HDCM-D — Harm Determination Criteria Module
OOF® Origin Open Foundation®
Independent Methodological Authority
Canonical Definition
The HDCM-D — Harm Determination Criteria Module governs how explicitcriteria, thresholds, conditions, sufficiency requirements, applicability
rules, uncertainty tolerances, exception conditions, and decision
boundaries are established and maintained for determining whether a
recognized and bounded Harm can legitimately be formally established.
HDCM-D establishes the transition:
Governed Harm Determination Input Set → Determination Purpose & Scope →
Applicable Criteria → Determination Thresholds → Criteria Structure →
Governed Harm Determination Criteria Set
Its central question is:
Against what explicit and governed criteria must the qualified Harm
Determination inputs be evaluated in order to determine whether Harm has
been established?
- Operational Role
HDIM establishes what information is fit to enter Determination.
HDCM-D establishes the rules against which that information will be
evaluated.
Its operational role is:
to prevent Harm Determination from becoming an intuitive, retrospective,
outcome-driven, inconsistent, or undocumented judgment by requiring the
basis for Determination to be explicit before the final conclusion is
reached.
HDCM-D does not decide whether Harm has been established.
It establishes:
what must be satisfied for that decision to become legitimate.
- Position Within HDS
The HDS internal flow is:
Qualified Inputs → Determination Criteria → Evaluation → Determination
State → Governed Harm Record
The Core Module sequence is:
HDIM → HDCM-D → HDEM → HDSM → HDRM
HDCM-D receives:
Governed Harm Determination Input Set
and produces:Governed Harm Determination Criteria Set
for HDEM.- Governed Object
The primary governed object is:
Harm Determination Criterion
A complete governed structure may contain:Required Criterion
Conditional Criterion
Supporting Criterion
Exclusion Criterion
Threshold Criterion
Composite Criterion
Domain-Specific Criterion
Uncertainty Criterion
Exception Criterion
Insufficient-Basis Criterion
Non-Establishment Criterion
Each criterion must have an explicit role within the Determinationmethodology.
- Foundational Criteria Principle
The foundational rule is:
Criteria must govern the conclusion; the conclusion must not reconstruct
the criteria.
Therefore:
Desired Conclusion → Retrospective Criteria
is not legitimate Determination governance.
The legitimate sequence is:
Criteria → Evaluation → Determination
- Determination Purpose
Criteria must correspond to the specific Determination being performed.
HDCM-D must establish:
what Harm object is being determined
what Determination question is being answered
what Harm Boundary applies
what scope is included
and:
what the resulting Determination is intended to establish
Without a defined purpose, criteria may be applied outside their
legitimate scope.
- Determination Scope
The criteria must remain aligned with the:
subject scope
temporal scope
effect scope
downstream scope
domain scope
and:
Harm object
established through HBS.Canonical principle:
Determination Criteria must not silently redefine the Harm Boundary.
- Criterion Definition
Each material criterion should specify, where applicable:
criterion object
required condition
applicable input
threshold
scope
method of evaluation
uncertainty treatment
exceptions
and:
effect on Determination
A criterion that cannot be meaningfully applied should not be represented
as a formal Determination criterion.
- Required Criteria
A Required Criterion must be satisfied for a specified Determination State
to be reached.
Failure to satisfy a Required Criterion may result in:
Pending
Insufficient Basis
Partially Established
or:Not Established
depending on why the criterion is unsatisfied.Therefore:
Unsatisfied Criterion ≠ Not Established by Default
- Conditional Criteria
Some criteria apply only where particular conditions exist.
Examples may include:
specific Harm domains
particular subject types
distributed operations
delayed Harm
collective Harm
physical systems
or:
cross-domain effects
Conditional criteria must state their activation conditions.
- Supporting Criteria
Supporting Criteria may strengthen Determination without being
independently mandatory.
Their contribution must remain distinguishable from Required Criteria.
Therefore:
Supporting Criterion ≠ Required Criterion
- Exclusion Criteria
HDCM-D may define conditions demonstrating that a particular Harm
proposition should not be established.
Exclusion Criteria must not be confused with missing support.
Canonical distinction:
Evidence Supporting Exclusion ≠ Absence of Supporting Evidence
- Determination Threshold
A Determination Threshold establishes how much governed support is
required for the applicable Determination conclusion.
Thresholds may differ depending on:
Determination purpose
Harm domain
criterion typeavailable methodology
required confidence
or:
material uncertainty
But they must remain explicit and defensible.
