value concern called "total benefit", the consequence theory being used, the excluded concerns such as equal access or avoidable harm to a subgroup, the evidence set, and the admissible use of the claim. If equal access or subgroup harm becomes live, D.3 maps the conflict and D.4 governs mediation or decision use.

#Bias-Annotation

#Conformance Checklist

#Common Anti-Patterns and How to Avoid Them

#Consequences

This pattern makes ethical claims portable across FPF without pretending that FPF has one final ethical theory. It also prevents a common failure: a technical pattern silently inherits one ethical theory because a word such as "safe", "fair", "beneficial", or "responsible" sounded ordinary.

The cost is extra explicitness. The gain is that ethics becomes reviewable in the same FPF body as systems, methods, work, evidence, assurance, architecture, and publication use.

#Rationale

D.1 is an entry boundary for ethical value plurality. It is intentionally modest: it does not settle ethical theory and does not decide an interlevel conflict. It makes the live value frame visible enough for neighboring FPF patterns to carry the next claim without hiding ethics inside technical adequacy, evidence, architecture, method, work, or publication wording.

This keeps FPF usable in engineering, research, organizational, public, and AI-enabled contexts where ethical traditions differ but value-bearing claims still need explicit affected entities, evidence, admissible use, and stronger-source return conditions.

#SoTA-Echoing

#Relations

• Builds on A.1 and A.7 for EntityOfConcern and description distinction.
• Coordinates with A.10 for evidence, source currentness, and source-use relations.
• Coordinates with B.3 when an assurance claim is current.
• Coordinates with D.2, D.3, D.4, and D.5 for multilevel entry, conflict structure, mediation, bias audit, causal-fairness audit consumption, and ethical assurance.
• Coordinates with C.28 for causal fairness use and with C.30.ILC when an architecture residual is current.

#D.1:End

#Multilevel Ethics For Holon Work

Type: D-family ethical entry pattern Status: Stable Pattern role: This compact pattern recognizes multilevel ethical concern and selects the next owner; it does not settle conflict or supply a fixed level ladder.

Use this when. Use this pattern when a system, holon, method, work plan, work occurrence, policy, recommendation, architecture move, or publication use may improve one declared level or scope while harming another, or when responsibility is assigned across levels.

Not this pattern when. If only the value frame is missing, use D.1. If the conflict structure is already current, use D.3. If the conflict has to be mediated or used in a decision, use D.4. If the current concern is bias, fairness, impact audit, causal-fairness audit consumption, or ethical assurance, use D.5.

What goes wrong if missed. A local improvement is treated as ethically sufficient while another declared level, scope, or affected holon carries the harm.

What this buys. The practitioner names the affected levels or scopes, the current value frame, and the next owner before mediation, assurance, bias audit, or architecture return.

#Problem Frame

Ethical trouble in system-holon work often appears because the current action is good for one level and bad for another. A person may benefit while a team is damaged. A project may benefit while a community pays the cost. A standard may improve coordination while excluding a minority case. A model may improve one metric while moving harm to a less visible scope.

Do not force these cases into a fixed ladder such as local, group, ecosystem, planetary. The relevant levels and scopes must be declared from the situation: person, team, organization, community, society, polity, economy, built asset, project, environment, episteme family, standard, publication, AI-enabled system, or another admitted holon or context.

#Problem

A change can be beneficial at one declared level or scope while imposing harm, exclusion, risk, or responsibility elsewhere. The failure is to treat the local gain as ethically sufficient before the affected levels, scopes, holons, epistemes, role assignments, work, evidence, and next owner are named.

#Forces

#Solution

Open a MultilevelEthicsEntry@Context:

MultilevelEthicsEntry@Context:
  ethicalConcernRef
  affectedEntityOfConcernRef
  boundedContextRef
  declaredLevelOrScopeRefs
  affectedHolonRefs
  affectedEpistemeRefs?
  roleAssignmentRefs
  interestOrConcernRefs
  capabilityOrFunctioningConcernRefs?
  methodOrWorkRefs?
  transformationRefs?
  expectedConsequenceRefs
  evidenceRefs
  uncertaintyOrCurrentnessCondition
  nextOwnerRef

The entry record has one job: recognize that multilevel ethics is live and choose the next owner. It does not itself resolve the conflict.

