Status of this policy

Public and revisable editorial standard. Last substantial update: 3 September 2026. It has not been endorsed by an academic institution; it is a commitment readers can test against the published work.

Author-published

Operating rule

Research begins with a bounded claim, not with a famous source and a search for something to say. Primary documents, original studies, and official data are preferred, followed by strong specialist syntheses. A source is rejected when its origin is unclear, when it only repeats another page, or when it agrees with the conclusion without actually supporting it.

A five-stage model from discovery and retrieval to support, extraction, and possible citation
Figure 1 — The trust path. Crawl access is only the first gate. The page must fit the question, support the claim, expose limits, and present extractable answer units before it can become a citation candidate.

Six tests before a source is accepted

1. Entailment

Does the source support this exact claim, at this level of specificity, or does it merely discuss the same subject?

2. Provenance

Who produced it, under what institution or publication process, and is it primary research, official documentation, reporting, or commentary?

3. Evidence type

Observation, experiment, review, textual interpretation, testimony, statistics, and philosophy answer different questions.

4. Currency

Changing law, medical guidance, software, and live data require current verification. Foundational historical sources may be old by definition.

5. Independence

Ten pages copying one release are not ten independent lines of evidence. Separate support is sought where possible.

6. Falsification pressure

The strongest counter-source or objection is actively sought. Claims are narrowed or withdrawn when they do not survive.

Source hierarchy

TierExamplesUse
A — direct primary sourceOriginal text, research paper, dataset, official law or judgment, mission reportDefault support for central claims when accessible and interpretable.
B — specialist synthesisSystematic review, scientific society, government agency, edited reference workField-level context, consensus, disputes, and interpretation of evidence.
C — secondary explanationHigh-quality science journalism, textbook, university explainerContext and accessibility; not a substitute for the source on sensitive claims.
D — lead, not evidenceSocial post, anonymous site, uncited or unattributed pageMay suggest a question or search term; not cited as factual proof.

The claim ledger

Extended dossiers are decomposed into claims with identifiers, source type, supporting references, confidence, and caveats. This does not mechanize judgment; it prevents broad assertions from hiding inside polished prose. The flagship dossier exposes both a human-readable appendix and a machine-readable JSON ledger.

Separation of layers

Religion-and-science dossiers distinguish the source text, interpretive history, observation, scientific model, and philosophical or theological inference. Medical pieces distinguish symptom description, diagnosis, cultural framing, and individual clinical decisions. Interaction is legitimate; category substitution is not.

Review labels

LabelMeaning
Author review completeThe author rechecked structure, sources, links, and language. It is not external peer review.
External specialist review completeAn independent specialist reviewed a stated scope, and the reviewer and material changes are disclosed.
External review pendingNo independent specialist has reviewed the work. This is the current state of most pages.
Peer reviewedUsed only for work actually published through a documented academic peer-review process. It does not currently apply.

Books and conflict disclosure

This website also presents Ahmed Alhafiz’s forthcoming books, and some research questions originated in those manuscripts. That creates an editorial and potential commercial interest. The manuscripts are therefore not used as evidence for public factual claims. Their relationship appears after the argument, while the evidence remains external and accessible.

Versions and corrections

Each extended dossier carries a version and a source-review date. A change that alters a conclusion, an important number, or attribution receives a logged revision. Minor typographical repairs may not create a new version. Current states are listed in the review and corrections register.

Discoverability is not authority

Pages use static HTML, descriptive headings, internal links, structured data, and extractable references so readers and search systems can understand them. This improves eligibility, not entitlement: no markup guarantees ranking or citation, and no technical layer can compensate for weak evidence.

Suggested citationAlhafiz, Ahmed. “Source Selection, Review, and Corrections.” Version 2.1, Official Website of Ahmed Alhafiz, 3 September 2026.