Emergence of Homophily in Computational Life
November - December 2024
Introduction
In this project, I explored whether homophily — the tendency of similar individuals to form connections — can emerge purely from agent selection mechanisms, without explicitly coding in similarity preferences. The work was inspired by Blaise Agüera y Arcas et al.'s paper, "Pre-Life Dynamics in Computational Universes", which demonstrated how self-replicating programs can arise from randomness.
My contribution is a detailed theoretical framework to extend their model to social interaction dynamics. Although I haven’t implemented the full simulation yet, the proposed architecture lays the groundwork for isolating and analyzing how homophily might emerge from randomness.
Can homophily emerge without explicitly coding it into the simulation?
Is the presence of a selection mechanism sufficient for homophily to emerge?
This project was completed as part of the COMS 6998 Networks and Crowds course taught by Augustin Chaintreau at Columbia University. The full research proposal PDF can be found below.
Full Report
What I Learned
- Complex systems and emergent phenomena are the coolest thing ever. Cells become minds, minds form civilizations — each layer is an echo of the simple, local decisions made by the system beneath it. The pattern doesn’t stop with us. Why assume we’re the final step?
- Structure and agency both shape networks, but figuring out what matters more is super difficult.
- Simulations are philosophy disguised as code. The assumptions you make shape the results.