For this pattern, holon work includes material systems and epistemes when they are the affected EntityOfConcern. An architectural description, standard, model card, policy publication, or research program may be the affected episteme; the pattern still asks which levels, scopes, affected holons, interests, responsibilities, and consequences are live.

#Recognition Matrix

#Boundaries

D.2 is an entry pattern, not a general ethics doctrine and not a conflict solver. It keeps ethics from being omitted when levels and scopes of holons matter. It also keeps multilevel ethics from replacing architecture, assurance, causal, evidence, or publication patterns before the ethical EntityOfConcern is clear.

D.2 does not create U.Level, U.Frustration, U.Emergence, or a fixed moral scale. Levels and scopes are declared relations in the current situation. If a mathematical lens is needed for scale, frustration, optimization, Pareto comparison, or renormalization-like reasoning, use C.29 and the owning pattern for the current object.

#Archetypal Grounding (Worked Slice)

A product team wants to reduce service cost by making a medical device harder to service outside authorized centers. The move may improve manufacturer quality control and reduce liability risk, but harm patients in regions where authorized service is unavailable. D.2 opens the entry: manufacturer, patients, service organizations, device fleet, and regulatory context are declared as affected scopes; value concerns include safety, access, responsibility, and maintainability; the work plan and expected consequences are named. D.3 then maps the conflict; D.4 governs mediation or decision use.

#Bias-Annotation

#Conformance Checklist

#Common Anti-Patterns and How to Avoid Them

#Consequences

This pattern makes ethical level structure visible early. It prevents two opposite errors: treating ethics as a late bias audit only, and treating every interlevel residual as architecture without first asking whether value, harm, responsibility, or admissible sacrifice is being claimed.

#Rationale

D.2 makes multilevel ethical concern visible before the work jumps to conflict mediation, bias audit, assurance, or architecture. This matters because many cases look technically local but ethically cross-level: a method, standard, architecture move, publication, or work plan can improve one scope while pushing cost, risk, exclusion, or responsibility elsewhere.

The pattern deliberately avoids a fixed ladder. It asks for declared levels and scopes from the situation, then selects the next owner. That keeps FPF holon-aware without making every cross-scope case an architecture residual or every ethical case a bias audit.

#SoTA-Echoing

#Relations

• Builds on D.1 for ethical value frame boundary.
• Builds on A.1, B.1, and C.13 for holon, level, scope, and part-whole grounding.
• Coordinates with D.3 for interlevel ethical conflict structure and with D.4 for mediation or decision use.
• Coordinates with D.5 for bias, fairness, impact audit, causal-fairness audit consumption, and ethical assurance.
• Coordinates with A.15, A.3.4, C.16, C.29, and C.30.ILC when method, work, transformation, measurement, mathematical lens, or architecture residual claims are current.

#D.2:End

#Interlevel Ethical Conflict Structure

Type: D-family ethical conflict-structure pattern Status: Stable Pattern role: This compact pattern owns the structure of an interlevel ethical conflict; mediation, decision use, assurance, causal use, and architecture residuals remain with their direct owners.

Use this when. Use this pattern when an ethical conflict spans declared levels or scopes and the conflict structure itself must be made inspectable before mediation, decision, assurance, or architecture return.

Not this pattern when. If only the ethical value frame is missing, use D.1. If only entry recognition is needed, use D.2. If the conflict structure is already mapped and the current question is mediation or decision use, use D.4. If the question is bias, fairness, impact audit, causal-fairness audit consumption, or ethical assurance, use D.5.

What goes wrong if missed. The team debates values or decisions before it has named the levels or scopes, carriers, harms, benefits, evidence, and residuals that actually conflict.

What this buys. The conflict becomes an inspectable structure that D.4, D.5, assurance, causal, and architecture owners can use without guessing.

#Problem Frame

Interlevel ethical conflict is not just disagreement between people. It may involve a system part and a whole, a person and an organization, one organization and a community, a project and a society, a collection and its members, an episteme family and the decisions it shapes, or an architecture move and the holon levels it affects.

The central move is structural: name what is in conflict, at which declared levels or scopes, through which methods, work, transformations, role assignments, evidence, value concerns, and consequence horizons. Do not turn the conflict into publication wording, assurance claim, or architecture residual unless that is the current governed object.

