serial killer
Enable an AI agent to assess whether evidence supports classifying a person as a serial killer under an explicit definition, track uncertainty, and govern responsible use of that classification.
Research draft, second pass
A second pass drafted this model: the structure a model of this thing needs, and what is known about it in the world. The line under this one says how the second half was obtained - researched against sources, or recalled without web access, in which case nothing here was read anywhere and every claim is a lead to verify. Unreviewed either way.
recalled by Codex without web access - no source was read
Researched by: Codex
Purpose and description
Enable an AI agent to assess whether evidence supports classifying a person as a serial killer under an explicit definition, track uncertainty, and govern responsible use of that classification.
A serial killer is a person who unlawfully kills multiple victims in separate events, commonly operationalized as two or more victims, although definitions differ over the minimum count and separation between events.
It can be Evaluate whether attributed events satisfy an explicitly selected definition.; Compare classification outcomes under alternative definitions without merging their criteria.; Trace qualifying counts to individual events and the sources supporting their attribution.; Identify disputed links, uncertain dates, and missing evidence that prevent a supported conclusion.; Generate a qualified account separating allegations, evidence assessments, and adjudicated findings.; Revise or withdraw the classification when event attribution or the governing definition changes..
Distinguishing features
Classification requires a stated threshold involving multiple victims across separate events; a single event with multiple victims does not establish this model's defining pattern.
Event separation must be evaluated under the selected definition, rather than inferred from victim count alone.
A person must be linked to the qualifying events by assessed evidence; similar offenses or media speculation alone do not establish attribution.
The classification concerns a pattern of attributed killings, not a job, an employment relationship, a diagnosis, or a personality type.
A suspected classification remains distinguishable from one supported by adjudicated findings, and neither status silently substitutes for the other.
Scope
+ The operational definition and counting rules used to classify a person
+ Links between a person and distinct killings or killing events, with attribution status
+ Temporal separation and boundaries between events in an alleged or established series
+ Evidence supporting or challenging the inclusion of each event
+ Differences between research classification, investigative suspicion, public labeling, and adjudicated findings
+ Conditions for assigning, qualifying, revising, or withdrawing the classification
- The person's complete biography, identity record, or employment history
- Detailed reconstruction of individual homicides and their forensic evidence
- Clinical diagnoses, treatment plans, or inferred psychiatric conditions
- Criminal charges, sentencing rules, and procedural law owned by jurisdiction-specific legal models
- Victims' complete personal records and support-service case management
- Operational methods for committing violence, selecting victims, or evading investigation
Characteristics
- Operational definition
- Named source and version specifying victim threshold, event separation, and exclusions The same evidence may produce different classifications under different definitions.
- Qualifying victim count
- People; record established and disputed counts separately under the selected definition An inflated or inconsistently counted total can change whether the classification applies.
- Qualifying event count
- Distinct killing events, with unresolved event boundaries identified Multiple victims do not necessarily imply multiple separate events.
- Inter-event intervals
- Elapsed time between events in hours or days, including uncertainty ranges or unknown values Temporal evidence supports evaluation of separation without assuming an unobserved psychological reset.
- Per-event attribution status
- Unassessed, alleged, supported, contested, adjudicated with outcome, or excluded The overall classification must remain traceable to the strength and status of individual event links.
- Participation in each event
- Alleged direct participation, alleged joint participation, other alleged involvement, established participation with stated basis, or unresolved Association with an offender or group does not by itself establish personal responsibility for a qualifying killing.
- Classification assessment
- Not assessed, insufficient evidence, criteria met under named definition, disputed, or withdrawn The label needs an explicit, revisable assessment rather than a permanent unsupported assertion.
- Assessment date
- Calendar date of the latest evidence and definition review Later evidence, corrected attribution, or changed adjudication may require reassessment.
Also called
Where this came from
wikidata · CC0 1.0
Drafted structure
Bundle to layer to finding to question, as the second pass will find it: 5 bundles · 9 layers · 14 findings · 22 questions.
Classification boundaries Defines the precise sense of serial killer used and the conditions required for its application.
A familiar label cannot support reliable recognition unless its threshold and exclusions are explicit.
Operational criteria
Identifies the definition governing victim counts and separate events.
Definition-dependent threshold
The model must record the selected definition rather than silently assume a universal victim threshold or separation rule.
- Which source and version define the minimum victim count and separate-event requirement for this assessment? definition
- Which exclusions or participation requirements does that definition impose? boundary
Neighboring classifications
Separates the classification from adjacent killing patterns and unrelated person categories.
Pattern and role boundaries
The assessment must distinguish separate-event killing from a multiple-victim single event and must not treat the label as an occupation or clinical diagnosis.
- What evidence establishes more than one event rather than multiple victims within one event? boundary
- Does the record improperly use employment, a diagnosis, or a personality description as sufficient evidence for the classification? boundary
Series composition Establishes which victims and events belong to the assessed series and how they are counted.
Classification depends on defensible event membership, not a headline total or an undifferentiated list of cases.
Victims and events
Separates victim identities, event identities, and uncertain inclusion.
Auditable series counts
Victim and event totals must be reconstructible from individual records, with duplicates and disputed inclusions handled explicitly.
- How many distinct victims and events qualify when only the stated evidence threshold is applied? measurement
- Which reported victims or events are duplicates, disputed, or excluded, and on what basis? provenance
Temporal separation
Records dates, uncertainty, and the basis for treating episodes as separate.
Observable event separation
Event separation should be assessed from documented chronology and the selected definition, without inventing emotional states between killings.
