D3

Media and Public Discourse

This document adapts Reality Audit to journalism, editorial communication, platform distribution, and public discourse. It examines the documented chain from event, source, and trace to selection, language, representation, publication, distribution, effect, and correction.

Audit types in the domain

Primary audit types

  • A4Representation Audit Tests traces, images, audio, video, clips, perspective, selection, editing, graphics, and presentation.
  • A1Language Audit Tests definition, reference, quotation, paraphrase, presupposition, modality, headlines, translation, and claim form.
  • A6Authority Audit Tests source status, expertise, editorial mandate, fact-checking, institutional trust, and demands for assent.

Supporting audit types

  • A5System and Institutional Audit Tests editorial workflow, ownership, incentives, platform rules, correction channels, metrics, and accountability.
  • A3Model Audit Tests polling, risk models, ranking systems, statistical presentations, estimates, and algorithmic distribution.
  • A7Practice Audit Tests actual sourcing, verification, publication routines, moderation, correction, reply, and review.
  • A2Identity Audit Used where person, role, group, affiliation, or label governs credibility, portrayal, participation, or responsibility.

Abstract

Media and public discourse give people access to events, sources, and perspectives they do not encounter directly. This mediation is necessary, but never without selection, medium, language, editing, sequence, interpretation, and distribution. The audit must therefore distinguish event, trace, source, claim, quotation, representation, editorial product, platform distribution, and public assignment to actuality.

The document establishes a traceable media chain; requirements for mandate, source provenance, verification, anonymity, and source protection; distinctions among fact, analysis, comment, and claim; audit of images, audio, video, headlines, graphics, and statistics; assessment of algorithmic distribution, advertising, and metrics; use of the seven audit types; F1–F7 findings; and correction through visible amendments, reply, version history, and review.

1

Purpose and scope

The purpose is to make public media representations traceable from the bounded event or matter to the sources, choices, representations, formulations, and distribution mechanisms sustaining the publication. The document does not determine whether media as a whole are trustworthy or biased; it examines what a particular product or distribution chain actually rests upon.

The scope includes journalism, documentary, interviews, editorial comment, fact-checking, press photography, audio and video, newsletters, podcasts, live coverage, social media, platform moderation, public posts, and other forms of broad social communication. Requirements must be adapted to genre, risk, time constraints, and publication form.

Reality Audit does not presume that a publication is manipulated, captured, or unreasonable. It begins with mandate, source chain, and distinctions. It may conclude with F1 — supported correspondence just as readily as with uncertainty, insufficient grounds, error, capture risk, or capture identified.

2

The media chain and decisive distinctions

A public representation is rarely produced in one step. It moves through event, trace, source, registration, verification, selection, editing, headline, publication, distribution, and interpretation. Each stage can add understanding, but also limitation, loss, framing, and uncertainty.

The audit must therefore distinguish what happened from the source reporting it; what the source said from the quotation used; what was registered from the clip presented; and the editorial product from later platform distribution and public reaction.

Decisive distinctions in the media chain
ConceptFunctionIs not automatically
Event or matterThe concrete condition, action, document, or question the publication concerns.Identical with the media product or dominant narrative.
SourceA person, institution, document base, or registration capable of providing information.A truth guarantee or neutral access to the whole matter.
TraceImage, audio, document, data, metadata, or other registered material referring beyond itself.The event, cause, or complete context.
ClaimA formulation testable for grounds, scope, and correspondence.Fact merely because it is published or repeated.
QuotationA marked reproduction of source wording within a particular context.The whole statement, intention, or complete source position.
ParaphraseAn editorial or journalistic reformulation of content.Verbatim source wording or interpretation-free transfer.
RepresentationA bounded presentation through text, image, audio, video, graphics, or layout.The object, event, or actuality itself.
EditingSelection, sequence, cutting, combination, and presentation for a publishing purpose.Manipulation in itself or guarantee of fair completeness.
Headline and leadCondensed presentation guiding expectation and attention.A complete summary or independent proof of the content.
PublicationMaking an editorial or public product available with identifiable version and responsibility.That later copies, excerpts, or commentary retain the same content and status.
DistributionHow content is ranked, recommended, shared, advertised, and made visible.A neutral reflection of public relevance or truth strength.
EngagementRegistered clicks, viewing time, reactions, sharing, or comments.Truth, quality, representativeness, or durable public opinion.
3

Mandate, roles, and publication object

The audit must bound which particular publication, clip, headline, source use, distribution chain, or public exchange is examined. It must not begin with a total judgment of a newsroom, journalist, platform, group, or the public sphere as a whole.

