iSociety: An Artificial Society of San Francisco for Policy Analysis

Inspiration

San Francisco is one of the most complex cities in the world, facing challenges in housing, inequality, transportation, and climate adaptation. Reading about how policymakers often rely on aggregate statistics and limited pilot programs, I wondered: what if we could experiment safely before implementing policies in the real world?

That question led me to build iSociety, an artificial society of San Francisco designed for policy analysis through simulation. The idea was inspired by agent-based modeling in economics, but I wanted to push it further, grounding the agents in real-world demographic, income, and behavioral data from the U.S. Census (ACS) and SF Open Data, so the model isn’t just theoretical, but reflective of the city itself.

What I Learned

• The power of agent-based modeling: I saw firsthand how simple local rules can lead to unexpected system-wide dynamics.
  
• The importance of real data: Grounding the simulation in ACS and census data revealed patterns I never would have anticipated if I had used randomly generated agents.
  
• Policy trade-offs are unavoidable: For example, increasing housing subsidies improved affordability but also increased congestion in certain districts, highlighting the complexity of policymaking.
  

Challenges

• Data integration: Translating raw census and city datasets into meaningful agent attributes was technically demanding. Different datasets used different geographic boundaries, and merging them required careful preprocessing.
  
• Balancing realism and simplicity: A fully realistic simulation is impossible; the challenge was deciding which features mattered most without overwhelming the system.
  
• Computational scale: Simulating hundreds of thousands of agents pushed the limits of my local machine. I had to implement efficient data structures and parallelization strategies to keep run times reasonable.
  

Impact

With iSociety, policymakers and researchers can test ideas in a low-risk digital environment before implementing them in reality. Beyond San Francisco, the same framework could be extended to other cities, making simulation-driven policymaking more accessible.

iSociety is not just a prototype; it is a vision for how AI, data, and computational modeling can help society make better, evidence-based decisions.

Built With

• node.js
• openrouter
• python
• react