On July 1, 2026, CBS organized its first Hackathon for Journalists. The event was hosted by CBS, Momus, and the Open State Foundation. The day almost had to be cancelled again: it was originally planned earlier, but then a fire broke out in the CBS data center in Almere. And on June 30, the day before the hackathon, StatLine went down and the CBS guest Wi‑Fi wasn’t working. In the end, everything worked out, but …..
The first Hackathon for Journalists brought together journalists, data analysts, developers and CBS experts to collaborate on Living and Energy (Wonen & Energie) datasets. Some participants had already attended a workshop on the Ai-tooling that would be used during the hackathon. The CBS experts normally focus on producing statistics and therefore rarely have time to write publications about the insights they encounter in the data.
Everyone present at the hackathon could pitch a topic to be explored as a journalistic piece. However, most of these ideas could not be carried out during the event because the available data lacked the necessary detail level. Many of the proposed journalistic investigations would require microdata to answer the research question properly. Due to CBS legislation (CBS-wet), only knowledge institutions (kennisinstellingen), such as universities, planning agencies, and similar organizations, are allowed access to micro‑level data. This means that if I, as a journalist, want to pursue one of the pitched ideas, I would need to collaborate with a knowledge institution to produce the journalistic piece, or rely on the same (public) sources CBS uses to create its statistics. As a side effect, I would then be verifying the accuracy of those CBS statistics.
I decided to join the group that wanted to investigate a neighbourhood with social housing in Almere that had been renovated, with residents promised lower energy bills and homes upgraded to energy label A or higher. The group examined CBS data, some of it was useful, but other parts required micro‑level data, which was not accessible. Because of this limitation, the group turned to other open data sources and discovered that the housing corporation had not fulfilled its obligation to report the energy labels of its homes to the institution responsible for collecting this information every ten years. The data simply wasn’t there. Rome wasn’t built in a day; to obtain the missing information, we would need to contact both the housing corporation and the energy‑labeling authority.
Part of Statline was unavailable during the hackathon, the CBS guest Wi‑Fi wasn’t working, and the AI tooling is still in its beta stage, yet every group still had to present a demo at the end of the day. Each team showed what they managed to accomplish in just one day and explained the obstacles they encountered, in addition to the issues already mentioned earlier in this blog. When using AI tools, verification of the data is essential. The AI tooling also requires significant computer storage and processing power to run effectively. Even with AI, Rome still cannot be built in a day. Journalistic pieces based on data demand thorough research, careful verification and this costs time.