Roles must be identified without becoming the whole person: source, reporter, photographer, editor, editor-in-chief, commentator, fact-checker, platform operator, advertiser, moderator, and affected party may hold different tasks, grounds, and responsibilities.

MM1

Bounded publication

Identify the particular article, broadcast, episode, post, clip, image, headline, poll, or distribution to be tested.

MM2

Purpose and genre

Register whether the product is news, comment, analysis, satire, advertising, documentary, live coverage, reader contribution, or another expressed genre.

MM3

Period and version

Bound publication and audit time and register updates, corrections, archived versions, and later excerpts.

MM4

Roles and responsibility

Map who gathered, verified, edited, approved, published, distributed, moderated, and could correct.

MM5

Affected parties

Identify persons and groups portrayed or affected and the opportunity they had for relevant reply or correction.

MM6

Platform and distribution

Document where the product was published, how it was recommended or advertised, and which other versions circulated.

MM7

Rights and protection

Register source protection, privacy, safety, copyright, confidentiality, and other limits affecting what may be documented or disclosed.

MM8

Express limitations

State what the audit cannot determine concerning motive, liability, legality, professional ethics, representativeness, or unavailable material.

4

Sources, provenance, anonymity, and source protection

Source work requires more than counting sources. The audit must show what each source can know, how the information arose, whether sources are independent, what interests or risks exist, and what was verified through other material.

Anonymity may be necessary for safety, whistleblowing, and access to information of public interest. It also reduces the public’s ability to assess role, proximity, and interest. The audit should therefore test process and grounds without demanding public exposure of a protected source.

MS1

Source provenance

Register source type, proximity, time, role, access, and how the information was obtained, within lawful and ethical protection.

MS2

Knowledge basis

Distinguish what the source observed, did, heard, was told, interpreted, believed, or inferred from other material.

MS3

Independence

Test whether multiple sources are actually independent or derive from the same document, person, press release, or rumour.

MS4

Interest and risk

Register relevant interests, conflicts, loyalties, retaliation risk, and other conditions affecting source use or protection.

MS5

Primary material

Seek original documents, complete recordings, data, first-hand registration, or a closer source where possible and responsible.

MS6

Anonymous source

Document internally why anonymity is necessary, who knows the identity, what verification occurred, and which claim the source actually carries.

MS7

Counter-material and reply

Obtain relevant counter-material and give the portrayed party a real opportunity to respond within the conditions of the matter.

MS8

Protection without invisible authority

Protect identity where necessary while making source type, verification method, uncertainty, and editorial responsibility as visible as responsible.

5

Claim types, verification, uncertainty, and counter-material

A media representation may contain observations, factual claims, estimates, explanations, characterisations, forecasts, legal claims, and value judgments. These require different grounds and must not be presented with the same degree of certainty.

Verification is not one simple stamp. It may include document control, time and location checks, source crossing, technical analysis, professional assessment, direct reply, and testing of alternative explanations. What was actually tested before publication and what remained unresolved must be visible.

MV1

Claim register

List decisive claims and mark them as observation, fact, estimate, explanation, forecast, legal assessment, comment, or another relevant type.

MV2

Grounds per claim

Connect each decisive claim to particular sources, traces, verification, and expressed uncertainty.

MV3

Alternative explanation

Test relevant alternatives and document why they are rejected, remain open, or are less supported.

MV4

Counter-material

Make material against the main representation visible and assess it before conclusion or publication.

MV5

Time-critical publication

Where publication precedes full verification, clearly mark what is confirmed, preliminary, reported, disputed, or unknown.

MV6

Expert control

Use relevant expertise within its field while showing competence, uncertainty, disagreement, and what expert status does not determine.

