Data Pipeline: We assembled data from 11 federal and local sources: U.S. Census ACS 5-year estimates (tract-level demographics, income, housing), Oakland PD CrimeWatch via Socrata Open Data (1.26 million incident records, 2005–2024), City of Oakland FY2024-25 Adopted Budget, Bureau of Labor Statistics (Alameda County unemployment), HUD Fair Market Rents, EPA EJScreen environmental justice indicators, FEMA National Risk Index, GreatSchools school quality ratings (87 Oakland Unified schools), Legistar council meeting records (134 sessions, 2019–2024), and Census TIGER tract boundary files.
Council Meeting Analysis: 134 Oakland City Council meeting transcripts were processed through a natural language pipeline that scored each session for topic frequency, speaker sentiment, contentiousness, and policy outcomes. Topics were classified into 12 categories (housing, public safety, budget, environment, infrastructure, etc.). Contentiousness was measured by the ratio of opposing-sentiment speaker turns to total turns. Results were validated against meeting minutes and vote records published on Legistar.
Quality-of-Life Indices: Four composite indices (0–100) built from weighted sub-indicators. Education (36.2): GreatSchools ratings, pupil-teacher ratios, OUSD fiscal status — low confidence due to state-level proxies for some metrics. Safety (31.2): violent and property crime rates from CrimeWatch, officer staffing levels — high confidence with city-level data. Prosperity (47.4): median income, unemployment, business tax revenue, commercial vacancy — moderate confidence. Affordability (54.1): rent-to-income ratio, HUD fair market rents, home values — moderate confidence. Each source is weighted by authority tier: federal data (Tier 1, weight 1.0), state/county data (Tier 2, weight 0.85), local/computed (Tier 3, weight 0.7).
Simulation Engine: A discrete-time affine dynamical system that models Oakland's 113 census tracts over a 10-year horizon. The engine applies budget allocation across four policy dimensions (education, safety, housing, economic development) with hard budget constraints — every dollar allocated to one dimension is subtracted from the others. Effects propagate through 5 feedback loops: crime suppresses commercial investment (crime→economy), prosperity raises land costs (prosperity→affordability), safety improves school outcomes (safety→education), economic growth reduces crime over time (economy→crime), and rising costs displace lower-income residents (displacement). Diminishing returns ensure no dimension can be driven to extreme values. Governance friction delays housing effects by 35% more than other dimensions, reflecting permitting and construction timelines.
Simulation Review: The model was reviewed from four independent perspectives: urban economics (feedback loop structure and magnitude), public finance (budget constraint realism), statistical modeling (parameter sensitivity and edge cases), and civic governance (political feasibility of scenarios). The simulation is an educational illustration of policy trade-offs, not a quantitative forecast. It cannot predict council decisions, state intervention, private-sector relocation, or macroeconomic shifts.
Tract-Level Mapping: The 3D cityscape map assigns each of Oakland's 113 census tracts a composite quality-of-life score based on weighted baseline indicators (income, crime density, housing cost burden, school proximity). Block height represents the composite score; color runs from red (struggling) to green (thriving). Tract geometries from Census TIGER/Line, neighborhood labels placed at approximate centroids of 8 key areas.
Limitations: No migration dynamics, housing construction pipeline models, or state/federal policy changes. Effect weights are calibrated from urban economics literature, not estimated from Oakland-specific panel data. Unemployment uses county-level data (Alameda County via BLS), not city-level. GreatSchools ratings cover 87 of OUSD's schools but not charter or private institutions. Crime data completeness varies by year and precinct. Simulation magnitudes are directional illustrations — the model shows which trade-offs bind hardest, not precise outcome values.
Date Ranges: Crime: 2005–2024 (1.26M records). Census: ACS 5-year ending 2022. Budget: FY2024-25 adopted. BLS unemployment: 2023. HUD rents: FY2024. GreatSchools: 2024 ratings. Council meetings: 2019–2024 (134 sessions). EPA EJScreen: 2023. FEMA NRI: 2023.