- Threshold ≠ Certainty
A threshold does not necessarily require complete certainty.
Canonical principle:
A legitimate Determination threshold governs uncertainty; it does not
require uncertainty to disappear.
- Emergence Threshold ≠ Determination Threshold
HEMS governs when Harm becomes real.
HDCM-D governs when the available governed basis becomes sufficient for
formal Determination.
Therefore:
Harm Emergence Threshold ≠ Harm Determination Threshold
A later Determination threshold must never be used to redefine when Harm
actually began.
- Recognition Threshold ≠ Determination Threshold
HRS may legitimately recognize Harm before HDS can formally establish it.
Therefore:
Recognition Confirmation Threshold ≠ Harm Determination Threshold
Recognition is intentionally capable of occurring under a different
governance requirement.
- Determination Threshold ≠ Legal Threshold
Legal systems may use standards such as jurisdiction-specific evidentiary
or liability thresholds.
HDS is a methodological Harm architecture.
Therefore:
Harm Determination Threshold ≠ Legal Liability Threshold
and:
Harm Determination Threshold ≠ Claim Eligibility Threshold
- Domain-Neutral Core Criteria
HDCM-D should maintain a common methodological core across Harm domains.
At minimum, Determination criteria should be capable of addressing:
whether the Harm object is sufficiently identified
whether the relevant Harm Boundary is established
whether Harm existence is materially supported
whether the supporting basis applies to the bounded Harm
whether material contradictions are governed
whether uncertainty remains within acceptable Determination limits
and:
whether the conclusion can be represented at the claimed scope
- Domain-Specific Criteria
Different Harm domains may require additional criteria.
Examples include:
medical measurement
physical impairment assessment
digital integrity state
environmental measurement
financial consequence measurement
cognitive effect assessment
infrastructure degradation
Domain-specific criteria may supplement the HAGA™ core.
They must not replace its methodological structure.
- Technology Neutrality
Determination Criteria must remain applicable regardless of whether Harm
arose through:
human action
AI behavior
autonomous executionindustrial process
physical interaction
software
infrastructure
environmental process
or:
composite operations
Canonical principle:
Technology may change the evidence and method; it does not remove the
requirement for governed Harm Determination criteria.
- Discrete Harm Criteria
Discrete Harm may arise through an identifiable transition or occurrence.
Criteria may therefore evaluate a relatively bounded:
event
effect
subject
and:
time
structure.
But HDCM-D must not assume all Harm is discrete.
- Continuous Harm Criteria
Continuous Harm may persist through an ongoing condition.
Criteria must therefore support Harm that cannot be reduced to one moment
or event.
Canonical distinction:
No Discrete Incident ≠ No Determinable Harm
- Accumulated Harm Criteria
Harm may emerge through cumulative exposure or repeated individually
insufficient conditions.
Criteria must support:
accumulation
threshold formation
progressive development
and:
aggregate effect
Therefore:
No Individually Sufficient Contribution ≠ No Determinable Accumulated Harm
- Composite Harm Criteria
Several inputs, actions, conditions, or effects may collectively establish
Harm.
Conceptually:
A + B + C → Determination Basis
where:
A ≠ sufficient
B ≠ sufficient
C ≠ sufficient
individually.
HDCM-D must permit criteria to operate at the composite level.
Canonical principle:
Individual Insufficiency ≠ Composite Insufficiency
- Distributed Harm Criteria
Harm may span:
agents
models
tools
systems
organizations
subjects
locations
or:
time periods
Criteria must be capable of evaluating the complete Harm object rather
than demanding that one local component independently satisfy the entire
Determination threshold.
Therefore:
Local Insufficiency ≠ Global Insufficiency
- Collective Harm Criteria
A Harm may exist at a group, organizational, environmental,
infrastructure, population, or societal level.
Criteria must not require the complete Harm to be reducible to identical
individually established effects.
Canonical principle:
Collective Harm ≠ Sum of Identical Individual Harm Determinations
- Non-Linguistic Harm Criteria
Determination criteria must not require a linguistic trace where Harm
arises through autonomous or physical operation.
A system may act through:
Perception → Interpretation → Decision → Authorization → Action
without producing a textual explanation.
Therefore:
No Linguistic Reasoning Trace ≠ No Determinable Harm
- Criteria for Harm Existence vs Harm Characteristics
HDCM-D must distinguish criteria required to establish:
that Harm exists
from criteria used to establish characteristics such as:
exact onset
complete subject set
full effect extent
severity
persistence
or:
downstream scope
Failure to establish every characteristic must not automatically prevent
establishment of Harm itself.