MV7

Update duty

Define how material new information, error, and reply will be assessed and connected to the original publication.

MV8

Strength and modality

Match terms such as may, likely, documented, confirmed, and proved to what the grounds actually support.

Claim types and required grounds
Claim typeMay be supported byMust not automatically become
Direct observationDocumented proximity, registration, time, place, and conditions.Explanation of cause, motive, or complete context.
Factual claimTraceable sources, documents, registrations, and counter-testing.Certain merely because quoted from authority.
EstimateDefined method, sample, uncertainty, error sources, and domain of validity.An exact number or actual count.
Causal explanationDocumented mechanism, temporal order, alternatives, and relevant professional grounds.Supported merely by covariance or simultaneity.
Motive claimExpressed reasons, actions, and material capable of bearing the attribution.Psychological certainty concerning inner state.
Legal characterisationRelevant legal grounds and competent assessment with visible status.A final legal determination where the matter is unresolved.
ForecastModel, assumptions, horizon, scenarios, and prior performance.Advance knowledge of what will occur.
Value judgment or commentVisible position, premises, and argument.An unmarked factual claim or editorial consensus.
6

Quotation, interview, context, translation, and representation

Quotations and interviews provide access to source wording but are always selected, bounded, and placed in context. The audit must distinguish verbatim quotation from paraphrase, full statement from excerpt, question from answer, and source wording from editorial characterisation.

Translation is not mechanical replacement of words. It involves choices of meaning, register, modality, and institutional force. Where alternatives could alter responsibility, certainty, right, or public impression, the choice and uncertainty must be visible.

MQ1

Verbatim accuracy and version

Check the quotation against recording, notes, or source text and register editing, language, and version.

MQ2

Relevant context

Include the preceding and following context necessary to understand meaning, qualification, irony, or condition.

MQ3

Question and premise

Test whether the question inserted an unestablished premise, constrained the answer, or was omitted in a way changing meaning.

MQ4

Quotation and paraphrase

Clearly mark source wording, editorial summary, interpretation, and characterisation.

MQ5

Representativeness

Do not treat one statement as representative of a person, group, institution, or period without relevant grounds.

MQ6

Translation and register

Document source language, translator, central choices, ambiguity, and alternatives where these may alter the claim.

MQ7

Cutting and combination

Make visible where answers from different times, questions, or contexts are combined and test whether the whole becomes misleading.

MQ8

Checking and reply

Distinguish any quotation or factual checking from a right to editorial control, and document how relevant objections were handled.

7

Images, audio, video, graphics, polling, and visualisation

Visual and auditory media can document concrete conditions with considerable force, but perspective, crop, time, recording conditions, compression, sound mix, captioning, and editing affect what the recipient can perceive and infer. Technical authenticity alone does not determine what the representation establishes concerning cause, scale, or completeness.

Graphics and polling may make complex material accessible. They may also obscure sampling, axes, weighting, missing data, model assumptions, and uncertainty. Numbers and visualisation must remain traceable to method and source.

MR1

Original and provenance

Register original file, capture source, date, location, equipment, metadata, copy history, and editing history as far as possible.

MR2

Crop and perspective

Test what lies outside the frame, sound field, time interval, or field of view and whether it matters to the claim.

MR3

Continuity and sequence

Document cuts, jumps, slow motion, combination, added sound, and other choices affecting time, cause, or dramatic impression.

MR4

Technical transformation

Disclose relevant filters, colour changes, compression, noise reduction, generation, reconstruction, and annotation.

MR5

Captioning and labelling

Check names, place, time, identification, translation, arrows, circles, labels, and other overlaid information.

MR6

Graphics and axes

Show scale, baseline, units, sample, weighting, missing values, and whether graphics amplify difference beyond the data.

MR7

Polling and representativeness

State population, sample, response rate, weighting, question wording, field period, margin of error, and limitations.

MR8

Registration and reconstruction

Distinguish photographed or recorded material from simulation, illustration, CGI, synthetic audio, generated images, or editorial reconstruction.

8

Headlines, selection, framing, dramaturgy, and editorial choices

No publication can include everything. Selection and form are therefore conditions of communication, not deviations in themselves. The audit does not ask whether representation lacks perspective; it asks whether perspective, relevance, omissions, and claim limits are visible and defensible in relation to the matter.