#Problem

Interlevel ethical conflict is often debated before it is structured. The failure is to argue over values or decisions while the affected objects, declared levels or scopes, value frames, methods, work, transformations, evidence, uncertainty, thresholds, and consequence horizons remain implicit.

#Forces

#Solution

Record an InterlevelEthicalConflictStructure@Context:

InterlevelEthicalConflictStructure@Context:
  conflictConcernRef
  boundedContextRef
  affectedEntityOfConcernRefs
  declaredLevelOrScopeRefs
  affectedHolonRefs
  affectedEpistemeRefs?
  collectionOrMembershipRelationRefs?
  partWholeRelationRefs?
  roleAssignmentRefs
  interestOrConcernRefs
  valueFrameRefs
  agencyOrResponsibilityThresholdRefs?
  methodOrWorkRefs?
  transformationRefs?
  evidenceRefs
  uncertaintyRefs
  consequenceHorizonRefs
  conflictRelationRefs
  nonConflictOverread
  nextUseOwnerRef

This structure may be represented by a table, graph, formal predicate, narrative case, or another selected description form. The representation is not the conflict itself. If a mathematical lens does work in the claim, cite [C.29](/generated/patterns/C.29); if the publication form changes admissible use, cite [E.17](/generated/patterns/E.17).

#Collection and Episteme Cases

A collection is ethically current only when whole-level characteristics, membership relations, environment-mediated effects, or aggregate consequences matter. Use A.14 for part-whole and membership relation vocabulary and C.13 for constructive grounding. Do not assign responsibility to a collection merely because it has a plural name.

An episteme is ethically current when its claim-bearing structure, source-use relation, publication relation, described EntityOfConcern, or model family changes affected systems or decisions. Use C.2.1 for the episteme slot relation and E.17 for publication claims. Do not turn every ethical conflict over a theory, standard, architecture description, or policy description into a problem about wording.

#Boundaries

#Archetypal Grounding (Worked Slices)

Engineering advice. A consultant improves the effectiveness of a client's harmful project. The conflict is not only "bad client, good method." D.3 maps the client organization, affected public, consultant role assignment, method, work occurrence, responsibility threshold, evidence uncertainty, and consequence horizon. D.4 governs refusal, conditions, escalation, or decision use.

Collection case. A fleet-level optimization reduces maintenance cost but increases failure risk for a small subfleet used in harsher conditions. D.3 names the fleet, subfleet, membership relation, affected users, evidence set, value concerns, and consequence horizon. It does not infer that the fleet is a responsible super-holon unless an admitted pattern allows that claim.

Episteme case. A published architecture description normalizes an interface assumption that excludes an alternative implementation option. The ethical conflict may involve the episteme whole, its source-use relation, affected suppliers, and system consequences. D.3 maps the conflict; C.30.AD governs architecture-description adequacy and E.17 governs publication-use claims.

#Bias-Annotation

#Conformance Checklist

#Common Anti-Patterns and How to Avoid Them

#Consequences

This pattern gives D.4, D.5, B.3, C.28, A.10, C.11, and C.30.ILC a conflict structure they can use without stealing the ethical object. The cost is that conflicts cannot be waved through by a slogan. The gain is that mediation and decision work start from a typed structure rather than from a politically convenient label.

#Rationale

D.3 provides the typed structure that ethical mediation, assurance, causal, architecture, and bias-audit patterns can use. Without it, D.4 receives slogans rather than inspectable conflicts; D.5 receives fairness claims without affected scopes; architecture receives value conflict disguised as residual; and evidence patterns receive claims with no declared consequence horizon.

The pattern therefore focuses on the conflict EntityOfConcern: affected objects, declared levels or scopes, value frames, role assignments, methods, work, transformations, evidence, uncertainty, thresholds, and consequence horizons. It keeps descriptions of the conflict useful but secondary.

#SoTA-Echoing

#Relations

• Builds on D.1 and D.2 for value-frame boundary and multilevel entry.
• Builds on A.1, A.14, B.1, and C.13 for holons, part-whole, membership, collections, and constructive grounding.
• Coordinates with D.4 for mediation and decision use.
• Coordinates with D.5 for bias, fairness, impact audit, causal-fairness audit consumption, and ethical assurance.
• Coordinates with C.2.1 and E.17 for episteme and publication-use claims.
• Coordinates with C.30.ILC, A.10, B.3, C.28, and C.29 when architecture residual, evidence, assurance, causal, or mathematical-lens claims are current.

