For over a decade, Airwars has followed a multi-stage research process to document thousands of incidents of civilian harm across more than 10 conflicts. These steps, outlined on our Casualty Recording page, include researching and assessing.
Alongside all the underlying sources for incidents published in our incident archive, Airwars’ Source Library also offers users first-look access at sources linked to particular harm events identified by our research team since Spring 2025. These sources have been linked to a particular event by our specialist open source research team, but are still undergoing the full analytical process that will eventually bring these events into our public archive.
This has been made possible by the introduction of a new and advanced digital archiving tool developed by longstanding Airwars web developers, Rectangle. The new tool allows the research team not only to digitally preserve each online source, but also to add some provisional labels and tags – such as the number of individuals reportedly killed as recorded in each source. These additional fields, alongside the text and media preserved through this tool, are now displayed in the ‘Sources’ and ‘Source media’ sections on each incident page. Sources archived prior to Spring 2025 will contain a link to open source web archivers, like archive.is, and will not contain these additional fields.
The source library is intended to maximise our ability to share our research with others, acknowledging that we are all now operating in an information environment that requires significant resource to comb through. We intend to help others with their own research and investigation by sharing our source identification process as soon as possible, advancing our timelines particularly for large scale and rapid response projects. As a support tool for a trusted ecosystem of actors, we request that both the source library and the outward links to ‘Researched’ (i.e. pre-publication) incidents are not shared publicly and are only circulated amongst known actors. Each page will come with a terms of use – violation of this trust-based system will result in a withdrawal of access to the Library.
‘Researched’ vs ‘Assessed’ incidents
Researched incidents are those where our open source researchers have identified an original claim of harm defined by a discrete date and location, and then found all open sources which refer to that claim. This work involves the painstaking task of carrying out advanced searches across social media sites, using key words often from an original harm claim.
In doing so, the researcher may discover the names of those individuals killed or injured, images from the site of the strike, testimony from those who survived the strike, etc. All of these sources are then digitally archived and linked together under a single event code.
Assessed incidents are those where our wider team of casualty recording specialists will review all of the sources that have been collected by the open source research team. This process follows a strict handbook and methodology, and adheres to the Standards of Casualty Recording. This includes summarising in narrative form the events of the strike, recording the range of those killed and injured and identifying the extent to which sources attributed the strike and resulting harm to the same belligerent. This process also involves where possible geolocating each event in line with Airwars’ graded geolocation methodology. Everything prepared by this team goes through a minimum of three rounds of review.
All values assigned to these events can be found in Airwars’ publicly available codebook.
Each of these incidents are made available in Airwars’ public incident archive as soon as they have completed their reviews.
As such, there are some key differences between a researched and assessed incident that should be kept in mind when navigating the Source Library.
In working with open sources, Airwars aims to collect all sources which attest to a specific incident of harm, and/or those reportedly impacted. Some sources collected by Airwars and included in the Source Library may contain insensitive, offensive, or even violent language. While Airwars collects sources, Airwars does not endorse insensitive, offensive, or violent claims.