AAGE — Amhara Advocacy Group in Europe
Documentation, research and advocacy for civilian protection in Ethiopia's Amhara region.
How sources are verified, how survivors are protected, how the limits of our reach are acknowledged, and the editorial standards that govern this work.
This page describes how the consortium works. It is intended for researchers, journalists, parliamentary staff, civil society partners, and any reader who needs to evaluate the rigour of what we publish before relying on it. The standards below apply to every investigation, every documentary, every report, and every record in the Drone Strike Monitor dataset.
We have written this page in plain language and at moderate length, on the principle that a methodology page is most useful when it can be read end-to-end in a few minutes. Where a section is shorter than a reader might expect, that brevity is deliberate; longer technical handbooks belong in the field-of-study literature, not on a public methodology page.
We rely on institutional sources first. The Office of the United Nations High Commissioner for Human Rights (OHCHR), the Ethiopian Human Rights Commission (EHRC), the U.S. Department of State, UNICEF, the World Health Organization, Physicians for Human Rights (PHR), Amnesty International, Human Rights Watch (HRW) and the Lemkin Institute for Genocide Prevention provide our baseline. We rely on press reporting from the BBC, Reuters, AFP, Addis Standard, Ethiopia Insight, The Reporter Ethiopia and other established outlets. We rely on partner diaspora documentation — in particular the continuous record kept by the Amhara Association of America (AAA) — for incidents that have been verified across multiple parallel networks.
Each contested factual claim that appears in our publications is cross-verified against at least two independent sources. For figures — casualty counts, displacement numbers, school closure totals — we cite the lowest verified value across our sources, not the highest. Where institutional sources disagree on a figure, we cite the divergence and let the reader see the range. We do not select between conflicting institutional records to favour the more striking number.
Where our reach is limited — and we say where it is limited — we either say so explicitly or we do not publish the claim. The Amhara region has been under regional telecommunications blackout since August 2023. Independent press access has been severely restricted. This is a forensic constraint on what we can know, not a stylistic choice. It is part of what we are documenting.
We are bound by the "Do No Harm" principle as articulated in the Documentary Filmmakers’ Statement of Best Practices and by the trauma-informed standards developed by the Dart Center for Journalism & Trauma. Where a tension between visual or narrative authenticity and the safety of a source or survivor has arisen during production, safety has always been chosen.
Concretely, this means the following: a survivor’s name is used only when it is already a matter of public institutional record (for instance through a BBC investigation, a Human Rights Watch report, or AAA documentation), or when it is provided to us with the explicit consent of the survivor or, in the case of a minor, of the survivor’s family. A survivor’s image is shown only with the same standards of consent. Medical detail is included only at the level already published by the underlying institutional source. Pseudonyms are used where appropriate, and identified as pseudonyms in the published material.
For sources who wish to communicate sensitive information to the consortium — witnesses, healthcare workers, teachers, displaced civilians — we do not currently operate a public end-to-end encrypted intake channel. A dedicated channel is in preparation, with end-to-end encryption (Signal) and a published PGP key for sensitive correspondence. Until that channel is published, sources should not transmit material that requires source protection through standard email or our contact form. We can be reached through trusted intermediary organisations on request.
For each entry in the Drone Strike Monitor dataset, the verification standard is documented in the dataset itself: incident date, location, source institution(s), method of cross-verification, and a confidence indicator. Records are versioned. Where a record is updated — because new institutional documentation has clarified figures, identified victims, or revised the date of an incident — the prior version remains visible in the change history.
For long-form investigations, each numerical claim in the published text is footnoted to a source. For documentary films, each figure read aloud in the narration is anchored to a source listed on the film’s editorial brief page on this site. Readers who want to verify a specific claim can therefore trace it directly to the institutional document or press report that supports it.
The figures we publish have been deliberately conservative. Where the AAA records 1,000 documented drone-strike civilian deaths and an institutional source records 800, we cite 800 with attribution and note the AAA figure with attribution. The point is not to minimise the harm. The point is to publish numbers a parliamentary office or a journalist can rely on without later finding that we had reached for a higher figure than the institutional record supports.
Each publication on this site — investigation, documentary, statement, dataset update — is reviewed by the consortium’s editorial coordination before publication. Editorial decisions are taken within the consortium’s research team, not voted on by the Federation’s member associations. Member associations supply field information, witnesses and translators; they do not direct the editorial line. This separation is constitutive of the project.
If you find an error in something we have published — a mistaken date, a misattributed quotation, a figure that has since been corrected by its institutional source — please write to [email protected] with the specific reference. We aim to acknowledge correction requests within five working days. Verified corrections are made publicly: the published page carries a corrections note, and the change log is preserved in the site’s public version-control history on GitHub.
We do not silently update publications in ways that would change what a citation of our work would have referred to. Substantive corrections are dated and disclosed.
This site is built with Astro (an open-source static-site generator), styled with native CSS, and deployed on Cloudflare Pages. Maps are built with the open-source Leaflet library on top of OpenStreetMap tiles. Documentary films are edited in Adobe Premiere Elements, with HTML/CSS/JavaScript-based infographics captured through OBS. We use open-source or correctly-licensed b-roll and stills throughout. Wherever a credible open-source alternative exists for a tool we need, we prefer it.
Our films use synthetic voice narration generated by ElevenLabs. We disclose this here, on each film’s editorial brief page, and in the films’ production credits. We use synthetic voices for three reasons. First, the consortium has no paid staff and the volume of multilingual narration we need to produce exceeds what volunteer human voice talent could sustainably provide. Second, using a recognisable diaspora voice for narration on sensitive material would create avoidable risk for the individual; using an institutional synthetic voice does not. Third, synthetic voice is a transparent tool, not a deception — we credit the voice on every film, and we do not represent the voice as that of any specific human source. The research, writing, fact-checking and editorial decisions remain the responsibility of the consortium’s contributors, not of any AI tool.
We do not use AI to generate the content of investigations, the figures published in the Drone Strike Monitor, or the editorial line of the consortium’s statements. AI tools assist with translation drafting, transcription, code, and visual layout. The intellectual responsibility for what is published is human.
The consortium is independent of states, of political parties, and of corporate funding. It receives no funding from the Ethiopian government, from any other state party to the conflict it documents, or from any commercial media organisation. It accepts no advertising on this site. It has no contractual relationship that would constrain its editorial line.
The consortium operates on volunteer labor. Production costs are absorbed by the Federation’s member associations and by the personal time of contributors — researchers, translators, designers, civic activists. We disclose this because it shapes what we can produce: a slow editorial cycle, a lean publication slate, careful pacing rather than continuous output. The constraint is part of how the work should be read.
We are not a press agency. We do not break news; we publish edited research after verification. We are not a litigation body. We do not represent victims in legal proceedings; we provide documentation to those who do. We are not a humanitarian aid organisation. We do not deliver assistance in the field; we point readers to the institutions that do, including OHCHR, EHRC, PHR and partner NGOs.
For questions about a specific source, a specific figure, or our methodology in general, write to [email protected]. For corrections, write to [email protected] with the specific reference. The full Contact page lists all the channels available to readers, journalists, partners and witnesses.