#D.3:End

#Ethical Mediation and Decision Use

Type: D-family ethical mediation and decision-use pattern Status: Stable Pattern role: This compact pattern owns the ethical use of an already mapped conflict: mediation, refusal, evidence demand, bounded decision use, and residual handling.

Use this when. Use this pattern when an interlevel ethical conflict structure from D.3 must be used for mediation, refusal, decision, evidence demand, causal return, assurance return, or architecture return.

Not this pattern when. If the conflict structure is not yet mapped, use D.3. If the issue is only value plurality, use D.1. If the issue is only entry recognition, use D.2. If the current work is bias, fairness, impact audit, causal-fairness audit consumption, or ethical assurance, use D.5.

What goes wrong if missed. A mapped ethical conflict is treated as solved, blocked, or decision-ready without naming mediation, refusal, evidence demand, return, accepted residual, or bounded decision use.

What this buys. The practitioner can use a D.3 conflict structure for one admissible mediation action or bounded decision use while keeping evidence, causality, assurance, architecture, and bias-audit claims with their owners.

#Problem Frame

Once an interlevel ethical conflict is visible, the next risk is premature closure. A team may declare a compromise before evidence is sufficient, turn an assurance input into ethical permission, use one level's value as a trump card, or hide a refusal behind technical language.

D.4 governs the use of the mapped conflict. It does not make FPF a final moral authority. It asks what move is admissible from the current conflict structure and what must return to evidence, causality, assurance, architecture, decision, or value framing before action is justified.

#Problem

A mapped ethical conflict can still be used badly. The failure is to treat a conflict structure, assurance input, formula, or architecture return as if it already selected a compromise, refusal, evidence demand, accepted residual, or bounded decision use.

#Forces

#Solution

Record an EthicalMediationDecisionUse@Context:

EthicalMediationDecisionUse@Context:
  conflictStructureRef
  boundedContextRef
  valueFrameRefs
  decisionQuestionRef?
  optionRefs
  proposedMediationRefs?
  refusalOrStopCondition?
  evidenceDemandRefs?
  causalReturnRefs?
  assuranceReturnRefs?
  architectureResidualReturnRefs?
  acceptedResidualRefs?
  decisionRecordRefs?
  admissibleUse
  inadmissibleOverread
  strongerSourceReturnCondition

The record names the current ethical use of the conflict: mediate, refuse, continue under explicit residual, demand evidence, ask a causal question, ask for assurance, return to architecture, or make a bounded decision.

#Mediation Moves

#Archetypal Grounding (Worked Slices)

Fair-share case. A service outage plan can protect hospitals, households, or industrial customers, but not all at once. D.3 maps affected scopes and value concerns. D.4 records the mediation use: options, accepted residuals, evidence demand, role assignments for decision responsibility, and return conditions. A mathematical allocation method may be cited through C.29, but the allocation formula is not the ethical decision by itself.

Override case. An assurance review says a release has the required technical assurance relation, but D.3 shows unresolved harm for a subgroup. D.4 does not let assurance override the ethical conflict. It records whether release is refused, conditioned, delayed for evidence, returned to C.28 causal-use analysis, or allowed with explicit residual and responsibility.

#Boundaries

D.4 does not own conflict structure, bias audit, ethical assurance, architecture residual, causal identification, evidence provenance, or decision theory in general. It owns the ethical use of a mapped interlevel conflict.

Do not name a mediation move "calculus" unless a mathematical lens is selected and the lens is actually doing work. Do not name a mediation move "operator" unless the current pattern explicitly governs an operation. Most D.4 use is a bounded decision-use record, not a mathematical object.

#Bias-Annotation

#Conformance Checklist

#Common Anti-Patterns and How to Avoid Them

#Consequences

This pattern makes ethical action reviewable without pretending that every conflict has a clean optimum. It preserves refusal, evidence demand, and residual acceptance as first-class outcomes. It also prevents architecture, assurance, or causal evidence from quietly becoming moral permission.