- Partial Establishment Criteria
Criteria must support:
Partially Established
where some material dimensions satisfy Determination requirements whileothers remain unresolved.
The criteria must specify:
what is established
what is unresolved
and:
whether the unresolved dimension prevents broader Determination
Canonical principle:
Partial Knowledge Does Not Require a False Binary Conclusion.
- Pending Criteria
A Determination may remain Pending where the criteria cannot yet be
legitimately evaluated because:
required inputs are pending
validation is pending
material contradiction remains unresolved
boundary reassessment is underway
or:
a required methodology has not yet been completed
Pending must be governed as an active state, not disguised as
non-establishment.
- Insufficient-Basis Criteria
HDCM-D must define when the available basis is insufficient for
Determination.
Insufficient Basis may arise from:
missing required inputs
unacceptable uncertainty
insufficient scope coverage
unresolved material contradiction
unfit evidence
or:
inability to satisfy a required criterion
Canonical distinction:
Insufficient Basis ≠ No Harm
- Non-Establishment Criteria
Not Established requires a governed basis for concluding that the Harm
proposition has not been established under the applicable methodology.
HDCM-D must preserve the difference between:
lack of sufficient support
and:
material support for non-establishment
Therefore:
Failure to Prove ≠ Proof of Non-Existence
- Uncertainty Criteria
Criteria should specify how uncertainty affects Determination.
Uncertainty may concern:
existence
scope
time
subject
effect
relationship
measurement
provenance
or:
method
HDCM-D must avoid a single undifferentiated uncertainty value where
different uncertainties have materially different consequences.
- Uncertainty Tolerance
A Determination methodology may permit some uncertainty while prohibiting
other uncertainty.
For example, Harm existence may be sufficiently established while exact
onset remains uncertain.
Therefore:
Uncertainty in One Dimension ≠ Uncertainty in the Entire Harm Object
- Confidence Criteria
Where confidence is used, HDCM-D must define:
what confidence refers to
how it is derived
how it affects Determination
and:
what it does not establish
Canonical distinction:
Determination Confidence ≠ Harm Severity
- Contradiction Criteria
HDCM-D must define how material contradictions affect Determination.
A contradiction may:
prevent Determination
reduce confidence
limit scope
produce Partial Establishment
require further evaluation
or:
have no material effect
depending on its relevance and strength.
Therefore:
Contradiction ≠ Automatic Determination Failure
- Missing Information Criteria
Missing information should affect Determination only according to its
methodological significance.
Canonical principle:
Missing information matters because of what the methodology requires, not
merely because something is missing.
This prevents unlimited information demands.
- Evidence Absence Criteria
Criteria must preserve:
Absence of Evidence
versus:Evidence of Absence
The former may create Insufficient Basis.The latter may support Not Established.
They are not interchangeable.
- Corroboration Criteria
Where corroboration strengthens Determination, criteria should
distinguish:
independent corroboration
from:
repetition of the same underlying source
Therefore:
Repeated Information ≠ Independent Support
- Source Neutrality
Criteria must not establish different truth requirements solely because
information comes from:
a subject
an organization
an authority
an AI system
a sensor
or:
an expert
Source characteristics may affect qualification.
They must not replace methodological evaluation.
- Criteria Consistency
Comparable Harm Determinations should use comparable criteria unless a
documented reason justifies variation.
Canonical principle:
Consistency means consistent methodology, not forced identical outcomes.
- Criteria Versioning
Determination methodologies may evolve.
HDCM-D must preserve:
criteria version
effective period
changes
reason for change
and:
applicability
A later criteria version must not silently rewrite an earlier
Determination.
- Criteria Change During Active Determination
If criteria materially change while Determination is underway, governance
must establish:
which criteria apply
whether reassessment is required
whether previous evaluation remains valid
and:
how the transition is recorded
Canonical principle:
Criteria Change ≠ Silent Methodology Replacement
- Criteria Independence
Criteria must not be manipulated because a particular conclusion would
create:
financial exposure
reputational consequences
legal consequences
insurance consequences
regulatory consequences
organizational inconvenience
or:
political pressure
Canonical principle:
The consequences of Determination must not become Determination Criteria.
- Outcome-Neutral Criteria
HDCM-D must protect against:
criteria tightening to avoid Harm
and:
criteria weakening to establish Harm
The methodology should be selected because it is appropriate to the Harm
Determination question.
