netseg: Free Tool Measures How Divided Your Online World Is

Two researchers, Onur Tuncay Bal and Michał Bojanowski, have published netseg — a free Python package, meaning a set of ready-made building blocks that programmers can plug into their own work. It measures two things in any network of connections: polarization (how sharply a community splits into rival camps) and segregation (how much different groups keep to themselves). Think of a social app, a town, or a workplace, and netseg puts a number on how separated it is.

That sounds niche, but the gap it fills is real. The statistical formulas behind these measures are famous and widely cited, yet the code implementing them is rarely shared and almost never tested for errors. netseg is documented and tested, and it improves on the usual approach in two ways: it works with more than two groups (most tools only compare two sides), and it handles one-way connections, like who follows whom but not back. It also adds newer measures, including Random Walk Controversy and Moran's I.

Speed is the headline claim. The package relies on igraph, a well-known engine written in fast compiled code, so calculations run "orders of magnitude" faster than other free options — meaning minutes instead of hours on large datasets. It's also more honest about comparisons: most measures judge a network against a purely random one, and netseg lets researchers pick a more realistic baseline while avoiding a common double-counting mistake.

To show it works, the authors ran every measure across a simulated model of shifting opinions, then applied them to a county-level railroad network built from nineteenth-century operator records joined to full census data. The honest caveat: this is infrastructure for researchers and data scientists, not something a regular person can use today. It requires network data and coding skill, and it measures division rather than fixing it. Still, better measuring sticks tend to lead to better evidence — and eventually better decisions about echo chambers, misinformation, and segregated communities.