#Rationale

D.4 exists because an inspectable ethical conflict still needs a bounded use. Some uses stop work. Some demand evidence. Some return to causal, assurance, or architecture owners. Some proceed under an accepted residual with named responsibility and return conditions. Without this pattern, teams either freeze because conflict exists or move too fast because the conflict was mapped once.

The pattern keeps refusal, evidence demand, and residual acceptance visible as ordinary outcomes. It also prevents formulas, assurance labels, architecture residual repairs, or causal claims from silently becoming moral authorization.

#SoTA-Echoing

#Relations

• Builds on D.3 for the mapped conflict structure.
• Coordinates with D.1 and D.2 when value frame or multilevel entry is incomplete.
• Coordinates with D.5 when bias, fairness, impact audit, causal-fairness audit consumption, or ethical assurance is current.
• Coordinates with A.10, B.3, C.11, C.28, C.29, and C.30.ILC when evidence, assurance, decision, causal, mathematical-lens, or architecture-residual claims are current.

#D.4:End

#Bias Audit and Ethical Assurance

Type: D-family bias-audit and ethical-assurance boundary pattern Status: Stable Pattern role: This compact pattern owns bias, fairness, impact-audit, causal-fairness audit consumption, and ethical-assurance boundary use; it does not replace D.1 through D.4.

Use this when. Use this pattern when a model, metric, policy, publication, decision system, recommendation, method, work plan, system, holon, or FPF claim may create bias, unfairness, human or group impact, causal-fairness overclaim, or ethical assurance risk.

Not this pattern when. If the ethical value frame is missing, use D.1. If the current question is multilevel ethics entry, use D.2. If the current question is interlevel ethical conflict structure, use D.3. If the current question is mediation or decision use of that conflict, use D.4. If the current question is only evidence, causality, assurance, measurement, or architecture residual without bias, fairness, human or group impact, or ethical assurance, use the direct owner.

What goes wrong if missed. A model, metric, policy, publication, or decision system passes ordinary evidence or assurance checks while representation, proxy, visibility, metric, language, or human-impact bias remains hidden.

What this buys. Bias, fairness, human-impact, causal-fairness, and ethical-assurance concerns become auditable without replacing D.1 through D.4, evidence, causal, measurement, or architecture owners.

#Problem Frame

Bias and fairness failures often survive ordinary verification. A metric may be accurate while hiding subgroup harm. A model may be predictive while reproducing past exclusion. A policy may look neutral while moving cost to people or groups who were not represented in the evidence. A publication may look technically clear while licensing a harmful use.

D.5 keeps this audit and assurance question explicit. It does not replace multilevel ethics. It asks whether the current object and its intended use are ethically unsafe because of bias, unfairness, impact, causal fairness without the required C.28 evidence value, or assurance without the required assurance relation.

#Problem

Bias, fairness, human-impact, causal-fairness, and ethical-assurance concerns can remain invisible after ordinary technical verification. The failure is to let the model, metric, policy, publication, method, work plan, system, or holon be treated as admissible for use while the audited EntityOfConcern, intended use, affected people or groups, evidence, mitigation, and residuals are not explicit.

#Forces

#Solution

Open a BiasAuditAssuranceFrame@Context:

BiasAuditAssuranceFrame@Context:
  auditedEntityOfConcernRef
  boundedContextRef
  intendedUseRef
  affectedPeopleOrGroupRefs?
  affectedHolonRefs?
  metricOrModelRefs?
  policyOrPublicationRefs?
  biasConcernRefs
  fairnessClaimRef?
  impactClaimRef?
  causalFairnessUseRef?
  evidenceRefs
  assuranceClaimRefs?
  mitigationOrConstraintRefs?
  acceptedResidualRefs?
  admissibleUse
  inadmissibleOverread
  strongerSourceReturnCondition

The frame is not a universal ethics owner. It is the local audit object used when bias, fairness, impact, or ethical assurance is current.

#Bias and Fairness Recognition

#Optional Audit Records And Depth

D.5 may use a compact BiasRegister@Context when the live need is to keep concerns visible during ordinary work:

BiasRegister@Context:
  auditedEntityOfConcernRef
  intendedUseRef
  affectedPeopleOrGroupRefs?
  biasConcernCode
  evidenceRefs
  mitigationOrConstraintRef?
  acceptedResidualRef?
  nextReviewTrigger?

