DOC #053 01_PUBLIC_WEBSITE_DOCUMENTS

SUBKUZ SOURCES, EVIDENCE AND METHODOLOGY DISCLOSURE STANDARD

Subkuz Global Governance Framework | Issued by Subkuz Media Networks Private Limited (CIN: U72900JH2020PTC014521)

Document #053 | Verified Master Version

How Subkuz Explains the Basis, Limits and Verification of Evidence-Intensive Journalism

PUBLICATION INFORMATION PUBLICATION INFORMATION
Issued by Subkuz Media Networks Private Limited
Corporate Identification Number CIN U72900JH2020PTC014521
Previous legal name during statutory disclosure period Subkuz Media Networks Private Limited
Effective date 6 August 2026
Last reviewed 7 August 2026

1. Purpose of this Standard

Readers should be able to understand not only what Subkuz concludes but how the conclusion was reached. This Standard defines the public disclosure expected for investigations, fact-checks, data journalism, technical tests, rankings, surveys, polls, major explainers, media-forensics work and other items in which the method materially affects credibility. It complements the Newsroom Standards and Verification Policy, Fact-Checking and Media Forensics Policy, Anonymous Sources and Source Protection Policy and Investigative Journalism and Legal Review Protocol.

The purpose is meaningful transparency, not the publication of every private note or source identity. A methodology section explains the evidence categories, reporting period, verification process, limitations and material conflicts while protecting confidential sources, personal data, legal privilege, security-sensitive systems and information whose disclosure would create disproportionate harm.

2. Evidence is assessed by function, not prestige

No category of source is automatically reliable in every circumstance. An official document may prove what an institution recorded but not that every underlying assertion is true. A company may accurately describe its announced policy while remaining interested in claims about performance. A qualified expert may explain a field without possessing direct knowledge of an event. An eyewitness may describe a personal observation without establishing a broader cause.

Subkuz assesses identity, basis of knowledge, proximity, expertise, independence, method, documentary support, date, jurisdiction, possible interest, consistency and corroboration. Several sources do not create independent confirmation when they depend on one originating claim, organisation, press release, dataset or campaign.

3. Principal source categories

The evidence base may include legislation, judgments, regulatory decisions, public filings, government records, audited accounts, direct interviews, direct observation, original documents, datasets, peer-reviewed research, preprints, technical documentation, source code, photographs, recordings, satellite or location information, public statements, company material, agency reporting, social media, user-generated content and confidential sources. The public methodology identifies the categories that materially support the work.

The methodology does not inflate a weak source by giving it a broad title. A court filing is described as a party's filing, not a judicial finding. An internal company figure is identified as company-supplied unless independently audited. A social post is identified as a post and verified for authenticity and context. A preprint is not described as peer-reviewed research.

4. Primary and secondary evidence

Subkuz seeks primary evidence where reasonably available. Primary evidence may include the original law, judgment, decision, filing, dataset, interview, recording, document, code, direct observation or technical output. Secondary sources may provide important reporting, historical context, expert interpretation or access to evidence that is otherwise unavailable.

Where another publication produced the original reporting, Subkuz credits the originating journalist or organisation and distinguishes any additional Subkuz work. Rewriting another report in different words does not create independent confirmation or original investigation. A link to a secondary article does not substitute for reviewing the primary record when the primary record is available and material.

5. Named and anonymous human sources

Named sources are preferred where identification is safe and relevant. An article describes the person's role, basis of knowledge and material interest sufficiently for readers to assess the account. A title alone does not prove direct knowledge, and a former employee may know historical practice without knowing current operations.

An anonymous source is used only under the source-protection policy. The responsible editor knows the identity except in an extraordinary approved circumstance, examines the reason for anonymity, tests motivation and seeks corroboration. The public description is as specific as safety permits. Anonymity is not used to publish unsupported personal attack, strategic political messaging or commercial accusation without evidence.

6. Documents and records

Documents are examined for origin, authorship, date, version, completeness, signatures, metadata, internal consistency, missing attachments, alteration, relevant legal status, confidentiality, privacy and source interest. A document can establish that a statement was made without establishing that the statement was true. A draft is not described as a final decision, and an unsigned or incomplete record receives appropriate qualification.

Leaked or confidential material is not published indiscriminately. The methodology explains the nature of the records and the verification performed without exposing a source or unrelated personal information. Passwords, private keys, identity documents, medical records, intimate content and information concerning uninvolved people are withheld or redacted unless publication is necessary, lawful and proportionate.

7. Data, calculations and statistical evidence

A data methodology identifies the source, collection period, population, fields, definitions, inclusion and exclusion criteria, missing values, duplicates, transformations, assumptions, calculations, sampling, error, revisions and limits of inference. A number is not presented as meaningful merely because it is large or precise. Downloads, accounts, active users, verified people, impressions, readers and subscribers are not treated as interchangeable measures.

Calculations distinguish totals from averages, percentages from percentage points, revenue from profit, booked amounts from cash movement, nominal values from inflation-adjusted values, estimates from audited figures and correlation from causation. Where the result cannot be reproduced from public information because the underlying data are confidential or licensed, the methodology explains the validation process and limitation.

8. Research, science and expert evidence

Research coverage identifies the authors, institution, publication status, study design, sample, population, geography, period, funding, conflicts, limitations, uncertainty and whether the result has been replicated. A single study is not described as settled consensus unless the wider evidence supports that conclusion. Statistical significance is distinguished from practical importance.

