Precise claims
A verdict applies to quoted or carefully restated wording—not to a person, group or entire subject.
Method & corrections
Do not trust the author.
Audit the work.
An anonymous project cannot borrow credibility from biography. Its reasoning, sources, limits and corrections must carry the entire burden—and readers should know where that accountability remains incomplete.
01CLAIMPrecise wording
→02EVIDENCETraceable record
↓03ASSESSVerdict + limits
↓04REVIEWDate + triggers
↺05CORRECTVisible history
Anonymity and accountability
Anonymity removes one appeal to authority. It also removes a layer of accountability.
The project uses no personality-led brand so claims are not accepted through status, credentials or loyalty. That is a legitimate design choice—not a credibility advantage.
Without a named author, readers cannot easily inspect conflicts or personal history. Transparent sources, dated reviews, a private challenge channel and visible corrections compensate for part of that loss. They do not fully replace an accountable author.
Editorial standard
Eight commitments for every substantive case.
These are publication tests, not aspirations. If a case cannot meet them, it should be labelled incomplete or not published as an assessment.
Visible reasoning
Facts, interpretations, causal steps and uncertainty must be distinguishable in the published analysis.
Source transparency
Readers should be able to open the strongest sources and trace repeated reporting to its origin.
Symmetrical standards
Institutional, anti-institutional, allied and opposing claims face the same evidential burden.
Competing explanations
A preferred explanation is compared with credible alternatives, not only with a convenient straw man.
Revision conditions
Every assessment states what evidence could weaken, strengthen or overturn it.
Dated review
Substantive cases show when sources and conclusions were last examined.
Visible correction
Material errors and assessment changes are labelled and preserved rather than silently rewritten.
Publication workflow
From interesting claim to defensible assessment.
The same workflow applies when the likely conclusion favours institutions, challenges them or remains unresolved.
- 1Select
Public relevance, testability and a claim precise enough to assess.
- 2Research
Original documents, source provenance, independent evidence and contrary findings.
- 3Analyse
Fact kernel, causal chain, alternatives, attribution limits and stop conditions.
- 4Stress-test
Symmetry test, strongest countercase and search for disconfirming evidence.
- 5Publish
Verdict, uncertainty, sources, review date and what could change it.
- 6Maintain
Trigger review when stronger evidence, official findings or source failures emerge.
Assessment language
A verdict and confidence answer different questions.
The verdict describes the relationship between the precise claim and the evidence. Confidence describes how stable that assessment is given source quality, independence, access and live alternatives.
Verdict
Confidence
Source standard
Primary does not mean true.
Official does not mean false.
Original document, full recording, dataset, physical or forensic record.
A separate evidence line with different access and origin.
Relevant expertise with methods and evidence exposed.
Useful synthesis whose sourcing must be traced backwards.
Anonymous post, screenshot, clip or assertion awaiting authentication.
A source can be high on the ladder and still be incomplete, deceptive or misinterpreted. Authority affects weight; it never eliminates verification.
AI-use disclosure
AI may assist the work.
AI is never the evidence.
AI tools may be used to structure drafts, generate search leads, compare wording, identify missing counterarguments and check consistency across the resource.
- AI-generated statements require verification against traceable sources.
- AI summaries do not replace the underlying document.
- Citations must support the exact published claim.
- Important quotations, dates and numbers require direct checking.
- Responsibility for the published assessment cannot be delegated to a model.
Corrections policy
Change the conclusion when the evidence changes.
Corrections are evidence that the method can lose—not proof that the work was untrustworthy. Silent rewriting is the credibility failure.
Correction
A substantive factual error affected the analysis. Show the original error, corrected wording, date and effect on the verdict.
Revision
New evidence changes the conclusion or confidence. Preserve the earlier assessment and explain why it moved.
Clarification
Wording was ambiguous or incomplete without making the underlying conclusion false.
Maintenance
A broken link, spelling error or formatting issue changed without affecting meaning.
Every material correction should showthe earlier wording · the corrected wording · date · reason · effect on verdict and confidence
Review triggers
A date is not a guarantee of freshness.
Cases are reviewed when a meaningful trigger appears. There is currently no automated monitoring promise. Readers should treat an old review date as a reason to check for newer evidence.
- New primary documents or datasets
- Court, regulatory or official findings
- Retraction or material correction by a key source
- Independent corroboration or contradiction
- A broken causal link or newly credible alternative
- A specific challenge containing traceable evidence
Public change log
Corrections and material method changes.
Individual applied cases also display their own review date and correction history.
DateItemChangeType
Method expansionAdded the full editorial standard, correction taxonomy, AI-use disclosure and challenge template.Enhancement
Case review baselineThirteen applied cases published with review dates, sources, change conditions and individual correction fields.Enhancement
Current material correctionsNo material corrections have been logged for the published applied cases.Status
Challenge an assessment
Make the objection as testable as the original claim.
Strong challenges identify exact wording, supply a traceable evidence line and explain what part of the conclusion should change.
Privacy and review
The application database contains no identity field and does not record IP addresses. Cloudflare may retain ordinary technical security and service logs. A submission does not promise a personal reply. Valid challenges will be reviewed, and any resulting change will be recorded in the public change log.
External reference points
Standards used as benchmarks—not borrowed authority.
The method is informed by established commitments to fairness, source and methodology transparency, independence and visible corrections. Test the Claim is not claiming membership, certification or equivalence with these organisations.