Evidence Methodology
How DECUR assigns evidence tiers, computes quantitative scores, classifies claims, and maintains editorial independence across its 72-case archive. This page is intended for researchers and journalists who need to understand what the platform asserts and how it justifies those assertions.
1. Evidence Tiers
Every case in the DECUR archive is assigned one of three evidence tiers. The tier reflects the nature of the primary supporting documentation, not editorial opinion about the event's significance. A case can be historically important and still be Tier 3 if the primary record consists of witness accounts without official documentation.
Official Documentation
Government or military records, formally authenticated sensor captures, congressional testimony under oath, or officially released investigation files. The existence of the event is corroborated at the institutional level.
Declassified Records
FOIA-released documents, formally declassified records, or officially acknowledged incidents where the government has confirmed the event occurred but has not necessarily characterized its nature. Includes cases with credentialed witnesses and supporting official documentation.
Credentialed Testimony
Cases resting primarily on accounts from credentialed witnesses (military, government, pilots, scientists) without formal official documentation of the event itself. Credentialing means the witnesses held verifiable roles relevant to what they observed.
Tier does not equal significance. Some of the most physically compelling cases in the archive are Tier 3 because they occurred before modern sensor infrastructure. The tier reflects documentation availability, not event importance. The EQI score provides a finer-grained measure that is not constrained to three buckets.
2. Evidence Quality Index (EQI)
The EQI is a 0-100 integer computed from six weighted components. It measures documentation quality: how well the event is recorded, corroborated, and officially examined. A high EQI means the event is thoroughly documented; it does not mean the event is confirmed anomalous.
EQI is distinct from the evidence tier. A Tier-1 case can score 62 if it has strong multi-sensor capture but a weak physical evidence record. A Tier-3 case may score 58 if it has exceptional physical evidence despite lacking official documentation. When a case's EQI diverges significantly from what the tier implies, the case detail page displays an editorial note explaining the discrepancy.
| Component | Weight | What it measures |
|---|---|---|
| Capture Technology | 25% | What instrumented the observation: multi-platform military sensor arrays at the top; unaided civilian eyewitness at the bottom. Physical trace evidence (GEPAN chemical analysis, radiation readings) scores as an independent capture pathway. |
| Witness Credential | 20% | Credential of the primary witness combined with the formality of their statement. Congressional testimony under oath scores highest; undocumented civilian account scores lowest. Formal declaration to an Inspector General or credentialed insider on-record interview score in between. |
| Corroboration Independence | 15% | Whether and how independently the observation was corroborated. Multi-platform sensor from separate platforms on independent missions scores highest. Multiple witnesses at the same event score lower because their observations are not fully independent. |
| Physical Evidence | 15% | Presence and quality of physical material: lab-verified anomalous samples (soil chemistry, metallurgy), documented radiation readings, recovered material, burn marks with independent analysis. Absence of physical evidence is not penalized in cases where the event type does not produce recoverable traces. |
| Investigation Quality | 15% | Depth of official investigation: multi-agency congressional acknowledgment scores highest; informal unofficial review scores lowest. Cases with FOIA-released investigation files receive credit for the resulting documentation. |
| Mundane Explanation | 10% | Status of conventional explanations. If an official investigation ruled out all mundane candidates, this component scores high. If a plausible conventional explanation was never examined or remains supported, it scores low. This component does not assert the cause; it reflects whether alternative explanations have been formally evaluated. |
Tier-Delta Alerts
When a case's EQI is more than 15 points above or below the expected midpoint for its tier (Tier-1: 82, Tier-2: 52, Tier-3: 35), the case receives a tier-delta editorial review. The review produces a written note explaining whether the divergence reflects a data limitation, a structural scoring asymmetry, or a genuine tier mismatch. All editorial notes are visible on the case's Scoring tab.
3. Behavioral Anomalousness Index (BAI)
The BAI is a separate 0-100 integer that measures the anomalousness of the object's behavior, independent of documentation quality. It answers a different question than the EQI: not "how well is this documented?" but "how far outside known aeronautical, physical, or biological constraints does this behavior fall?"
