Our Methodology

Transparency is not optional — it's the foundation.

Confidence Levels

Every result in TRUTH is assigned one of five confidence levels. Here is exactly what each one means and how it is determined.

Verified

Criteria

Multiple independent, high-reliability sources confirm the claim
Peer-reviewed research or official government documentation
Reproducible evidence with transparent methodology
No significant contradicting evidence from credible sources

Example

The Antikythera Mechanism is an ancient Greek analog computer dating to approximately 205-87 BCE. Confirmed by radiometric dating, CT scanning, and multiple independent archaeological analyses.

High Confidence

High

Criteria

Strong evidence from at least 2-3 reliable, independent sources
Minor gaps or ambiguities do not undermine the core conclusion
Expert consensus supports the finding
Methodology is sound with minor limitations acknowledged

Example

The Somerton Man has been identified as Carl Webb via DNA analysis (2022) and genealogical research. High confidence due to strong DNA match, though formal exhumation confirmation was pending at time of identification.

Moderate

Criteria

Credible evidence exists but requires additional corroboration
Sources may have methodological limitations or potential biases
Expert opinion is divided or evolving
Circumstantial evidence is strong but not conclusive

Example

The correlation between ocean floor magnetic anomalies and coastal archaeological sites is statistically significant but the causal mechanism is still under investigation. Multiple research teams have replicated the finding with varying degrees of agreement.

Low Confidence

Low

Criteria

Limited evidence, often from a single source or anecdotal accounts
Significant methodological concerns or uncontrolled variables
Claims have not been independently reproduced
Contradicting evidence from credible sources exists

Example

The Baltic Sea Anomaly: sonar imagery is suggestive, but rock samples show natural mineral composition. Reports of electronic equipment malfunctions near the site have not been independently reproduced under controlled conditions.

Unverified

Criteria

Claims that have not yet been subjected to rigorous analysis
New submissions awaiting source verification
Oral testimony or social media reports without documentary evidence
Potentially valid information that simply needs more investigation

Example

User-submitted reports or emerging claims that have entered the TRUTH pipeline but have not yet completed the cross-referencing and verification process.

Source Types

Not all sources are created equal. TRUTH categorizes and weights sources by type, reliability track record, and potential for bias.

Government Documents

High

Official records, FOIA releases, declassified materials, legislative records, census data. Reliability is high for factual content, though institutional bias and selective disclosure are accounted for.

Subject to redaction, delayed release, and potential classification errors.

Academic Papers

High

Peer-reviewed research published in indexed journals. Pre-prints are accepted at lower confidence. Replication status and citation count are weighted factors.

Publication bias, funding source conflicts, and replication crisis are tracked.

Witness Accounts

Low-Moderate

First-person testimony, interview transcripts, depositions. Weighted by corroboration count, temporal proximity to the event, and consistency across multiple accounts.

Memory degradation, suggestibility, and motivated reasoning reduce reliability over time.

Open Source Intelligence

Moderate

Publicly available data including satellite imagery, shipping records, corporate filings, social media analysis, and geospatial data. Cross-referenced against official records.

Data can be manipulated, selectively curated, or taken out of context.

Expert Analysis

Moderate-High

Domain-specific expert opinions, forensic analysis reports, and professional assessments. Weighted by the expert's track record, institutional affiliation, and potential conflicts of interest.

Expert disagreement is documented and presented, never hidden.

Cross-Reference Process

How TRUTH connects information across domains that traditional search engines treat as separate silos.

Entity Extraction

Natural language processing identifies people, places, dates, organizations, and events from every source document.

Graph Mapping

Extracted entities are mapped into a knowledge graph. Edges represent relationships — temporal, geographic, causal, or organizational.

Pattern Detection

Machine learning models identify statistically significant clusters, anomalies, and correlations that span multiple lenses and time periods.

Source Verification Matrix

This matrix shows the minimum cross-referencing requirements and confidence floor for each source type.

Source	Type	Base Reliability	Min. Cross-Refs	Confidence Floor
Peer-Reviewed Journal	Academic	High	1+	High
Declassified Document	Government	High	2+	Moderate
FOIA Release	Government	High	1+	High
Eyewitness Testimony	Witness	Low-Moderate	3+	Low
Satellite Imagery	OSINT	Moderate-High	2+	Moderate
News Report	Media	Moderate	2+	Low
Forensic Analysis	Expert	High	1+	High
Social Media Post	OSINT	Low	5+	Unverified

What TRUTH Cannot Do

Intellectual honesty demands that we are transparent about our limitations. No discovery engine is infallible.

AI Is Not Omniscient

TRUTH uses machine learning to identify patterns and connections, but AI can hallucinate, miss context, and amplify biases present in training data. Every AI-generated insight is flagged and subjected to additional verification.

Sources Have Biases

Government documents reflect institutional perspectives. Academic papers can suffer from publication bias. Media reports optimize for engagement. We account for these biases but cannot eliminate them entirely.

Correlation Is Not Causation

When TRUTH identifies connections between disparate events or data points, it is surfacing statistical correlations and temporal/geographic overlaps — not proving causal relationships. Interpretation requires human judgment.

Absence of Evidence Is Not Evidence of Absence

TRUTH can only analyze what exists in accessible records. Destroyed documents, unreported events, and classified materials create gaps that no engine can fill. We flag known data gaps rather than pretending they do not exist.

Human Judgment Is Irreplaceable

TRUTH is a tool for discovery, not a replacement for critical thinking. We present evidence, confidence levels, and connections. What you conclude from that evidence is your responsibility.

The important thing is not to stop questioning. Curiosity has its own reason for existing.
Albert Einstein