Use a fuller BiasAuditReport@Context only when the object is being released, relied on by other work, exposed to affected people or groups, used for assurance, or used after a material source-currentness, population, context, model, metric, or policy change. The report is a Description episteme or publication-use object; it does not make the audited object fair by existing.

Lightweight scan is enough when the intended use is local, reversible, low-impact, and the scan finds no affected group, proxy, metric, representation, causal-use, or publication-use concern. Deeper review is required when the use is consequential, repeated, automated, cross-context, externally published, safety-relevant, regulatorily or deontically constrained, or when an affected group, missing group, proxy variable, threshold, causal fairness claim, accepted residual, or assurance claim is current.

#Compact Bias Concern Taxonomy

The codes are only concern locators. They do not replace the governed object, affected people or groups, intended use, evidence, mitigation, or accepted residual.

#Causal Fairness Boundary

A fairness claim can be associative, interventional, or counterfactual. D.5 records the ethical-audit use of that claim, but C.28 owns the causal-use question, causality-ladder rung, estimand, identification, realizability, evidence design, CausalEvidenceSupportBasis, and causal-use verdict.

Metric-only fallback: if only metric disparity is claimed and no causal fairness use is made, record it as metric or evaluation use. Do not add causal-fairness machinery by vocabulary alone.

Fairness escalation rule: an interventional-action proxy may admit bounded interventional fairness use, but it cannot be published as counterfactual fairness without the needed C.28 evidence value and verdict.

#Ethical Assurance Boundary

Ethical assurance is not a stamp of moral permission. It is an assurance claim that bias, fairness, impact, and accepted residuals have been examined for the current use.

Use B.3 for the assurance relation. Use A.10 for evidence provenance and source currentness. Use D.3 and D.4 when the audit exposes an interlevel ethical conflict. Use C.30.ILC when the issue is an architecture residual rather than a bias or fairness audit.

#Archetypal Grounding (Worked Slice)

A hiring-screening model has high aggregate accuracy and an internal note says it is "fair." D.5 first asks what fairness claim is being made. If the claim is only a metric disparity comparison, the audit records the metric, affected groups, intended use, missing evidence, and admissible use. If the team claims the model would have prevented unfair outcomes under an intervention or counterfactual, C.28 must supply the causal-use evidence value and verdict before D.5 can treat the fairness claim as admissible for that ethical-audit use. If the audit exposes a conflict between company efficiency and applicant harm across declared scopes, D.3 maps that conflict and D.4 governs decision use.

#Bias-Annotation

#Conformance Checklist

#Common Anti-Patterns and How to Avoid Them

#Consequences

This pattern keeps bias, fairness, impact, causal-fairness audit consumption, and ethical assurance from being scattered across technical patterns. It also prevents D.5 from swallowing all ethics. The cost is that teams must say which bias or fairness claim they are making. The gain is that ethical assurance becomes a typed assurance or evidence claim rather than a comforting label.

#Rationale

D.5 exists because bias, fairness, human-impact, causal-fairness audit consumption, and ethical assurance often survive ordinary technical checks. It keeps those concerns in one audit frame while preserving direct owners: metrics and measurement remain with measurement patterns, causal fairness remains with causal-use patterns, assurance remains an assurance relation, and multilevel ethical conflict remains with D.2 through D.4.

Audit record depth is selected by use, reliance, exposure, source currentness, and residual risk. A compact register is enough for local low-impact use when no live concern remains; a fuller report is required when release, reliance, affected people or groups, source-currentness change, causal fairness, accepted residual, or assurance use is current.

#SoTA-Echoing

#Relations

• Builds on D.1 and coordinates with D.2, D.3, and D.4 for value frame, multilevel entry, conflict structure, and mediation or decision use.
• Coordinates with A.10 for evidence and source currentness.
• Coordinates with B.3 for assurance relation and reliance.
• Coordinates with C.16 for metric and measurement construction.
• Coordinates with C.28 for causal fairness and causal-use evidence value.
• Coordinates with E.17 when publication or publication-use relation changes admissible use.

#D.5:End

Last Updated: 2026-07-03 — upstream FPF commit f7c7e93f (github.com/ailev/FPF)