HDIM — Harm Determination Input Module
OOF® Origin Open Foundation®
Independent Methodological Authority
Canonical Definition
The HDIM — Harm Determination Input Module governs how informationrequired for formal Harm Determination is identified, admitted, qualified,
structured, associated, differentiated, challenged, limited, updated, and
prepared as a governed Determination Input Set.
HDIM establishes the transition:
Governed Harm Recognition State + Governed Harm Boundary +
Determination-Relevant Information → Input Identification → Input
Qualification → Input Structuring → Governed Harm Determination Input Set
Its central question is:
What information is legitimately fit to enter formal Harm Determination,
and under what qualifications, limitations, uncertainty, and provenance?
- Operational Role
HDS cannot legitimately determine Harm from an uncontrolled collection of:
reports
observations
records
claims
system outputs
measurements
assumptions
or:
opinions
HDIM establishes the governed input layer before Determination Criteria
are applied.
Its operational role is:
to ensure that Harm Determination begins from an explicit, qualified,
traceable, and appropriately bounded input set rather than from whatever
information happens to be available or convenient.
- 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
HDIM receives:
Governed Harm Recognition State
and:Governed Harm Boundary
It produces:Governed Harm Determination Input Set
for HDCM-D.- Governed Object
The primary governed object is:
Harm Determination Input
Possible input states include:Candidate Input
Admitted Input
Qualified Input
Qualified with Limitations
Corroborative Input
Contradictory Input
Duplicate Input
Superseded Input
Disputed Input
Pending Qualification
Excluded Input
Each material input must retain its governance state.- Foundational Input Principle
The foundational distinction is:
Available Information ≠ Qualified Determination Input
Information does not become suitable for Harm Determination merely because
it:
exists
has been reported
appears in a system
comes from an authority
supports a preferred conclusion
or:
is easy to access
- Input Identification
HDIM identifies information materially relevant to determining the bounded
Harm.
Potential inputs may include:
Recognition outputs
Harm Boundary outputs
subject informationtemporal information
effect information
downstream information
operational records
observations
measurements
subject reports
system records
machine detections
expert findings
validated findings
historical information
contradictory information
Identification means the information should be considered.
It does not mean it should automatically be relied upon.
- Recognition Inputs
HRS provides the:
Governed Harm Recognition State
HDIM preserves the Recognition basis relevant to Determination.However:
Recognized Harm ≠ Determined Harm
Recognition information is an input to Determination, not its conclusion.
- Boundary Inputs
HBS provides the:
Governed Harm Boundary
including relevant:subject
temporal
effect
and:
downstream
dimensions.
HDIM must preserve the boundary against which Harm is being determined.
Canonical principle:
Determination Inputs must correspond to the Harm object being determined.
- Input Relevance
An input must have a materially supportable relationship to the
Determination question.
Information may be:
directly relevant
indirectly relevant
contextually relevant
corroborative
contradictory
or:
irrelevant
Volume alone does not create relevance.
- Input Fitness
A relevant input may still be unfit for a particular Determination
purpose.
Fitness may depend on:
scope
method
timing
subject applicability
domain applicability
measurement conditions
source limitations
validation state
or:
integrity state
Therefore:
Relevant Input ≠ Fit Input by Default
- Source Qualification
HDIM must preserve what kind of source produced the input.
Examples include:
affected subject
witness
operator
organization
sensor
AI system
autonomous systemtechnical record
expert
authority
external institution
Source type may affect qualification.
It must not predetermine truth.
- Authority Status ≠ Input Truth
An input must not become methodologically privileged solely because it
originates from an institution or authority.
Therefore:
Authoritative Source ≠ Automatically Correct Input
Likewise:
Non-Authoritative Source ≠ Automatically Invalid Input
- Subject-Reported Inputs
A harmed or potentially harmed subject may provide important information
about:
effects
timing
experience
function
persistence
or:
change
Subject reports must remain governable without either automatic acceptance
or automatic dismissal.
