Municipal Elections 2026 in the Netherlands: Part IV
Analysis: Network visualizations
Directed graph
Which accounts attracted the most engagement? Who are the most active responders, and in which direction does the interaction flow?
In the directed graph, every line contains an arrow. The arrow shows the direction of engagement: it points from the creator of the post toward the person who responded.
In other words: if Account A posted something and Account B responded to it, an arrow is drawn from A to B. An account with many outgoing arrows (high out-degree) is a content creator whose posts attracted responses from many unique accounts.
In this network, for example: telegraaf.nl attracted 292 unique responders, DENK attracted 230, and so on. Larger dots in the graph therefore represent accounts whose content received responses from a larger number of unique accounts.
What High In-Degree Means
An account with many incoming arrows (high in-degree) is an account that responds to content from many different accounts. A responder with high in-degree is an active participant spreading its responses across a wide range of content creators. When combined with a high bot score, this pattern is worth investigating.
How to use this graph
Look for the largest dots — these are the accounts attracting the most engagement.
Hover over a dot to see: the account name, follower count, and bot score.
Follow the arrows leaving a large dot. The accounts at the end of those arrows are the accounts responding to that account’s content.
Pay attention to the border color around each dot (green, orange, or red ring).
In the directed graph: community membership is shown through the fill color + while the bot score is shown through the colored ring surrounding the dot.
See “Reading the Colors” for the meaning of each color.
Look for accounts with many incoming arrows (high in-degree). These are highly active responders.
If these accounts also have an orange or red border, they may represent suspicious responders interacting with a large number of different accounts.
Cluster Colours: In the directed graph, the fill color of each dot shows which cluster the account belongs to.
There are six clusters in this network:
-
16 accounts. Modularity score: 0.60.
Contains sports, fitness, and training-related accounts as well as several anonymous accounts.
-
18 accounts. Modularity score: 0.71.
Contains a mix of public broadcaster accounts (Nieuwsuur) and Forum for Democracy (FvD) affiliated accounts — a striking cross-ideological cluster.
-
18 accounts. Modularity score: 0.68.
Dominated by mainstream Dutch news media: telegraaf.nl, nu.nl, nosstories, omroepwnl. The largest media hub in this network.
-
13 accounts. Modularity score: 0.74
— the most internally cohesive cluster. Contains local political accounts (VVD Rotterdam, luna.hoogendam). Deserves further investigation.
-
Only 4 accounts. Modularity score: 0.72. An exceptionally small, very close-knit cluster. The identities and shared theme of these four accounts call for manual investigation.
-
18 accounts. Modularity score: 0.60.
Includes DENK (denk_nl), espnnl, and various regional and local media accounts. A mixed cluster centered on a political party.
Clusters in detail
A breakdown of each of the six clusters detected in the directed graph, including their size, cohesion score, and known prominent accounts.
What Modularity Measures
The modularity score measures how clearly defined a cluster is, on a scale from 0 to 1. A score of 0.74 means that the members of the cluster interact significantly more with each other than would be expected by chance. All six clusters score above 0.60, which is generally considered strong clustering. A higher score means a more self-contained bubble.
-
Contains sports, fitness, and training accounts alongside a few anonymous accounts. One anonymous account in this cluster has an exceptionally high red bot score (0.91). The lower modularity suggests that this cluster is less self-contained than others.
-
An unusual mix: includes both the public broadcaster Nieuwsuur and the right-wing populist party Forum for Democracy. This cross-ideological clustering is significant — it means that these accounts share a common interaction audience. This may reflect coverage of the FvD or audience similarity. Deserves further investigation.
-
The dominant mainstream media cluster: telegraaf.nl, nu.nl, nosstories, omroepwnl, and FvD politician Lidewij Devos are all in this cluster. This is the cluster with the highest reach in the network — it contains accounts that attracted hundreds of commenters. The co-clustering of an FvD politician with mainstream media is remarkable.
-
The most internally cohesive cluster. Contains local political accounts including vvd.rotterdam and luna.hoogendam. The high modularity (0.74) means that the members of this cluster interact almost exclusively with each other. The identity of all 13 accounts and the content they exchanged warrant manual investigation.
-
An exceptionally small, very close-knit cluster of only 4 accounts. High modularity means that these four accounts interact almost exclusively with each other. This pattern—a small, closed, highly cohesive group—may indicate coordinated or carefully directed involvement. Manual examination of these accounts is recommended.
-
A diverse cluster centered around denk_nl (DENK party), espnnl (sports), and various regional and local media accounts. The presence of a political party alongside unrelated accounts (sports, local news) suggests that the cohesion of this cluster is driven by shared audience engagement rather than ideological alignment.
-
The descriptions above (e.g., "mainstream media cluster", "local political corner") are based on observing which well-known accounts appear in each cluster. The algorithm itself does not assign labels — it groups accounts only based on interaction patterns. Many accounts in each cluster are anonymous or less well-known, and the full cluster composition requires manual inspection of all 87 nodes.