The #DutchArms investigation: Where do Dutch weapons end up?

Results and lessons learned from an open source investigative bootcamp

Picture produced by KRO-NCRV data.

In January, a collective of journalists set up a makeshift newsroom in Amsterdam. They opened their laptops and set to work, for two weeks straight, in the #DutchArms bootcamp. The aim was simple: to track and trace the Dutch global arms trade.

The Netherlands is a major international arms dealer. Since 2004, the export value of this industry has accumulated to almost 15 billion euros. This amount represents thousands of deals with 122 different countries. The business is great for the national economy, but it also implicates risks: after all, weapons, even the legally traded one, remain war-fighting tools. The Netherlands does not want these weapons to end up in the wrong hands. We started the track and trace program to find out whether they have.

Starting off: analyzing the export permits

When the Netherlands joined international negotiations for the International Arms Trade Treaty, the Dutch government positioned itself as an advocate for transparency. The state has taken it upon itself to publish monthly updates of its licensed military exports.

At first glance that monthly spreadsheet gives us information on the wares, customers and amounts involved. But the licenses list also indicates where the weapons export control ends: as soon as the goods have left the Dutch distribution points.

It doesn’t tell us whether the weapons systems were used, and if so, where and by whom. Would the Dutch government know about these things? Could this information be known, even without the state’s explicit disclosure of such facts?

Enter the open source verification challenge the #DutchArms team set itself: to assess to what extent post-export controls can be done. The team tried to follow up on the whereabouts of over a dozen different deals, including:

  • armoured vehicles sold to the Philippines
  • patrol ships, to Libya
  • radar systems, to Turkey
  • Fennek armoured vehicles, to Germany and Qatar
  • trucks and military vehicles, to Sudan and Rwanda
  • components for F16 fighter jets, to the US
  • patrol vessels, to the UAE
  • communications and radio systems, to Saudi Arabia
  • components for MQ-9 Reaper drones, to the US
  • components for armoured vehicles, to Turkey
  • components for Apache helicopters, to the US
  • armoured vehicles YPR-765, to Egypt
  • dogs, to Israel
  • M109 Howitzers, to the UAE

Investigation: the results

Taking freely published and accessible online information — from press releases, social media messages, video/photo uploads and satellite imagery — we have been able to verify the trajectory that some of the Dutch weapons have travelled.

Based on these open sources, we have found that:

  • Shipbuilder Damen skirts the Dutch licensing protocols to export patrol ships, which qualify as ‘military goods’, as civilian wares;
  • Communication equipment developed by Dutch-French company Thales is installed in tanks that belong to the Saudi Arabian army, which are undergoing a modernization programme and are deployed in the war in Yemen;
  • Multiple sales to Turkey include components for the maintenance of armoured vehicles, of the same type of those vehicles that have been donated to a division of the Free Syrian Army by the Turkish defense forces;
  • M109 Howitzers that were sold off to the United Arab Emirates, are now deployed in the war in Yemen — a conflict in which the Netherlands has refused to be an actor;
  • YPR-765 armoured vehicles, which were supplied to the Egyptian army in the 1990s, are deployed in Egypt’s war against IS. Some of them have been taken by IS and used against the Egyptian army;
  • Components produced for Reaper drones and Apache helicopters, supplied by Dutch companies, are employed with the ‘finished products’ in conflicts that the Netherlands does not endorse.
Research by the #DutchArms team concluded: geolocation confirms the presence of at least three M109 Howitzers of the United Arab Emirates in southern Yemen in December 2015. These Howitzers were sold to the UAE by the Netherlands in the 1990s.

With thousands of permits that could potentially warrant a follow-up, a strict scope — dictated by the open source feasibility — proved very useful.

Even as the research was guided by this scope, several research leads led us to complications that appear characteristic to the nature of the traded goods: components that are built into larger weapon systems by the original buyer, become ‘lost’ and cannot be traced as individual pieces.

Such cases present a combined challenge of ‘verified speculation’: the weapon system can be easily detected on open sources imagery (satellite images, video uploads), but it becomes impossible to ascertain that this particular carrier operates on the Dutch component concerned.

This issue complicated several leads, with both parts that were delivered for hardware, as well as the Dutch communications systems. Equally confusing were the permits themselves. While at first glance seemingly transparent, this administration does not lend itself for third-party scrutiny. Several licenses only carried vague information on the exported goods or the intended end-user, making selected searches impossible.

In other cases, however, we have been able to combine company and end-user information to fill in ´holes´ in the lists or even conclude invalidities in the control system, as when we proved the licensing issue at Damen Shipyards.

The method: lessons learned

What made our #DutchArms research effective was the combined effort of what we called our bootcamp: a two-week makeshift newsroom with experienced journalists, open source specialists and regularly visiting subject experts. By working in the same place, with a shared focus, we created a sense of community that propelled our research forward and resulted in a very steep learning curve.

To help you learn from our experience, here are the most important take-aways:

  1. Be willing to invest in preliminary research. You need a set of very specific research questions if you want to work effectively, especially if you want a big group of people to collaborate. And even more so if you want to engage online communities.
  2. Do not work alone, but share your questions, results and dilemmas. It will help you keep a relevant focus and makes your learning curve a lot steeper. And you have someone to share successes with.
  3. Archival discipline is crucial. When starting it might seem easy to digest your open sources and the information streams, but within hours you don’t know what you’ve seen before, whether it was relevant or if someone else already found it. So make sure to use a clear spreadsheet to archive the findings.
  4. Protect what you’ve found. Make sure to download or archive the information you’ve collected so it doesn’t get lost when the original owner decides to take it offline. We recommend Archive.org to keep a historical record of the websites you’ve perused.

There are plenty of online tools to help your open source research. These tools proved most useful to us:

  • Google maps can be of great value if you look for specific coordinates, often crucial in verification processes (tap ‘what’s here’). It can also give you geolocated photos that can give you relevant visual clues. If you need more specific information go to Wikimapia or use Terreserver (paid)for high-res sattellite imagery;
  • To gather messages, text-based information and eyewitness accounts from a certain relevant geographical area you focus on: search for geolocated social media updates. Echosec or WarWire are excellent tools for this task;
  • Languages don’t have to limit your research. Messages, searches or social media updates can easily and be translated using Google translate. It opens up the web to more local social updates and media reports;
  • It is important to authenticate and verify videos and photos you think might serve as evidence in your investigation We we very pleased with the reverse image search supported by RevEye (a browser plugin) and the Amnesty dataviewer to find out more about the history of a youtube video;
  • If you’re looking for more detailed or technical search and verification tools, check out this exhaustive list that is kept up-to-date by the members of the Bellingcat community.

If you want to know more or think about organizing a journalistic and investigative bootcamp yourself: feel free to drop us a line at info@lighthousereports.nl


The #DutchArms bootcamp was conceived by the Lighthouse Reports journalism collective. We teamed up with Bellingcat and the data journalism team at Dutch broadcaster KRO-NCRV.