Canonical distinction:
Subject Report ≠ Harm Determination
- Machine-Generated Inputs
Inputs may be generated through:
sensors
monitoring systems
AI models
automated detectionlogs
analytics
or:
autonomous systems
Machine generation does not automatically establish accuracy.
Therefore:
Machine Output ≠ Qualified Determination Input by Default
- Human-Generated Inputs
Human observations, evaluations, testimony, and expert conclusions may
contain:
error
bias
uncertainty
incomplete perception
interpretation
or:
conflict
Human origin does not automatically establish reliability.
- Direct and Indirect Inputs
Determination may rely on both:
Direct Inputs
and:Indirect Inputs
An indirect input must not be excluded merely because an intermediaterelationship exists.
Canonical distinction:
Indirect Input ≠ Irrelevant Input
Its relationship must instead be represented.
- Historical Inputs
Harm may be determined retrospectively.
Historical inputs may therefore include:
archived records
earlier measurements
past reports
historical system states
later-discovered evidence
reconstructed timelines
Historical age does not automatically eliminate relevance.
- Late-Discovered Information
Information may become available after:
Recognition
boundary establishmentinitial evaluation
or:
earlier Determination
HDIM must support admission of new information without rewriting the
history of what was previously known.
Canonical principle:
Late Discovery ≠ Late Harm
- Temporal Applicability
Every materially time-dependent input should be associated with the period
to which it legitimately applies.
An input generated at T3 may describe:
T1
T2
T3
or:
a range
The timestamp of the record must not automatically become the timestamp of
the Harm condition described.
Therefore:
Record Time ≠ Harm Time by Default
- Subject Applicability
An input concerning Subject A must not automatically be generalized to:
Subject B
a groupan organization
or:
a population
without sufficient basis.
Therefore:
One Subject Input ≠ Universal Subject Input
- Domain Applicability
Information valid in one Harm domain may not automatically establish
another.
For example:
operational degradation
may inform but not automatically establish:
medical Harm
or:
financial Harm
Cross-domain use must remain qualified.
- Scope Qualification
Every material input should have an understood scope.
Scope may include:
subject
time
effect
system
location
domain
operation
downstream relationship
An input must not support a conclusion broader than its legitimate scope.
Canonical principle:
Input Scope Must Bound Input Use
- Input Provenance
Material Determination inputs should preserve sufficient provenance to
understand:
where the input originated
how it was produced
when it was produced
what transformation occurred
which system or actor produced it
and:
whether its origin remains known
OBIDENITY® may govern relevant identity and provenance structures.
HDIM uses those structures for Harm Determination input governance.
- Transformed Inputs
Information may be:
summarized
aggregated
translated
normalized
filtered
inferred
converted
or:
AI-processed
Transformation may change what the information can legitimately support.Canonical principle:
Validated Source ≠ Validated Derivative
A transformed input must preserve its transformation history where
material.
- Derived Inputs
A Derived Input is produced from one or more other inputs.
Conceptually:
Input A + Input B → Derived Input C
The derived input must preserve:
source relationships
derivation method
assumptions
limitations
and:
uncertainty
- Aggregated Inputs
Several inputs may be combined.
Aggregation can reveal patterns invisible at the individual-input level.
But:
Aggregation ≠ Automatic Sufficiency
The quality of the aggregate depends on the structure and legitimacy of
the contributing inputs.
- Composite Determination Inputs
Some Harm can only be represented through multiple individually incomplete
inputs.
Conceptually:
Input A + Input B + Input C → Composite Determination Basis
HDIM must preserve such inputs so HDEM can later evaluate them
collectively.
Canonical principle:
Individual Input Insufficiency ≠ Composite Input Insufficiency
- Distributed Inputs
Determination information may be distributed across:
subjects
agents
systems
organizations
sessions
locations
domains
and:
time periods
Technical fragmentation must not make the complete input basis invisible.
- Evidence Fragmentation
Canonical principle:
Evidence fragmentation must not make a composite operation unauditable or
a materially coherent Harm undeterminable by default.
HDIM must therefore support cross-source association while preserving
provenance.
- Local Input ≠ Complete Reality
A single system may contain only one part of the relevant information.