Headline, lead, image, sequence, repetition, music, conflict axis, and personification may guide interpretation before documentation is encountered. A product must be assessed both part by part and as a total presentation.

MH1

Headline correspondence

Test whether headline and lead remain within what the body, sources, and uncertainty markers actually support.

MH2

Selection criteria

Document why particular events, sources, numbers, quotations, and images were selected or prioritised.

MH3

Omission

Identify relevant material not included and whether omission changes understanding of the central claim.

MH4

Conflict and dramaturgy

Test whether conflict, personification, or narrative progression makes the matter more determinate than the material bears.

MH5

Balance and relevance

Distinguish relevant counter-material from mechanical equal time or false equivalence; weighting should answer to grounds and relevance.

MH6

Person and role

Prevent identity, appearance, group affiliation, prior matter, or social role from replacing the current action and documentation.

MH7

Placement of uncertainty

Do not hide decisive uncertainty, contradiction, or conditions at the end while headline and lead communicate certainty.

MH8

Overall impression

Test whether the combination of text, image, audio, graphics, and sequence communicates something stronger or different from the individual parts.

9

Distribution, platforms, algorithms, metrics, advertising, and sponsorship

Public visibility is not determined by editorial publication alone. Search engines, social platforms, recommendation systems, notifications, advertising, influencer networks, and sharing patterns affect what is encountered, in what order, and with which appearance of social confirmation.

Distribution systems may be proprietary or difficult to inspect. Limited access is not by itself proof of manipulation, but it bounds what can be established concerning ranking, representativeness, discrimination, causation, and corrigibility.

MD1

Distribution map

Register publication channel, platform, account, recommendation, notification, search, sharing, paid distribution, and later copies.

MD2

Ranking basis

Document known signals and objectives affecting visibility and mark what remains unavailable for examination.

MD3

Metrics and engagement

Distinguish clicks, views, viewing time, sharing, and comments from quality, truth, representativeness, and public importance.

MD4

Paid and editorial content

Clearly distinguish advertising, sponsorship, product placement, partnership, and financial interest from editorial content.

MD5

Ownership and interest

Map relevant ownership, funding, collaboration, and dependency without treating interest as automatic proof of error.

MD6

Algorithmic amplification

Test whether the system systematically amplifies sensation, conflict, misinformation, or particular actors and which material supports the finding.

MD7

Moderation and labelling

Document rules, decision basis, human or automated control, appeal, correction, and consequence of removal or labelling.

MD8

Visible correction

Test whether correction and reply reach the same channels, audiences, and copies that carried the original erroneous or insufficient representation.

10

Public discourse, participation, identity, moderation, and power

Public discourse extends beyond editorial media. Citizens, organisations, professionals, politicians, anonymous accounts, and platforms participate with unequal resources, risks, and power. The audit must distinguish the statement, person, role, network, reaction, and institutional consequence.

Criticism, objection, and sharp language are not by themselves error or capture. At the same time, person-labelling, pile-ons, harassment, coordinated exclusion, or identity reduction may move attention from grounds to social status. Such patterns must be documented concretely rather than merely felt or attributed as hidden motive.

MO1

Statement and person

Distinguish the concrete claim or action from total characterisation of the person, identity, or group.

MO2

Role and power

Map who can publish, rank, moderate, sanction, define categories, or set the agenda.

MO3

Conditions of participation

Document access, language, format, accessibility, identity requirements, deadlines, and risk affecting who can meaningfully respond.

MO4

Coordinated action

Distinguish spontaneous disagreement from documented coordination, automation, campaign, or institutional mobilisation.

MO5

Harassment and safety

Identify concrete threats, personal data exposure, repetition, contact pattern, and consequence without treating all criticism as harassment.

MO6

Moderation and reasons

Test whether rules are applied consistently, whether content or identity is judged, and whether relevant appeal and correction routes exist.

MO7

Labelling and credibility

Test whether labels such as expert, extremist, conspiratorial, activist, biased, or credible replace examination of the particular claim.