Not because it produces a preferred result.
- Severity-Neutral Criteria
Greater potential severity may justify urgency or stronger governance
response.
It does not automatically lower the criteria required to formally
establish Harm.
Therefore:
Severity ≠ Determination Sufficiency
- Visibility-Neutral Criteria
Highly visible Harm and difficult-to-observe Harm must both remain
methodologically determinable.
Therefore:
Visibility ≠ Determination Threshold
- Accountability-Neutral Criteria
Criteria must not depend on whether an accountable actor can be
identified.
Therefore:
Unknown Accountability ≠ Undeterminable Harm
AGA™ governs accountability separately.
- Liability-Neutral Criteria
Harm Determination Criteria must not be constructed around whether
liability can later be imposed.
Therefore:
No Liability Path ≠ No Determinable Harm
- Claim-Neutral Criteria
The existence, scope, or admissibility of a claim must not define the Harm
Determination methodology.
Therefore:
Claim Eligibility ≠ Harm Determination Criterion
- Value-Neutral Criteria
Financial or other Value change may support particular Harm analyses.
But:
Value Loss ≠ Universal Harm Determination Criterion
and:
No Measurable Value Loss ≠ No Harm
VFM® remains separate.
- Operational Failure Criteria
Operational failure may be relevant to some Harm Determinations.
But it cannot become a universal criterion.
Therefore:
Operational Failure ≠ Required Condition for Harm
Harm may occur even where the producing system operated as designed.
- Governance Violation Criteria
A violation of AI, autonomous system, organizational, or other governance
rules may be relevant.
But:
Governance Violation ≠ Harm by Default
and:
No Governance Violation ≠ No Harm
HDS determines Harm, not rule violation.
- Validation Dependency
Where a criterion depends on validated information, the required
validation state should be explicit.
But:
Validation ≠ Harm Determination
VALIDOS™ governs whether information or methods satisfy validation
requirements.
HDCM-D governs what those validated inputs must establish for Harm
Determination.
- Integrity Dependency
Where integrity is material, criteria may require an acceptable integrity
state.
But:
Integrity ≠ Harm Determination
INTEGROS® remains the governing architecture for integrity.
- Evidence Dependency
Criteria may specify evidence requirements.
ADIT® governs the evidence itself.
HDCM-D governs:
what evidentiary support is required for the Harm Determination question.
This preserves the architecture boundary.
- Operational Reality Dependency
Where Determination requires understanding what actually happened, ORA™
outputs may be required.
But:
Operational Reality Finding ≠ Harm Determination Criterion Satisfaction by
Default
The criterion must explicitly state how the operational finding relates to
Harm.
- Simulation and Prediction Boundaries
Simulations and predictions may inform criteria application.
But:
Simulated Harm ≠ Actual Harm
and:
Predicted Harm ≠ Existing Harm
Criteria for actual Harm Determination must require appropriate
actual-world basis.
- Criteria Applicability Record
HDCM-D should preserve which criteria:
apply
do not apply
apply conditionally
cannot yet be evaluated
or:
have been superseded
This prevents silent omission of inconvenient criteria.
- Criteria Conflict
Two criteria may create conflicting implications.
HDCM-D must preserve and resolve methodological conflicts before final
evaluation where possible.
A criteria framework that contains unresolved internal contradiction
should not silently produce a Determination conclusion.
- Criteria Hierarchy
Where criteria differ in importance, their relationship should be
explicit.
Possible structures include:
mandatory
conditional
supporting
exclusionary
or:
supplementary
A hierarchy must not be invented after evaluation merely to justify the
outcome.
- Criteria Completeness
The Criteria Set should cover the material dimensions necessary for the
intended Determination.
Completeness does not mean creating criteria for every imaginable fact.
It means the methodology is sufficient to answer the defined Determination
question.
- Criteria Minimality
Criteria must not impose requirements unrelated to legitimate Harm
Determination.
Canonical principle:
A Determination methodology should require everything necessary — and
nothing merely convenient for preventing Determination.
This protects against impossible or artificially burdensome thresholds.
- Determination Readiness
HDCM-D may classify the Criteria Set as:
Ready
Ready with Conditions
Incomplete
Conflicted
or:Not Applicable
Criteria readiness concerns the methodology.It does not determine Harm.
- Handoff to HDEM
HDCM-D produces:
Governed Harm Determination Criteria Set
HDEM then asks:When the qualified Determination inputs are evaluated against these
criteria, what does the governed basis establish?