Therefore:
Local Input Set ≠ Complete Determination Input Set by Default
This is particularly important for distributed AI and autonomous systems.
- Supporting Inputs
A Supporting Input materially supports one or more propositions relevant
to Harm Determination.
Its support must be represented without converting it directly into the
conclusion.
Therefore:
Supporting Input ≠ Determination Conclusion
- Contradictory Inputs
HDIM must actively preserve information that materially conflicts with the
emerging Determination basis.
Contradictory information must not be excluded merely because it makes
Determination more difficult.
Canonical principle:
Contradiction is a governance input, not a data-cleaning inconvenience.
- Counterevidence
Counterevidence materially supports a conclusion inconsistent with another
Determination proposition.
HDIM must preserve:
what it contradicts
its scope
its qualification
its provenance
and:
its limitations
HDEM later evaluates its significance.
- Missing Inputs
HDIM should identify materially expected information that is unavailable.
Possible states include:
Missing
Unavailable
Not Collected
Destroyed
Unknown Whether Existing
Inaccessible
These states must remain distinct where material.- Missing Information ≠ Negative Evidence
Canonical distinction:
Missing Input ≠ Evidence That the Opposite Is True
Missing information may affect Determination sufficiency.
It does not automatically determine the conclusion.
- Evidence Absence vs Evidence of Absence
HDIM must distinguish:
Absence of Evidence
from:Evidence of Absence
This distinction is essential for later HDS evaluation.- Input Uncertainty
Every material input may carry uncertainty concerning:
accuracy
scope
timing
identity
interpretation
measurement
completeness
or:
relationship
Uncertainty should be represented rather than hidden.
- Input Confidence
Where appropriate, HDIM may associate confidence with an input.
Possible states may include:
High
Moderate
Low
Unknown
Disputed
Confidence must remain tied to the specific input dimension beingassessed.
- Confidence ≠ Truth
Canonical distinction:
High Confidence ≠ Guaranteed Truth
and:
Low Confidence ≠ False Information
Confidence informs governance.
It does not replace evaluation.
- Input Limitations
Material limitations should travel with the input.
Examples include:
partial observation
limited sample
measurement range
missing context
uncertain identity
temporal gap
model limitation
transformation loss
source dependency
An input stripped of material limitations may become misleading.
- Duplicate Inputs
The same underlying information may appear in several records.
HDIM must distinguish:
independent corroboration
from:
duplicated representation
Canonical principle:
Signal Replication ≠ Independent Corroboration
- Corroboration
Independent inputs may strengthen the Determination basis where they
materially support the same proposition through sufficiently independent
pathways.
Corroboration must preserve:
independence
scope
relationship
and:
limitations
- Source Dependence
Two apparently independent sources may derive from the same underlying
information.
Conceptually:
Source A → Report B
Source A → System C → Report D
B and D may not constitute independent corroboration.
HDIM must preserve material source dependence.
- Input Conflict
Inputs may conflict regarding:
whether Harm occurred
when it occurred
who was affected
what effect existed
how long it persisted
whether a downstream consequence belongs
HDIM records the conflict.
It does not resolve the final Determination.
- Input Exclusion
An input may be excluded from the Determination Input Set where it is:
irrelevant
materially unfit
outside scope
duplicative without independent value
unusable
or:
methodologically incompatible with the determination purpose
Exclusion must be documented.
- Excluded Input ≠ Deleted Input
Canonical principle:
Exclusion From Determination ≠ Erasure From Governance History
Where material, the existence of the excluded input and the reason for
exclusion should remain traceable.
- Input Bias
HDIM must protect against input selection driven by:
desired conclusion
organizational interest
financial exposure
reputational concern
legal strategy
technical convenience
or:
availability bias
Canonical principle:
Input selection must follow the Determination question, not the preferred
Determination outcome.
- Confirmation Bias
A Determination process must not collect only information supporting the
initial Recognition state.
Recognition creates a governance object.
It does not predetermine Determination.
Therefore:
Recognized Harm ≠ Presumed Determined Harm
- Visibility Bias
Highly visible information must not automatically receive disproportionate
weight merely because it is easy to observe.
