City Opportunity Simulator

A city can fund a program without residents ever hearing about it. Residents can complete training without employers having enough openings. This Python/Mesa model makes those gaps visible.

MESA 3.5.1 RUNS · SEED 42

Budget → awareness → skills → work

Budget $600kOutreach 8%Training seats 36Openings 90Run folder workforce_001-baseline-seed42

Residents reached21991% of population

Completed training18210 still enrolled

Employed12653% of population

Budget remaining$0kafter 16 weeks

Run artifacts2923runs/workforce_001-baseline-seed42

AwareIn trainingTrainedEmployed

Week 0Week 16

Generated by the Python/Mesa source model. Rebuild with generate_mesa_results.py; it writes run folders plus the index consumed here.

The model follows 240 Mesa resident agents for 16 weeks. Choose a documented scenario to inspect the actual output generated by the Python model. Each scenario is backed by a dedicated run folder under runs/ with its own config, metrics, agent states, events, and narrative beats.

The standalone internal dashboard replays those same artifacts as a navigable world for agent following, system observation, and report development. See the research cockpit and world roadmap for the boundary between Mesa, Phaser, Codex-assisted world building, and public documentation.

Read the model

• ODD protocol documents the purpose, entities, behaviors, assumptions, and schedule.
• Code notes explain the deterministic engine and extension points.
• Tutorial walks through a first experiment.
• Download the tutorial PDF for a printable version with both captured runs.

This is a learning model, not a forecast. Its parameters are explicit hypotheses that should be replaced with local program data before using the results to support decisions.