The EQI and BAI are designed to diverge. Levelland (1957) has a relatively low EQI because it has no video and no military investigation, but a high BAI because seven independent witnesses reported vehicle electrical failures correlated with object proximity, a behavioral signature that has no known conventional explanation. Displaying both scores allows readers to distinguish cases that are well-documented but behaviorally unremarkable from cases that are poorly documented but behaviorally extreme.
| Component | Weight | What it measures |
|---|---|---|
| Kinematic Anomalousness | 35% | Severity of flight characteristics that have no known aeronautical analog: extreme acceleration without propulsion, velocity changes that would destroy any known airframe, silent hovering with abrupt departure, vertical takeoff and landing with no rotor wash or runway. |
| EM / Physical Effects | 25% | Whether the object produced detectable physical effects on its environment: electromagnetic interference with vehicle or aircraft systems, radiation injuries to witnesses, structured EM emissions detectable by radar or ELINT, or documented heating of surrounding material. |
| Agency / Awareness | 20% | Whether the object demonstrated reactive behavior toward observers, military assets, or the local environment, including vectoring toward pursuers, departing in response to interception, or maneuvering in ways that correlate with observer actions. |
| Trans-Medium Behavior | 20% | Whether the object transitioned between atmospheric and aquatic environments, or between atmosphere and space, in ways that contradict known aeronautical or hydrodynamic constraints. |
BAI scores are computed from structured claims data and physical evidence type tags in each case file. A BAI of 0 does not mean nothing anomalous occurred; it means the case file does not yet contain the structured behavioral claims data required to compute the score. Cases with BAI=0 and a documented reason are annotated with an editorial note.
4. AATIP Five Observables
The AATIP Five Observables are a framework originally developed within the Advanced Aerospace Threat Identification Program (attributed to Luis Elizondo and colleagues) for characterizing UAP behavior that exceeds known human aerospace capability. DECUR assesses each case against all five observables and reports the status as Confirmed, Possible, Not Observed, or Unknown.
Observable status is determined from structured claims in the case file. "Confirmed" means at least one verified or probable claim in the taxonomy explicitly documents the behavior. "Possible" means the behavior is mentioned but not formally verified. "Unknown" means the case data does not contain sufficient information to assess the observable.
Instantaneous Acceleration
Immediate velocity change with no measurable inertial buildup: no propulsion plume, no structural deformation signal, no sonic precursor.
Hypersonic / No Thermal
Supersonic or hypersonic velocity with no associated sonic boom, no infrared thermal signature, and no plasma sheath.
Low Observability
Transparent to some sensor systems while simultaneously visible to others: present on visual but absent on radar, or present on FLIR but absent on electro-optical.
Trans-Medium Travel
Demonstrated transition across the air-water or air-space boundary without structural change, loss of integrity, or trajectory discontinuity.
Positive Lift Without Surfaces
Sustained controlled flight without visible wings, rotors, lift surfaces, or conventional propulsion systems of any kind.
Source: Elizondo, Luis; Mellon, Christopher. AATIP internal briefing framework, later described in public testimony and interviews (2017-present). DECUR's observable scoring is based on this framework but is not an official AATIP product.
5. Claims Taxonomy
Every case file contains a structured claims taxonomy with four confidence buckets. Each claim in the taxonomy also has a type tag that drives BAI component scoring.
Verified
Confirmed by independent scientific analysis, authenticated official records, or multi-source corroboration. The claim has been subjected to external scrutiny and has not been refuted.
Probable
Reported by multiple credible independent sources, internally consistent with verified claims, and without strong evidence against. Not yet formally validated by an independent institution.
Disputed
Asserted by some sources but directly contradicted by others, or asserted by a single source with limited corroboration. The claim is in the archive but its evidentiary status is unresolved.
Speculative
Extrapolated from other evidence without direct documentation, or asserted as a hypothesis that cannot be evaluated against available records. Included to acknowledge that the claim exists, not to endorse it.
Claim Type Tags
Each claim is additionally tagged with one of seven types: kinematic, physical-effect, witness-account, official-record, sensor-data, psychological-effect, or institutional-acknowledgment. These tags drive automated BAI scoring — the kinematic claims feed the Kinematic Anomalousness component; physical-effect and sensor-data claims feed the EM/Physical component.
6. Credibility Assessments
Each case has a credibility assessment presented on the Assessment tab as two arrays: Supporting and Contradicting. These are not scores; they are structured editorial lists of the factors that strengthen or weaken the case as an unexplained event.