Likewise:
Invisible Information ≠ Irrelevant Information
- Authority Bias
Information from powerful or formal institutions must not automatically
override materially stronger contradictory inputs.
HDS requires methodological qualification, not institutional hierarchy
alone.
- AI-Generated Summaries
An AI-generated summary may support Determination workflows.
But the summary may:
omit information
compress uncertainty
merge sources
lose provenance
introduce inference
Therefore:
AI Summary ≠ Original Determination Basis
Where materially relied upon, source traceability must remain available.
- Synthetic and Simulated Inputs
Simulated information may support understanding, testing, or comparison.
But:
Simulated Evidence ≠ Actual Harm Evidence
SIMULOS® governs simulation.
HDIM must clearly classify simulated inputs so they cannot silently become
evidence of actual Harm.
- Predicted Inputs
Predictions may identify future or likely consequences.
But:
Predicted Harm ≠ Existing Harm Input by Default
Prediction must remain distinct from information about actual Harm.
- Operational Reality Inputs
ORA™ may provide information about:
actual execution
system state
operational sequence
failure
constraint
intervention
HDIM may admit these as Determination inputs where relevant.
But:
Operational Reality ≠ Harm Determination
- Evidence Governance Boundary
ADIT® governs:
evidence formation
integrity
correlation
verification
audit
HDIM does not recreate those functions.
It asks:
Is this evidence fit to enter this Harm Determination, and under what
qualification?
- Validation Boundary
VALIDOS™ may establish whether information has been validated for a
particular purpose.
HDIM must preserve the validation state where materially relevant.
But:
Validated Input ≠ Sufficient Harm Determination Basis
Validation establishes input fitness or validity.
HDS establishes Harm.
- Integrity Boundary
INTEGROS® may establish relevant integrity states.
HDIM may use those states in qualification.
But:
Integrity State ≠ Determination Input Conclusion
- Identity Boundary
OBIDENITY® may govern:
origin
identity
ownership
provenance
continuity
HDIM uses those outputs where necessary.
It does not independently establish identity.
- Input Change
An input may later be:
corrected
superseded
invalidated
expanded
reinterpreted
or:
withdrawn
HDIM must update its current governance state without erasing previous
states.
- Input Reassessment
Input qualification must be reassessed where:
new provenance appears
validation changes
integrity concerns arise
scope changes
the Harm Boundary changes
a source is corrected
a transformation is discovered
new contradictory information appears
- Historical Input Integrity
HDIM must preserve:
Input State₁ → New Information → Input State₂
including:
original input
original qualification
limitations
change
new qualification
reason
Canonical principle:
Current Input State ≠ Complete Input History
- Determination Input Set Completeness
HDIM should assess whether the available input set sufficiently covers the
dimensions required for the next Determination stage.
Completeness does not require all imaginable information.
It requires sufficient coverage to apply the relevant Determination
Criteria meaningfully.
- Complete Input Set ≠ Sufficient Determination
Even a complete input set may fail to establish Harm.
Therefore:
Input Completeness ≠ Determination Sufficiency
HDEM later evaluates sufficiency.
- Determination Readiness
HDIM may conclude that the input layer is:
Ready
Ready with Limitations
Conditionally Ready
or:Not Ready
for criteria application and evaluation.Readiness concerns the input layer.
It does not determine Harm.
- Handoff to HDCM-D
HDIM produces:
Governed Harm Determination Input Set
HDCM-D then asks:Against what explicit criteria must these qualified inputs be evaluated to
determine Harm?
The separation is essential:
Input Qualification ≠ Determination Criteria
and:
Input Qualification ≠ Determination Evaluation
- Minimum Implementation Framework
A minimum HDIM implementation requires eight capabilities.
- Input Identification
Identify materially relevant Recognition, Boundary, subject, temporal,
effect, downstream, operational, evidentiary, historical, supporting, and
contradictory information.
- Input Qualification
Represent relevance, fitness, scope, provenance, temporal applicability,
subject applicability, uncertainty, confidence, and limitations.
