11/
The work was led by Jolien Cremers, Benjamin Kohler, Benjamin F. Maier and done in collaboration with Stine N. Eriksen, Johanna Einsiedler, Frederik K. Christensen, @sunelehmann.com, David D. Lassen, Laust Hvas Mortensen. We are grateful for funding by the Villum Foundation.

10/
We build on earlier work by @drtomemery.bsky.social and colleagues, who constructed the Dutch administrative network 🇳🇱.
Our network adds temporal depth and enables computationally scalable analysis at the national level.

9/
This dataset opens new doors for research on:
– Inequality & opportunity
– Social cohesion
– Spatial structure
– Peer effects
…all grounded in real, registry-based, population-wide data.

8/
To handle 1.4 billion ties, we used a bipartite network representation — linking individuals to shared foci like workplaces, schools, or households.
This reduced edge count by up to 98%, making large-scale analysis feasible.

7/
Social connectivity is highly unequal:
The top 20% of individuals hold nearly half the ties in some layers.
Over time, the well-connected in one layer become central in others too — interlayer centrality accumulates.

6/
📍 Geography matters.
People are much more likely to be socially connected if they live nearby.
Probability of connection drops steeply after 100m — confirming that spatial proximity underpins social proximity.

5/
By combining and weighting layers, we uncover a familiar pattern: The average shortest path is around 6 — echoing the “six degrees of separation” concept.
But only when accounting for the structure and strength of ties.

4/
Social network position reflects income:
Higher-income individuals tend to have more ties — especially when past ties are included.
The more central you are, the higher you tend to be in the income distribution.

3/
We find that social ties persist and resurface — former classmates become neighbors, colleagues become family.
Connections across life stages create dense, overlapping social paths.

2/
We mapped social ties through five layers:
👨‍👩‍👧 Family
🏠 Households
📍 Neighborhoods
🎓 Classmates
💼 Colleagues
…across 2008–2021. The result is a dynamic, multiplex network of the entire population.