Insights from our meetup with Marie Skau and Inge Maria Falbe.

When we talk about defending democracy, we rarely talk about bipartite graphs or the Louvain algorithm. But at our recent event hosted by DemAI and Next Generation Democracy, a data-driven approach to understanding policy-making took center stage.

Marie Skau and Inge Maria Falbe, students from the IT University of Copenhagen, joined us to walk through a network analysis project they developed in collaboration with DemAI. By mapping an entire year of EU lobbying data—encompassing over 28,000 lobbying organizations and 648 Members of the European Parliament—they showed how software design can be used to monitor complex political ecosystems.

Their research presented an important perspective: Influence isn't just transactional; it's relational. In the following, we have highlighted the central insights from their presentation.

Beyond the Volume of Meetings

To understand how this relational influence works, we first have to look past the sheer volume of data.

If you evaluate the data purely on the number of meetings held with politicians, tech companies appear to be the most prominent actors. Meta, Apple, and Google log the highest number of overall meetings in Brussels.

However, Marie and Inge explained that raw volume isn't the same as strategic coordination. When the researchers applied a mathematical filter called "backbone extraction" to remove coincidental meetings and highlight meaningful connections, a different picture emerged. Tech companies dropped in the relative rankings, while heavy industries—such as large chemical and energy sector associations like BASF—showed up as highly central. These groups operate with significant structural coordination rather than just a high volume of individual meetings.

Uncovering Transparency Gaps

A key takeaway from the evening was the identification of "transparency gaps." In data science, the conclusions you draw depend entirely on the datasets you analyze.

Marie and Inge demonstrated how looking beyond standard MEP (Member of European Parliament) meetings reveals less visible, yet highly impactful, channels of influence:

• Targeting the Commission: While MEP data shows broad sector clusters, adding European Commission data to the network reveals focused, coordinated activity around specific policy areas (like AI or the Green Claims Directive). This is where much of the detailed legislation is shaped, and access at this level is highly concentrated. The researchers found that the top 10% of actors have 80.6 times more connections at the Commission level than the bottom 10%.
• The Role of Memberships: Joining umbrella organizations is a substantial lobbying channel. For example, Hydrogen Europe has 96 declared members in the network, making it the single most connected organisation in the membership layer. Marie and Inge’s network graph showed that 84.7% of membership ties have no overlap with official meetings. This indicates that umbrella memberships function as a separate channel of influence that traditional meeting data doesn't fully capture.

The Danish Context

For a Danish audience, the EU data provides a relevant point of comparison.

The European Union has a transparency register and public meeting disclosures. While the system isn't perfect, this public data exists—which is exactly how Marie, Inge, and DemAI were able to map the network.

In Denmark, there are currently no systematic lobbying disclosure requirements and no public registers of political meetings. The takeaway is not necessarily that Denmark has more or less influence activity happening—it is simply that we lack the systemic visibility to map these networks locally.

Looking Forward

Despite the complexity of these lobbying networks, Marie and Inge ended on a constructive note regarding the role of accessible technology.

Network analysis is a useful lens. By utilizing open-source tools and public data, civil society can reconstruct and visualize these complex structures relatively quickly, helping us better understand how policy is shaped.

A massive thank you to Marie Skau and Inge Maria Falbe for joining us and collaborating with DemAI on this research.

Are you a student interested in using technology for democracy? At DemAI we are always looking for new talent! We love collaborating with students on projects that merge data science, software design, and civic tech. Let's build the next democratic lens together—reach out to us!

‍