- Provenance & Transformation Governance
Preserve source, derivation, transformation, aggregation, and dependency
relationships where materially relevant.
- Composite & Distributed Input Governance
Support inputs distributed across subjects, systems, agents, sessions,
organizations, domains, locations, and time without losing their
collective structure.
- Conflict, Missing Information & Duplication Governance
Represent contradictory inputs, counterevidence, missing information,
evidence absence, duplicate records, and independent corroboration
distinctly.
- Input Bias Controls
Prevent preferred outcomes, visibility, authority status, reporting
convenience, organizational interests, or technical availability from
determining the admitted input set.
- Reassessment & Historical Continuity
Support correction, supersession, invalidation, withdrawal,
requalification, and historical preservation.
- Governed Output
Produce a traceable Governed Harm Determination Input Set with explicit
readiness status for the next HDS stage.
A system that simply collects available information and passes it directly
into a Harm conclusion does not fully implement HDIM.
- Use Case 1 — Conflicting Human, Operational, and System
Information
A subject reports persistent Harm.
An operational record supports part of the reported timeline.
A system record appears to contradict another part.
Additional historical information is incomplete.
HDIM does not decide which conclusion is correct.
It qualifies:
the subject report
the operational record
the contradictory system record
the missing historical information
and:
their respective scope, provenance, uncertainty, and limitations
The resulting Governed Harm Determination Input Set allows HDS to evaluate
the complete basis without silently removing inconvenient information.
The governing principle is:
Contradiction must be governed before it can be evaluated.
- Use Case 2 — Distributed AI Harm Produces Fragmented
Determination
Inputs
A Harm-relevant AI operation spans:
Agent A
Model B
Tool C
External System D
and:a physical execution environment
Each system contains only part of the relevant operational record.
No local record independently represents the complete Harm-producing
operation.
HDIM reconstructs the Determination Input Set across those partitions
while preserving:
source identity
execution relationships
transformations
time
limitations
duplicate information
and:
material gaps
The governing principle is:
Fragmented information must not make a materially coherent Harm invisible
to Determination.
- Governance Requirements
HDIM requires that:
- available information remain distinct from qualified
Determination
input; - Recognition and Boundary outputs remain preserved as governed
inputs; - relevance remain distinct from fitness; - source authority not
determine truth; - subject reports and machine outputs not automatically
determine Harm; - direct and indirect inputs remain governable; -
late-discovered information not redefine Harm timing by default; - record
time remain distinct from Harm time; - input scope constrain legitimate
input use; - provenance and material transformations remain traceable; -
derived and aggregated inputs preserve source relationships; - composite
and distributed inputs remain governable; - supporting and contradictory
information both remain visible; - missing information remain distinct
from negative evidence; - absence of evidence remain distinct from
evidence of absence; - uncertainty and limitations travel with material
inputs; - confidence remain distinct from truth; - duplicate
representation remain distinct from independent corroboration; - excluded
inputs remain historically traceable where material; - input selection
remain protected from outcome bias; - simulated and predicted information
remain distinct from evidence of actual Harm; - ADIT®, VALIDOS™,
INTEGROS®, OBIDENITY®, ORA™, and SIMULOS® boundaries remain preserved; -
input changes and reassessment preserve historical integrity; -
Determination readiness remain distinct from Determination itself.
- Governance Logic
HDIM applies the following logic:
Has HRS produced a governed Recognition state and HBS established the Harm
Boundary?
If no:
→ formal Determination Input governance is not ready.
If yes:
→ identify Determination-relevant information.
Is the information relevant to the bounded Harm?
If no:
→ exclude it from the Determination Input Set while preserving the reason
where material.
If yes:
→ assess fitness.
Is the input fit for its intended Determination use?
If yes:
→ qualify it.
If partly:
→ qualify it with limitations.
If unresolved:
→ maintain Pending Qualification.
Does the input conflict with other information?
→ preserve the conflict.
Is the information duplicated?
→ distinguish duplication from independent corroboration.
Is provenance or transformation materially relevant?
→ preserve it.
Is information missing?