Canonical distinctions:
Criteria ≠ Evaluation
Criteria Satisfaction ≠ Determination State Until Governed Evaluation Is
Completed
- Minimum Implementation Framework
A minimum HDCM-D implementation requires eight capabilities.
- Determination Purpose & Scope Definition
Define the Harm object, Determination question, applicable Harm Boundary,
and scope to which the criteria apply.
- Criterion Definition & Classification
Represent Required, Conditional, Supporting, Exclusion, Threshold,
Composite, Domain-Specific, Uncertainty, and other materially necessary
criteria.
- Threshold Governance
Define explicit Determination thresholds while preserving separation from
Emergence, Recognition, legal, liability, claim, and other external
thresholds.
- Composite & Distributed Criteria
Support accumulated, continuous, collective, composite, distributed, and
non-linguistic Harm without requiring one locally sufficient action or
input.
- Uncertainty, Contradiction & Insufficient-Basis Governance
Define how uncertainty, missing information, conflicting inputs, evidence
absence, and incomplete support affect Determination.
- Criteria Independence & Bias Controls
Protect criteria from outcome preference, severity, visibility, authority,
accountability, liability, claim, Value, financial, organizational,
political, or reputational pressure.
- Criteria Lifecycle & Historical Continuity
Preserve applicability, versioning, changes, supersession, conflicts, and
historical criteria states.
- Governed Output
Produce a traceable Governed Harm Determination Criteria Set ready for
HDEM evaluation.
A system that chooses, modifies, or interprets its Harm Determination
criteria only after seeing which conclusion they produce does not fully
implement HDCM-D.
- Use Case 1 — Harm Exists but Exact Onset Remains Uncertain
A recognized and bounded Harm has strong support across subject, effect,
and operational information.
The exact day on which the Harm began cannot be established precisely.
HDCM-D defines criteria that distinguish:
Harm existence
from:exact temporal precision
The methodology may therefore permit Harm to be formally established while
preserving an estimated or bounded temporal start.
The governing principle is:
Uncertainty in one characteristic must not automatically become
uncertainty about the existence of the entire Harm.
- Use Case 2 — Distributed AI Harm Requires Composite Criteria
An AI-mediated operation spans multiple agents, models, tools, and
systems.
No single action independently satisfies the complete Harm Determination
threshold.
HDCM-D defines criteria capable of evaluating:
the composite operational structure
the combined effects
the bounded subjects
and:
the complete qualified input set
rather than demanding one independently sufficient output.
The governing principle is:
Harm that exists at the composite level must be determinable at the
composite level.
- Governance Requirements
HDCM-D requires that:
- criteria be established before the conclusion
- Determination purpose and scope remain explicit
- criteria not silently redefine the Governed Harm Boundary
- Required, Conditional, Supporting, and Exclusion Criteria remain
distinguishable; - thresholds remain explicit and governable; -
Determination thresholds remain separate from Emergence and Recognition
thresholds; - methodological Harm thresholds remain separate from legal,
liability, and claim thresholds; - domain-specific criteria supplement
rather than replace the common HAGA™ methodology; - continuous,
accumulated, composite, distributed, collective, and non-linguistic Harm
remain determinable; - Harm existence criteria remain separable from
criteria for exact Harm characteristics; - Partial, Pending, Insufficient
Basis, and Not Established outcomes remain methodologically supportable; -
uncertainty, contradiction, and missing information have explicit
treatment; - absence of evidence remain distinct from evidence of absence;
- corroboration remain distinct from repeated source information; -
criteria remain protected from authority, severity, visibility,
accountability, liability, claim, Value, and outcome bias; - criteria
changes and versioning preserve historical continuity; - criteria remain
complete enough to answer the Determination question but not artificially
burdensome; - criteria readiness remain distinct from Harm Determination;
- the complete governed Criteria Set be handed to HDEM.
- Governance Logic
HDCM-D applies the following logic:
Has HDIM produced a Governed Harm Determination Input Set?
If no:
→ criteria application is not ready.
If yes:
→ define the Determination purpose and scope.
What must be established for this bounded Harm to reach a legitimate
Determination State?
→ define the applicable criteria.
Are any criteria conditional?
→ establish their activation conditions.
Does the Harm require composite, accumulated, distributed, collective,
continuous, or non-linguistic evaluation?
→ structure criteria at the level at which Harm exists.
What threshold applies?
→ define it explicitly.
Does uncertainty affect only one Harm characteristic or the Harm existence
proposition itself?
