Cases and Documentation

Cases and Documentation is not a gallery of conclusions. The section separates pedagogical examples, protocols, public registers, archive functions, and possible actual cases so that each part has the correct status and reading mode.

An example may be synthetic and pedagogical. A case may rest on historical material, public documents, or anonymised actual material. Its status must be visible so that a constructed illustration is never presented as an actual event and a case presentation is never represented as more complete than its source access allows.

Publication is a separate methodological act. Material may be relevant to the audit without being appropriate for public release. Privacy, source protection, safety, rights, contradiction, potential harm, and public interest must be assessed separately.

A credible case requires several functions to remain distinct. Especially important is the distinction among material, a claim about the material, the auditor's analysis, and the final finding.

The case document is itself a representation. It must therefore remain auditable for selection, language, transformation, anonymisation, missing material, and conflicts of interest.

Every published case must be traceable through the same ten-part structure. Simple examples may be brief, while serious or complex cases require more material, fuller contradiction, and stronger publication safeguards.

Each stage must produce an identifiable output. Where a stage cannot be completed, the absence must be recorded as a limitation and must affect the strength of the finding.

1. S1

   Identity and status

   Assign case code, case type, language, owner, version, lifecycle status, and publication level.

   Case header with stable identifier and status date.
2. S2

   Mandate and questions

   Bound what is examined, why, by whom, during which period, and with which explicit exclusions.

   Mandate, principal question, subquestions, and limitations.
3. S3

   Parties, roles, and rights

   Distinguish person, role, institution, source, auditor, decision-maker, and affected party, and record rights and conflicts.

   Role and rights matrix with accountability and impartiality notes.
4. S4

   Source and material register

   Record origin, date, version, access, integrity, transformation, relevance, and limitation for each item.

   Numbered source register and material chain.
5. S5

   Claim and question matrix

   Divide the case into testable claims, references, grounds, counter-material, inferences, and open questions.

   Claim matrix with P-codes and status.
6. S6

   Audit type and analysis

   Select primary and supporting audit types A1–A7 and document why they are relevant.

   Analysis plan, domain requirements, and traceable examination.
7. S7

   Provisional findings

   Formulate provisional F1–F7 findings with grounds, strength, limitation, and conditions that could change them.

   Provisional findings matrix, not final publication conclusion.
8. S8

   Contradiction and response

   Give affected parties a real opportunity to correct facts, submit material, and answer relevant criticism before final findings.

   Response log, new sources, and visible treatment of objections.
9. S9

   Final finding and correction

   Revise findings after contradiction and set proportionate correction, responsibility, deadline, and control method.

   Versioned final document and correction plan.
10. S10

    Effect, revision, and closure

    Test implementation and effect, record what remains open, and decide whether to close or reopen the case.

    Review note, closure grounds, or new version.

The case code must remain stable across languages and versions. Title, summary, and anonymisation level may change; the identifier must not be replaced without preserving the former identifier in the revision history.

Status describes where the case stands in the work, not how credible or serious it is. Publication level describes who may see which material.

Lifecycle status

Publication levels

The source register must show where material comes from, what has happened to it, and what it can bear. A file, screenshot, or official document is not self-explanatory.

Copies of one origin must not be counted as independent sources. Missing material must be recorded together with who should have created it, whether it may have been lost, and what the absence can actually support.

The claim matrix prevents a large narrative from being treated as one indivisible assertion. Every decisive link must be testable against particular sources, counter-material, and inference rules.

Observation, reported information, interpretation, causal explanation, motive claim, legal characterisation, and normative assessment must be marked as different claim types.

Findings should be formulated after contradiction where possible and safe. A party response is not automatically true, but it is relevant material that must be tested and made visible.

Every finding receives an individual finding code and one F1–F7 category. F7 requires the complete capture threshold; disagreement, error, power, category, or non-response is not enough alone.

Publication must be examined as a separate action with its own effects. Openness may be important, but maximal publication is not the same as maximal reviewability or right action.

Anonymisation is more than removing names. Combinations of place, time, role, event, quotation, and rare details may permit reidentification. Excessive anonymisation may also remove context necessary to understand the case. Both risks must be documented.

A case may require correction at several levels: source, claim matrix, finding, measure, published text, and actual consequence. A visible textual change is insufficient if the error still governs registers, decisions, or practice.

New information must not be suppressed to protect the original presentation. At the same time, every new allegation should not automatically reopen the whole case without relevance and grounds testing.

The published architecture precedes public cases. This prevents the rules from being designed retrospectively to fit a particular opponent or conclusion.

The case and register section distinguishes synthetic learning examples, protocols, public register entries, archive/export functions, and possible actual cases. No actual case counts as published until it has its own code, status, contradiction, publication level, and revision history.