→ identify the missing state without converting absence into negative
evidence.
Are several individually incomplete inputs collectively relevant?
→ preserve their composite relationship.
Does the available input set sufficiently support criteria application?
If yes:
→ establish Ready or Ready with Limitations.
If no:
→ establish Not Ready or Conditional Readiness.
Then produce:
Governed Harm Determination Input Set
for HDCM-D.At every stage:
Information ≠ Qualified Input
Recognition ≠ Determination
Relevance ≠ Fitness
Authority ≠ Truth
Duplication ≠ Corroboration
Missing Information ≠ Negative Evidence
Input Completeness ≠ Determination Sufficiency
- Governance Outputs
HDIM may produce:
- Governed Harm Determination Input Set
- Candidate Input Set
- Qualified Input
- Qualified Input with Limitations
- Excluded Input
- Pending Qualification State
- Supporting Input
- Contradictory Input
- Counterevidence State
- Input Relevance State
- Input Fitness State
- Input Scope
- Input Provenance Record
- Input Transformation Record
- Derived Input Record
- Composite Input Structure
- Distributed Input Structure
- Missing Input State
- Input Uncertainty State
- Input Confidence State
- Input Limitation Record
- Duplicate Input State
- Corroboration Relationship
- Source Dependency Record
- Input Conflict State
- Input Reassessment Record
- Historical Input Record
- Determination Input Readiness State
The principal output is:
Governed Harm Determination Input Set
- Critical Distinctions
Available Information ≠ Qualified Determination Input
Recognized Harm ≠ Determined Harm
Relevant Input ≠ Fit Input
Authoritative Source ≠ Automatically Correct Input
Subject Report ≠ Harm Determination
Machine Output ≠ Qualified Determination Input by Default
Indirect Input ≠ Irrelevant Input
Late Discovery ≠ Late Harm
Record Time ≠ Harm Time
One Subject Input ≠ Universal Subject Input
Input Scope Must Bound Input Use
Validated Source ≠ Validated Derivative
Aggregation ≠ Automatic Sufficiency
Individual Input Insufficiency ≠ Composite Input Insufficiency
Local Input Set ≠ Complete Determination Input Set
Supporting Input ≠ Determination Conclusion
Missing Input ≠ Evidence That the Opposite Is True
Absence of Evidence ≠ Evidence of Absence
High Confidence ≠ Guaranteed Truth
Low Confidence ≠ False Information
Signal Replication ≠ Independent Corroboration
Excluded Input ≠ Deleted Input
Recognized Harm ≠ Presumed Determined Harm
AI Summary ≠ Original Determination Basis
Simulated Evidence ≠ Actual Harm Evidence
Predicted Harm ≠ Existing Harm Input by Default
Operational Reality ≠ Harm Determination
Validated Input ≠ Sufficient Harm Determination Basis
Input Completeness ≠ Determination Sufficiency
Input Qualification ≠ Determination Criteria
Input Qualification ≠ Determination Evaluation
- Canonical Boundary
The HDIM — Harm Determination Input Module governs the identification,
admission, qualification, structuring, provenance, transformation, scope,
fitness, uncertainty, limitations, conflict, contradiction, missing
information, duplication, corroboration, composite relationships,
distributed information, reassessment, and historical continuity of
information used for formal Harm Determination. It produces the Governed
Harm Determination Input Set but does not establish Determination
Criteria, evaluate whether the criteria are satisfied, determine the Harm
Determination State, create the final Governed Harm Record, establish
general causation, assign accountability or liability, or replace
evidence, validation, integrity, identity, operational reality, AI
behavior, autonomous execution, simulation, Value, prediction, or
jurisdiction-specific governance.
- Canonical Closing Statement
Harm Determination is only as legitimate as the governed basis from which
it begins.
Information must not become evidence merely because it is available,
become reliable merely because it is authoritative, or disappear merely
because it contradicts the expected conclusion.
HDIM ensures that Determination begins with a qualified representation of
what is known, what is uncertain, what conflicts, what is missing, and
where each material input came from.
Before Harm can be determined, the information used to determine it must
itself become governable.