Methodology

Sources

Every significant claim is linked to one or more identified sources. For each source we record its type, author or institution, date, language, means of access and the function it serves as evidence.

No source is treated as accurate simply because of where it comes from. A government report, an academic study, a participant's memoir and a court judgment each carry different value and limitations. For that reason every source record includes a note on its perspective and limitations, and distinguishes the primary document from later analysis. When a source cannot be consulted directly, we create only a bibliographic record and flag it for editorial verification.

The archive does not reproduce complete books, paywalled articles or copyrighted reports. We link to the original publisher, use short quotations only when necessary, and prefer paraphrase with precise citation. Sources hosted by the archive are marked separately and published only with permission.

Conflicting evidence

When serious sources disagree, we do not quietly choose the most convenient version. The positions are set out side by side, with their sources and reasoning, and the claim is labelled “Disputed”.

We distinguish disagreement over facts — a date, a figure, the identity of perpetrators — from disagreement over interpretation, such as the causes or significance of an event. In the first case we look for further sources and state what would settle the question; in the second we present the arguments and do not turn a debate into a false conclusion. A single event can include both judicially established elements and elements that remain disputed; these are kept clearly apart in the text.

Testimony

Personal testimonies are invaluable sources of lived experience. We publish them only with documented consent, label them “Personal testimony” and compare factual details with other sources.

Memory changes over time and is shaped by trauma, by later retellings and by political context. This does not diminish testimony, but it calls for care: we do not rely on a single testimony as the only evidence for a contested claim, and we do not alter a witness's account to fit another version. Where accounts differ, we say so openly.

We never create first-person quotations. Where no verified transcript exists, we provide an editorial summary and label it as such. For living witnesses, names, places of residence and identifying details are published only according to the level of permission they have given.

Court findings

Court decisions are presented with their exact procedural status. An indictment is not proof of guilt, and a first-instance judgment is not final while an appeal is possible or pending.

The label “Judicially established” is used only for findings contained in a final judgment, and only for what the court actually found. A judgment may find, for example, that a crime occurred at a given place and date without ruling on other events. The status of every case is checked against the court's official website, and the date of the last check is shown next to the case.

Criminal responsibility is individual. A conviction for acts committed by some members of a structure is not a conviction of a people, of a war or of every member of an organization. Likewise, an acquittal does not necessarily mean that a crime did not happen, but that the accused person's guilt was not proven.

Historical interpretation

Historical interpretation is an argued reading of sources, not a single fact. We label it clearly, link it to the scholars who hold it and, where other serious readings exist, set them side by side.

Questions such as why the strategy of peaceful resistance did not produce a political settlement, how much weight the KLA and NATO carried in ending the war, or how Rambouillet should be understood, have no single answer accepted by everyone. This archive has a declared editorial position — it recognizes the importance of Albanian resistance — but that position does not replace evidence and does not hide the counter-arguments.

Corrections

Mistakes are corrected openly. Every significant correction is noted on the page concerned, with the date, what changed and why; texts are not quietly altered after publication.

Anyone can report an inaccuracy through the contribution page, stating the page, the problem and the source supporting the correction. The editors check the report against the sources, consult a second reviewer on sensitive issues and decide on the change. Minor language corrections are made without a separate note; corrections of facts, figures, legal status or attribution are always noted.

Translation

Albanian is the archive's primary language. English translations are editorial rather than word-for-word, and are reviewed by a second person before publication.

Where an English translation is missing, the page shows the Albanian text with the note “English translation in progress.” and does not substitute machine translation. Legal terms follow the official terminology of the courts concerned, and institutions are given their official names. Place names keep their preferred Albanian form, with alternative names recorded for search.

Image permissions

Every image records its creator or institution, date and place where known, source, licence, permission status and alternative text. Images without verified permission are not published; clearly labelled placeholders are used instead.

We do not download random images from search engines, and we do not use AI-generated images as historical evidence. When the identification of a photograph — place, date or people — is uncertain, this is stated beside it. Graphic material is published only with a warning, documented publication rights and a written editorial justification, and is shown only after a deliberate action by the reader.

Uncertainty: dates, places and figures

Uncertainty is not hidden behind confident wording. Approximate dates, approximate coordinates and figures that vary by source are labelled as such, with the relevant source and definition.

On the map, every location states its precision: exact, approximate or municipality level. We draw no boundaries or zones without verified cartographic data, and refugee lines are shown as schematic. For dates, we distinguish a known day from a known month or year; an event lasting several days is recorded with its start and end.

Evidence labels

The archive uses five labels: “Judicially established”, “Documented”, “Personal testimony”, “Disputed” and “Historical interpretation”. Each has its own shape and text, so that none is distinguished by colour alone.

A label is attached to a claim, not to a topic. The same event may carry “Documented” claims, such as its date and place; “Judicially established” claims, such as the responsibility of a person convicted by final judgment; and “Disputed” claims, such as the exact number of victims. The “How do we know this?” panel shows, for each claim, its sources, status and date of last review.

Language and the attribution of responsibility

We make no collective accusations against Serbs, Albanians, Roma, Bosniaks or any other group. When describing crimes, we name the responsible institution, unit, authority or person as precisely as the sources allow.

Formulations such as “Serbian police forces”, “Yugoslav Army units”, “state-security structures”, “KLA members accused in a specific case” or “unidentified armed individuals” are more accurate than ethnic labels. When a source does not identify the perpetrator, we do not fill the gap with speculation. Individual responsibility, institutional responsibility, historical interpretation and collective identity always remain separate concepts.

Place names

The preferred Albanian name appears first. Alternative names — Serbian, historical or other transliterations — are kept in the place record and used in search, so that readers can also find sources written in other languages.

For example, Reçak also appears in sources as Račak, Gjakovë as Đakovica, Pejë as Peć and Drenas as Glogovac. International court documents often use Serbian forms; when citing a document we keep its usage in the title and in quotations, while the editorial text uses the Albanian form.

Figures and casualties

We do not invent figures, round them for effect or pick the most dramatic one. Where figures differ, we give the range, the definition, the time period and the source of each.

The number of people killed, missing or expelled depends on the definition — civilians or combatants, killed or missing — on the period covered and on the date of the count. Figures are therefore always cited with these elements, and when a source updates its own figures we note the year of each. The archive does not reproduce casualty databases without a licence and verification; it links to projects that have documented names using a public methodology.

Sensitive material

War, death, sexual violence and disappearance are subjects that touch living families. Distressing content carries a warning, no media plays automatically and graphic material is shown only after a deliberate action.

We do not publish private addresses, telephone numbers, medical information or details that identify survivors of sexual violence, children or protected witnesses without clear authorization. The remembrance space is kept apart from the analytical pages, and our language aims at people's dignity, not at effect.

The review process

Every text passes through four stages: editorial draft, verification against sources, review by a second person, and publication with the date of the last review.

The initial content of this release has been drafted from the cited sources, but many materials are still awaiting full editorial verification; this is clearly marked on every page. Court cases are reviewed whenever their procedural status changes, and all pages are reviewed periodically. The checklist below is completed before any article is published.

Wikipedia and social media

Editors may use Wikipedia privately, and only as a tool for finding sources; it is never displayed as a source and never cited as evidence.

Social-media posts are not used as evidence for historical claims. They may be an object of study — for example, of how an event is remembered today — but never a source for the facts of that event. When a claim circulates widely online without a verifiable source, we do not repeat it.