Experts are selected for relevant competence rather than public prominence alone. The methodology identifies a material employment, funding, advisory or advocacy relationship where relevant. Where qualified experts disagree, Subkuz explains the basis and weight of the disagreement instead of manufacturing equal balance between positions with unequal evidence.

9. Polls, surveys and public opinion

A poll or survey disclosure identifies the commissioner, conducting organisation, field dates, target population, sample size, sampling method, collection mode, question wording, weighting, margin of error where applicable, undecided or non-response treatment, geography and material limitations. A self-selected online poll is not described as representative of the general public.

A ranking, score or index explains the criteria, weighting, source data, review period, exclusions and conflicts. Subjective weighting is not hidden behind numerical precision. A sponsor or client does not receive the right to design a public ranking that appears independent without a clear commercial classification and disclosure.

10. Technical testing and cybersecurity

A technical methodology identifies the device, operating system, application or software version, account type, date, network conditions, configuration, test procedure, sample, expected behaviour, observed result and known limitations. Testing one device, account or environment does not automatically establish universal performance. A result produced through AI-generated code or automation is reviewed by a competent person and validated independently.

Security testing is lawful, proportionate and designed to minimise harm. A vulnerability is not exploited beyond what is necessary to demonstrate it safely. Personal data, credentials, source identities and unrelated systems are not accessed unnecessarily. Technical detail capable of enabling immediate harm may be withheld during coordinated remediation.

11. Images, audio, video and media forensics

Media verification may include original-source tracing, reverse search, metadata, chronology, geolocation, weather, architecture, shadows, audio continuity, frame analysis, compression, comparison with known records, direct confirmation, expert review and detection tools. No single automated detection score is treated as conclusive proof that material is authentic or synthetic.

The public methodology states what was verified and what remains uncertain. An authentic image may still be misleading if it is old, cropped, unrelated or presented with a false caption. A synthetic or materially altered item is labelled according to the AI and Synthetic Media Policy. Forensic methods are described without publishing a technique that would compromise an active investigation or security control.

12. Company, government and campaign material

Company documents, founder interviews, government statements, campaign claims, advocacy reports and public-relations material may be important sources. They are identified as interested-party material and are independently assessed. A company is authoritative about what it announced, not automatically about whether its performance, security, sustainability, user count or market leadership claim is correct.

A government is authoritative about an official decision, but its political justification or disputed account remains subject to scrutiny. A campaign filing or opposition document may contain valuable evidence while also serving a strategic purpose. The methodology explains the origin and the corroboration performed.

13. Right of reply and contrary evidence

A serious criticism or allegation normally receives a meaningful pre-publication right of reply. The methodology may state when and how the subject was contacted, the deadline, whether a response was received and whether the response changed the work. It does not publish confidential source information merely to prove that contact occurred.

Failure to respond is not evidence of guilt. A response is not automatically true. Subkuz tests the response against documents, chronology, independent sources and the existing evidence. Material contrary evidence is addressed rather than omitted because it weakens a preferred conclusion.

14. Limitations and uncertainty

A credible methodology states what the evidence cannot establish. Limitations may arise from incomplete records, unavailable data, restricted access, small sample, translation, legal restrictions, confidential sources, technical constraints, selection bias, disputed definitions, changing conditions or the inability to reproduce a claim independently.

Uncertainty is expressed precisely. Phrases such as according to the company, the available records indicate, the estimate is based on, the allegation is denied, the claim could not be independently confirmed and alternative explanations remain possible are used where they accurately describe the evidence. Qualification is not used mechanically to publish an otherwise unsupported accusation.

15. Standard Sources and Methodology structure

A full methodology section may include the reporting question, evidence period, jurisdictions, interviews, documents, data, calculations, technical tests, visual verification, source anonymity, right of reply, conflicts, funding, exclusions, limitations and correction route. The sequence may be adapted to the work, but the disclosure must enable a reasonably informed reader to understand how the central conclusion was reached.

The following table identifies the principal disclosure fields and their public purpose.

Methodology field Public purpose
Reporting question Defines what the work examined and what it did not
Period and jurisdiction Locates the evidence in time and law
Evidence categories Identifies interviews, records, data, media, tests and other sources
Selection method Explains inclusion, exclusion and sampling
Verification Describes corroboration, authentication and independent checks
Calculations and analysis States assumptions, transformations and reproducibility
Right of reply Explains the opportunity given to materially criticised parties
Conflicts and funding Identifies relationships affecting reader assessment
Limitations States what remains uncertain or unavailable
Correction route Provides a direct process for challenging the evidence or conclusion

16. When a public methodology is not required

A routine short news item, event notice, public-service update, simple interview, ordinary opinion or brief report may not require a separate methodology section when the source and basis are clear from the article. The underlying newsroom verification standard still applies. The absence of a separate block does not permit unsupported claims or vague attribution.

A methodology section is added when the method is central to credibility, when the evidence is complex, when a ranking or quantitative claim could otherwise be misunderstood, when anonymous sourcing materially supports a conclusion, when the item is an investigation or fact-check, or when a reader would reasonably need more information to assess the work.

17. Corrections and challenges to evidence

A person may challenge a source attribution, calculation, dataset, quotation, document, image, translation, methodology or limitation through the Correction, Clarification and Retraction Request Form. The request should identify the item, exact claim, alleged defect and supporting evidence. Subkuz may reproduce a calculation, inspect the original record, consult a specialist, contact the source or update the methodology.

A confirmed material error is corrected visibly. A new development is recorded as an update rather than falsely described as an original error. A methodology correction may affect only the explanation or may require a change to the central conclusion, rating, headline, data visualisation or entire publication. The public notice states the effect.