MO8

Corrigible public sphere

Assess whether errors, misleading impressions, and unreasonable omissions can be visibly corrected without participation or status becoming a barrier.

11

Combining audit types and findings

Media cases often require several audit types. A misleading headline may be a language finding; a clip may be a representation finding; polling may require model audit; source and expert status may require authority audit; and failed correction may be a practice or system finding.

A finding in one track must not automatically be transferred into another. A translation error does not prove system capture. An unreasonable moderation decision does not prove that all platform content is biased. A correct source statement does not prove that the headline or overall representation is fair.

The seven audit types in media and public discourse
Audit typeTypical objectExample of a bounded question
A1 LanguageClaim, headline, quotation, paraphrase, translation, label, and modality.Is the headline stronger than the documentation and body text?
A2 IdentityName, role, group, label, credibility, and participation.Is the person reduced to a category replacing the current action or claim?
A3 ModelPolling, estimate, ranking, recommendation, risk score, and statistical graphics.Which assumptions and validity limits sustain the number or ranking?
A4 RepresentationImage, audio, video, clip, graphics, layout, perspective, and metadata.What does the material show, and what is assigned beyond what was registered?
A5 SystemEditorial workflow, platform, ownership, metrics, moderation, and correction system.Can relevant correction change publication and distribution before error is propagated?
A6 AuthoritySource, expert, editor, fact-checker, institution, and consensus.Is status used as grounds or as a substitute for grounds?
A7 PracticeSourcing, verification, publication, moderation, reply, correction, and review.Does actual practice correspond to expressed editorial standards?
12

Correction, reply, archives, visibility, safeguards, and further programme

Correction must answer to the finding and reach the stage where the deviation arose. A typographical error does not necessarily require republication, while a false headline, misleading clip, or incorrect identification is not corrected merely by silently changing the body after distribution has occurred.

Correction, clarification, reply, withdrawal, new version, labelling, linking, and redistribution serve different functions. The choice must consider scope, harm, audience reach, archival value, source protection, privacy, and the need for a trace of what changed.

Further development should include appendices for source protection, live coverage, conflict and war reporting, polling, synthetic media, fact-checking, platform moderation, and public archiving. Such appendices require relevant editorial, technical, and legal competence.

  • Develop a dedicated source and publication-chain record with protected and public fields.
  • Build anonymised and synthetic cases distinguishing language, representation, model, system, and practice findings.
  • Develop a visible-correction standard across articles, podcasts, video, newsletters, and platform copies.
  • Test the method on both disputed and ordinary publications — including cases ending in F1.
MC1

Correct the source of error

Correct text, quotation, image, audio, metadata, graphics, numbers, classification, or workflow where the deviation arose.

MC2

Bound or withdraw the claim

Adjust certainty, scope, headline, and conclusion to what the material bears; withdraw where the grounds no longer stand.

MC3

Make correction visible

Mark what changed, when, why, and by whom and connect the correction to the original publication and relevant copies.

MC4

Reach a corresponding audience

Assess whether correction must be distributed through the same channels, notifications, or platforms that carried the original representation.

MC5

Reply and contradiction

Provide a relevant and proportionate opportunity to answer material allegations or error without transferring editorial control to the portrayed party.

MC6

Archive and version history

Preserve necessary public trace and documentation while considering privacy, danger, rights, and continued circulation of outdated error versions.

MC7

System correction

Where deviation arises from incentive, template, moderation, ownership, technical system, or workflow, test whether change reduces the same risk in new cases.

MC8

Review and close

Define responsibility, deadline, effect measure, side effects, appeal or renewed assessment, and criteria for closure.

Grounds and references

  1. DET SOM ERFoundational work for the relations among actuality, language, representation, authority, systems, and practice.
  2. Corrigible RealismPhilosophical placement of correspondence and corrigibility.
  3. Reality AuditMethodological overview and common principles.
  4. Audit StandardNormative requirements for documented audit.
  5. Applicable media law, professional ethics, source protection, privacy, copyright, platform rules, and editorial standardsMust be added according to medium, jurisdiction, and case; this document does not replace them.
V

Revision history

Document version
1.0
First published
18 June 2026
1.0

First public edition.