- What are the known or bounded dates of each event, and what intervals follow from them? measurement
- Would plausible alternative dates or event boundaries change whether the series meets the selected definition? boundary
Attribution and evidence Connects each proposed killing to the person while preserving uncertainty, source dependence, and contrary evidence.
A correctly counted series still cannot justify classifying a person unless their involvement is supported event by event.
Person-event links
Assesses the evidence and participation basis for each attributed event.
Individualized attribution
Each event requires its own attribution assessment; resemblance to other events, association with another offender, and a reported confession require evaluation rather than automatic acceptance.
- What evidence links this person to this event, and what involvement does that evidence support? provenance
- Which links depend primarily on similarities, association, or a confession whose corroboration remains unresolved? provenance
Corroboration and challenges
Distinguishes independent support from repeated claims and records evidence against attribution.
Source dependence and counterevidence
Repeated reporting of one allegation must remain distinguishable from independent corroboration, and conflicting evidence must stay visible.
- Which sources supply independent evidence, and which repeat a common originating claim? provenance
- What contrary evidence, retractions, or corrected identifications challenge inclusion of each event? provenance
Assessment and responsible use Controls how the classification is expressed, reviewed, and propagated.
An agent must preserve distinctions between suspicion and established findings and respond to corrections without perpetuating unsupported labels.
Classification status
Separates the model's assessment from investigative, judicial, and public descriptions.
Qualified label use
Any use of the classification must identify its definition, evidential basis, and unresolved allegations; judicial outcomes remain separately recorded.
- Who assigned the label, under which definition, and on what evidential basis? provenance
- What qualification must accompany the label so that disputed attribution is not presented as an established finding? action
Revision and propagation
Defines when changed evidence requires reassessment and correction of dependent outputs.
Classification reassessment
Removing an event, changing an attribution assessment, or selecting a different definition must trigger reconsideration of the overall classification.
- Which evidence changes or definition changes would reduce the qualifying series below the applicable threshold? boundary
- Which linked records or generated accounts require correction if the classification is qualified or withdrawn? action
Evidence and external alignment What the world already says about this thing, gathered so the model can be checked against it.
A model that cannot be lined up against existing standards, identifiers and practice cannot be adopted by anyone who already uses them.
Reported evidence
Findings from the breadth pass, kept separate from the structural claims.
Check these first
Recalled without web access and unsourced; every item is a lead to verify.
- This describes a person classified by repeated criminal conduct, not an occupation, qualification or psychiatric diagnosis.
- Check the definition used by the relevant research dataset or investigative body before comparing counts; a cooling-off period is not required by every definition.
- Recalled knowledge only: no sources were consulted, and motivational typologies are omitted because their reliability and boundaries are disputed.
- Which of these check these first hold for the sense of serial killer this model covers, and on what evidence? provenance
Real-world use
Recalled without web access and unsourced; every item is a lead to verify.
- Criminological classification of offenders and repeated homicide.
- Investigation of potentially linked homicide cases.
- Research on victimization patterns, offending histories and investigative outcomes.
- Which of these real-world use hold for the sense of serial killer this model covers, and on what evidence? provenance
Typical measurements
Recalled without web access and unsourced; every item is a lead to verify.
- Number of victims attributed to an offender - Two or more under a common definition; some definitions require at least three - people
- Time between separate killing events - No universally accepted numerical threshold - elapsed time
- Which of these typical measurements hold for the sense of serial killer this model covers, and on what evidence? provenance
Failure modes and hazards
Recalled without web access and unsourced; every item is a lead to verify.
- Further homicide and lasting harm to survivors, victims' families and communities.
- Incorrectly linking unrelated deaths or failing to connect related cases.
- Treating suspected, confessed and independently established victim counts as equivalent.
- Using stereotyped offender profiles as substitutes for evidence.
- Sensational coverage that glorifies offenders or marginalizes victims.
- Which of these failure modes and hazards hold for the sense of serial killer this model covers, and on what evidence? provenance
Regional variation
Recalled without web access and unsourced; every item is a lead to verify.
- Researchers and investigative bodies use different victim-count thresholds and event-separation criteria.
- Differences in death investigation, missing-person reporting and record sharing affect identification and comparison of cases.
- Which of these regional variation hold for the sense of serial killer this model covers, and on what evidence? provenance
Neighbouring kinds and how to tell them apart
Recalled without web access and unsourced; every item is a lead to verify.
- homicide offender - The broader category does not require multiple victims killed in separate events.
- mass murderer - Mass murder concerns multiple victims within a single event; serial killing requires separate events.
- spree killer - Traditionally denotes multiple killings in a continuous sequence without an intervening cooling-off period; its boundary with serial killing is contested.
- contract killer - Defined by killing for payment or another contracted reward; a contract killer may also meet a serial-killing definition.
- psychopathy - A psychological construct, not a repeated-homicide category; neither label establishes the other.
- Which of these neighbouring kinds and how to tell them apart hold for the sense of serial killer this model covers, and on what evidence? provenance
What the second pass must settle
- Which authoritative operational definition should govern this registry entry, and how should alternative victim thresholds be represented?
- How do relevant definitions distinguish serial killing from spree killing, and do they require temporal separation, a cooling-off concept, or other criteria?
- How should killings involving joint offenders, contracted violence, organized crime, or armed conflict be included or excluded under the selected definition?
- What evidence threshold should permit an unqualified classification when some qualifying events have not been adjudicated?
- Does an existing Vercy world model already own this classification, requiring this registry entry to link to it rather than establish a separate model?