The Two-
Dimensional
Divide

Our choice of social ties depends on who we are: our gender, nationality, socioeconomic class, and more. These spontaneous preferences often go unnoticed, but they can unintentionally create inequalities. In particular, people in minority groups can end up with fewer connections—something that matters because social networks influence access to opportunities, support, and visibility. In this project, we built a mathematical model to understand how our multifaceted identities shape who we connect with—and what that means for inequality.

With our model, we simulate how individuals with different characteristics form social networks. But to make things more fun and accessible, let’s go back to our visualization example and imagine we’re simulating networks of cute animals instead of people.

Each group—cats, dogs, black animals, and white animals—has a preference for connecting with others, represented by a number between 0 and 1 called homophily: the tendency to prefer similar others. A homophily of 1 means the group only connects with others from the same group; 0 means they only connect with those from a different group; and 0.5 means they’re indifferent.

To build a network, we simulate random encounters between pairs of animals. When two animals meet, they look at their characteristics and evaluate their homophily along each dimension (species and color). For each dimension, they “flip a biased coin” based on the homophily level. If all the coins land in favor, a connection is formed.

For example, if homophily is 0.8 in both dimensions, that means that when two animals of the same group meet, the coin has an 80% chance of landing in favor of the connection. If they are from different groups, the coin has only a 20% chance.

So, if a white cat meets a black cat, they toss two coins: one with 80% heads probability (they’re both cats) and another with 20% heads probability (they have different colors). A connection is only made if both coins land heads. The more identity dimensions matter, the more coins must be flipped, and the harder it becomes to form ties across differences. Conversely, if there is only one dimension (like on planet Monos), only one coin is flipped.

This process creates a network that reflects individual preferences. Once the network is built, we analyze how connected different groups are. We’re especially interested in how often members of minority groups—those making up less than half of the population—end up with fewer links than members of majority groups. In our simulations, we focus on cases where the minority in one dimension (say, color) is also numerically smaller than the minority in the other dimension (say, species). This way, we can isolate the effects of group size and identity structure without being biased by which dimension is labeled as “species” or “color”—since the model treats them symmetrically. It also helps us understand how different forms of underrepresentation interact to shape network inequalities.