Fluent language is not evidence.
A clean argument, confident summary, or elegant interface can make a claim easier to process. Ease is not verification.
RESEARCH METHODOLOGY CENTER
Fluent language is not evidence.
A public method for separating source from repetition, context from screenshot, documentation from interpretation, and confidence from truth.
EVIDENCE WITHOUT VERDICT
The record describes. It does not condemn.
No judgment whatsoever of the person or the mind. Research here may test a proposition, trace a source, expose a contradiction, or record an unknown. It does not assign moral worth, diagnose identity, purify belief, or decide who deserves to speak.
Documented, disputed, unsupported, corrected, incomplete, and unknown are conditions of a record. They are not verdicts on the author, subject, source, reader, or dissenter. A person may reject the Ministry’s interpretation and still use the evidence, inspect the receipts, contribute a correction, or leave.
Challenge the claim. Preserve the person. Keep correction open.
QUICK ANSWER
In one sentence
A realism-first research methodology for source literacy, claim classes, context, AI-assisted research, adversarial inquiry, corrections, and ethics.
THREE AXIOMS
Repetition is not corroboration.
If one flawed source is copied across a thousand pages, the evidentiary chain may still have only one origin.
A link is not automatically a source.
A URL proves that a destination can be named. It does not establish authorship, method, provenance, stability, or contextual accuracy.
The archive must show its work.
Consequential claims should leave enough evidence for another reader to reconstruct why they were published.
THE FOUR-MOVE CHECK
Add friction before belief.
STOPNotice the claim and your reaction.
SOURCEWho created this, how, and for what context?
COMPAREFind independent coverage, contradiction, and better evidence.
TRACEReturn quotes, images, statistics, and claims to the earliest usable context.
THE SOURCE LADDER
Weight is contextual, but provenance matters.
Primary or physical evidence
Original records, datasets, artifacts, recordings, direct measurements, official filings, contemporaneous documents. Primary does not mean infallible.
Method-bearing analysis
Research or investigation that explains how evidence was collected, selected, tested, and limited.
Independent reporting and synthesis
Useful when it identifies sources and distinguishes reporting from interpretation.
Later recollection and commentary
Potentially valuable, especially for experience and reception, but sensitive to memory, incentive, and retrospective framing.
Repetition, aggregation, and unsupported assertion
Discovery clues, not automatic corroboration. Trace them backward.
CLAIM TYPES
Name what kind of sentence you are publishing.
OBSERVEDDOCUMENTEDREPORTEDINTERPRETEDDISPUTEDHYPOTHESISDOCTRINESPECULATIONUNKNOWN
These labels do not decide truth automatically. They prevent one type of claim from quietly dressing as another.
A QUOTE IS NOT CONTEXT
Recover the room around the sentence.
Ask who spoke, when, to whom, in which document or recording, through which edition or translation, and what the surrounding material changes. A sentence cut cleanly from its setting can remain verbatim and still mislead.
THE SCREENSHOT PROBLEM
An image of text can sever the chain.
A screenshot may preserve appearance while removing URL, date, author, revision history, surrounding language, and whether the image itself has been altered. Treat it as a lead until the underlying artifact is recovered.
RECOVERY ORDER
Recover the original URL or file. Preserve the exact bytes and available document or response metadata. Record when and how the artifact was found. Compare revisions or archived captures. Hash the preserved artifact. Keep the screenshot as rendering evidence—not as automatic proof of what a server delivered, what an author meant, or whether the pictured record is authentic.
THE AI PROBLEM
The model can help inspect the archive. It cannot become the archive.
AI can summarize, compare sources, extract disagreements, organize timelines, and generate hypotheses. External factual claims still need evidence outside the model. Fluency is especially dangerous here because a machine can generate the texture of confidence without the provenance.
ADVERSARIAL INQUIRY
A hypothesis that cannot lose is a costume.
What evidence would make this claim weaker?
What is the strongest competing explanation?
Are the sources actually independent?
What would we expect to observe if the claim were false?
Did the conclusion arrive before the evidence?
ABSENCE
A gap is not a verdict.
Missing evidence may matter. It does not by itself prove suppression, innocence, guilt, rescue, selection, conspiracy, or nonexistence. State the gap. State what would have been expected. Keep the inference separate.
TIMELINES
One event can have many dates.
EVENTDOCUMENTPUBLICATIONDISCOVERYREVISIONDIGITIZATIONLAST REVIEWED
Do not let one timestamp impersonate another.
RUN THE METHOD
Do not stop at the slogan. Build the receipt.
Bound the claim
Write one sentence. Name its claim type, material consequence, and what the sentence does not establish.
Recover the source
Find the earliest inspectable artifact you can support. Record edition, issuer, acquisition path, and exact bytes where available.
Test independence
Trace repeated coverage backward. Ten copies of one origin remain one evidentiary dependency.
Rebuild context
Recover speaker, date, audience, document, surrounding passage, translation, edition, and revision state.
Separate the clocks
Keep event, document, publication, discovery, revision, digitization, and review dates from impersonating one another.
Record how the claim can lose
Name contrary evidence, unknowns, the strongest competing explanation, and the exact rewrite required if a load-bearing source fails.
RESEARCH RECEIPTS
Separate the claim from the source and the correction.
The local Receipt Toolkit provides printable Claim, Source, and Correction receipts without submitting entered data to this site.
Open the Claim ReceiptRESEARCH ARTICLE RECEIPT
Every substantial article should be reconstructable.
- Summary
- What is being claimed?
- Question
- What bounded question is the article trying to answer?
- Claim class & confidence
- What type of claim is this, and how strong is the evidence?
- Why it matters
- What material consequence makes the question worth asking?
- Sources
- Primary and secondary evidence with provenance.
- Competing interpretations
- The strongest serious alternatives.
- Unknowns
- What remains unresolved.
- Corrections
- What changed after publication and why.
PUBLISHED SOURCE WORK
Research that has crossed the publication gate.
Your Conscience Is Not TerritoryLegal-source analysis + normative philosophy.Can an Algorithm Become an Unanswerable Authority?NIST AI RMF 1.0 + EU AI Act consolidated text dated 27 July 2026 + institutional analysis.A Hash Proves Bytes, Not TruthNIST/Library of Congress fixity sources + archival limits.What Makes Authority Legitimate?UN/OECD governance sources + political-legitimacy theory.The Right to Change Your MindHuman-rights sources + correction and research-integrity practice.Fear Is Not EvidenceRisk-perception research + public risk-communication guidance.
RESEARCH ETHICS
Evidence is not permission to build a dossier on a person.
Minimize unnecessary personal information. Do not publish private addresses, diagnoses, genetic maps, family dossiers, or human target lists as research artifacts. The purpose is to understand claims and systems, not manufacture vulnerability.
THE ARCHIVE MUST REMEMBER THAT IT CAN BE WRONG.
THE COST OF ASKING
Inquiry can be formally permitted while institutionally expensive. The difference requires evidence, boundaries, and a reason trail.
Read How Institutions Punish QuestionsRELATED LIBRARY
Follow the question into the evidence-complete Library.
CONTINUE THE TEST
Build an archive that can admit error
NEW SCRIPTURE STUDY
What does the Bible say about the Antichrist?
Compare the Johannine passages with the Ministry’s deliberately contested Light-Bringer interpretation—text, history and theological proposal kept distinct.