→ preserve the distinction.
What does missing or contradictory information mean under the methodology?
→ specify its effect before evaluation.
Would the criteria change because the potential conclusion creates an
undesirable consequence?
If yes:
→ reject the outcome-driven modification.
Are criteria internally complete, consistent, applicable, and ready?
If yes:
→ produce the Governed Harm Determination Criteria Set.
If not:
→ classify the Criteria Set as Conditional, Incomplete, or Conflicted and
resolve the methodological deficiency before final Determination
evaluation.
At every stage:
Criteria → Evaluation → Determination
not:
Conclusion → Criteria
- Governance Outputs
HDCM-D may produce:
- Governed Harm Determination Criteria Set
- Determination Purpose
- Determination Scope
- Required Criterion Set
- Conditional Criterion Set
- Supporting Criterion Set
- Exclusion Criterion Set
- Determination Threshold
- Composite Harm Criteria
- Distributed Harm Criteria
- Collective Harm Criteria
- Continuous Harm Criteria
- Accumulated Harm Criteria
- Domain-Specific Criteria
- Uncertainty Criteria
- Contradiction Criteria
- Missing Information Criteria
- Insufficient-Basis Criteria
- Non-Establishment Criteria
- Partial Establishment Criteria
- Criteria Applicability Record
- Criteria Hierarchy
- Criteria Conflict Record
- Criteria Version
- Criteria Change Record
- Criteria Readiness State
The principal output is:
Governed Harm Determination Criteria Set
- Critical Distinctions
Criteria ≠ Conclusion
Unsatisfied Criterion ≠ Not Established by Default
Supporting Criterion ≠ Required Criterion
Evidence Supporting Exclusion ≠ Absence of Supporting Evidence
Threshold ≠ Certainty
Harm Emergence Threshold ≠ Harm Determination Threshold
Recognition Confirmation Threshold ≠ Harm Determination Threshold
Harm Determination Threshold ≠ Legal Liability Threshold
Harm Determination Threshold ≠ Claim Eligibility Threshold
No Discrete Incident ≠ No Determinable Harm
Individual Insufficiency ≠ Composite Insufficiency
Local Insufficiency ≠ Global Insufficiency
Collective Harm ≠ Sum of Identical Individual Harm Determinations
No Linguistic Reasoning Trace ≠ No Determinable Harm
Insufficient Basis ≠ No Harm
Failure to Prove ≠ Proof of Non-Existence
Uncertainty in One Dimension ≠ Uncertainty in the Entire Harm Object
Determination Confidence ≠ Harm Severity
Contradiction ≠ Automatic Determination Failure
Repeated Information ≠ Independent Support
Severity ≠ Determination Sufficiency
Visibility ≠ Determination Threshold
Unknown Accountability ≠ Undeterminable Harm
No Liability Path ≠ No Determinable Harm
Claim Eligibility ≠ Harm Determination Criterion
Operational Failure ≠ Required Condition for Harm
Governance Violation ≠ Harm by Default
Validation ≠ Harm Determination
Simulated Harm ≠ Actual Harm
Predicted Harm ≠ Existing Harm
Criteria Change ≠ Silent Methodology Replacement
Criteria ≠ Evaluation
- Canonical Boundary
The HDCM-D — Harm Determination Criteria Module governs the definition,
classification, applicability, threshold structure, sufficiency
requirements, uncertainty treatment, contradiction treatment, exception
conditions, domain-specific supplementation, composite and distributed
applicability, independence, consistency, versioning, lifecycle, and
readiness of the criteria used for formal Harm Determination. It produces
the Governed Harm Determination Criteria Set but does not qualify
Determination inputs, perform the Determination evaluation, assign the
final Determination State, create the Governed Harm Record, establish
general causation, assign accountability or liability, or replace
evidence, validation, integrity, operational reality, identity, AI
behavior, autonomous execution, simulation, prediction, Value, claim, or
jurisdiction-specific governance.
- Canonical Closing Statement
A Harm Determination cannot be methodologically legitimate if the rules
for reaching it appear only after the desired conclusion is already known.
HDCM-D establishes those rules before evaluation: what must be shown, how
much support is sufficient, what uncertainty can remain, what
contradictions matter, and where the limits of the conclusion lie.
The standard is not whether a methodology can produce a conclusion. The
standard is whether the same governed methodology would remain legitimate
if the evidence pointed the other way.
Criteria must govern the conclusion — never the conclusion the criteria.