Supporting factors include: independent corroboration by unconnected witnesses, physical evidence with documented chain of custody, institutional acknowledgment that the event occurred, failure of known conventional explanations under investigation. Contradicting factors include: presence of plausible mundane explanations, contradictions between witness accounts, absence of physical evidence in contexts where it should exist, known incentive or capacity for fabrication.
Credibility assessments are human-authored and represent editorial judgment. They are not computed from the structured data. Each assessment is reviewed when new information is added to a case.
7. Sources Policy
DECUR distinguishes primary, secondary, and tertiary sources in its evidence hierarchy.
Original government documents (FOIA releases, national archives, congressional records), authenticated sensor captures, sworn testimony transcripts, original scientific reports, and direct witness statements collected from first-person publications or verified interviews.
Published academic papers, books by credentialed authors, investigative journalism with named primary sources, and official investigative summaries. Used to supplement primary sources and provide context.
Aggregator sites, anonymous forum posts, uncorroborated secondhand claims, and undated assertions without traceable sourcing are not used as evidence. They may appear in the background sections of cases to document the public record of the event, but are not cited as evidence for specific claims.
Every case file includes a Sources section listing the primary and secondary sources used, with direct URLs where available. The platform-level sources page at /sources documents the reference materials used across the archive.
8. AI Disclosure
DECUR uses Claude (Anthropic) as a research and drafting assistant throughout its editorial workflow. This is disclosed explicitly here in the interest of transparency.
What AI does
- +Researches source material and synthesizes information from primary documents
- +Drafts case summaries, figure profiles, and claim descriptions for human review
- +Applies structured scoring rubrics to normalized case data
- +Assists with schema normalization and data consistency checks
- +Suggests hypotheses and alternative explanations for editorial review
What AI does not do
- -Make tier assignments (all tiers are human-assigned and human-reviewed)
- -Determine claim confidence levels (verified/probable/disputed/speculative)
- -Set rubric weights or define scoring level thresholds
- -Make editorial decisions about what to include or exclude from the archive
- -Generate probability estimates of non-human origin or any equivalent framing
Every factual claim, tier assignment, and credibility assessment in DECUR is human-reviewed before publication. The structured scoring (EQI, BAI, AATIP observables) is computed from human-authored normalized data fields; the scoring script applies the rubric mechanically, but the underlying data it operates on was written and reviewed by humans.
9. Editorial Limits
DECUR is an evidence archive, not an advocacy platform. The following statements define what the archive claims and does not claim.
What does a high EQI or BAI mean?
A high EQI means an event is well-documented. A high BAI means the documented behavior has no known conventional explanation. Neither score is a measure of extraterrestrial origin or non-human intelligence. The rubric does not compute, imply, or estimate probability of any specific causal hypothesis.
What does a Tier-1 designation mean?
Tier 1 means the event is supported by official documentation. It means government or military institutions created records acknowledging the event. It does not mean the event has been officially characterized as anomalous, confirmed as unexplained, or attributed to any specific cause.
Do multiple figures claiming non-human intelligence constitute evidence?
Several key figures in the archive, including David Grusch (under congressional oath), Karl Nell, Luis Elizondo, and Eric Davis, have made on-record claims attributing observed phenomena to non-human intelligence. DECUR documents that these claims exist and links to the source testimony. DECUR does not adjudicate their truth. The existence of sworn testimony from credentialed individuals is itself a documented fact; the underlying claim remains unverified by independent evidence.
What would constitute definitive evidence?
DECUR has not defined a specific evidentiary threshold for "definitive proof" of non-human intelligence, because doing so would require assumptions about what non-human intelligence would look like. What the archive can establish is what human institutional processes have and have not confirmed: that unexplained aerial phenomena with anomalous performance characteristics have been documented by governments; that no comprehensive official explanation has been published; and that multiple credentialed insiders have made formal on-record claims that go further than official acknowledgment.
What are the limits of this methodology?
The scoring rubric measures what the data in the case files contains. It cannot account for information that exists but has not been declassified, witnesses who have not come forward, or sensor data that was never recorded. The archive represents the publicly available and reasonably verifiable record. It is not the complete record.
Rubric version 1.1. This methodology page reflects the current scoring design. Updates are versioned; the rubric version on each case score record corresponds to the version of this document in effect when the score was computed.
Questions about methodology: contact page.