DATA PROVENANCE & TECHNICAL ARCHITECTURE
Complete Directory of Live Data Feeds, Geographies & Models
Every layer in this application is powered by verified, programmatic datasets downloaded and cross-indexed by
data_pipeline.py into local GeoJSON/TopoJSON/JSON caches. Below is the complete inventory of where each
dataset originates, how it is joined, and how the economic and electoral metrics are calculated.
Live race calls, candidate vote totals, party vote shares, Estimated Elected Vote Percentage (EEVP%), precinct reporting counts, and flipped-seat flags for yesterday's November 3, 2026 midterm elections across all 435 US House districts, 35 US Senate races, and 36 Gubernatorial races.
Household-level distributional impacts (in 2025 dollars per household per year over the 2025β2034 budget window) across 3,143 US counties and 435 Congressional Districts, synthesized from Congressional Budget Office (CBO), Tax Foundation, Urban Institute, and CBPP microsimulations.
Why does the unadjusted average look positive (green) almost everywhere?
Under the unadjusted baseline, OBBBA borrows
$3.4 Trillion in national debt to finance
$4.5T in tax cuts (heavily concentrated in the top 20% of earners) while cutting
$1.1T from SNAP & Medicaid transfers. Because large tax cuts for top earners pull up the arithmetic mean of each county, raw "All Households" averages are positive (+$500 to +$4,000) in nearly every county.
How to see the true distributional contrast on the map:
- Budget-Neutral Mode ($3.4T Debt Adjusted): Following VisualizePolicy's CBO formula (
fed_tax × 0.25 + fed_transfer + state_tax + other_spending), scaling tax cuts to what the bill actually pays for turns most working-class and rural counties Red (Net Loss) while wealthy enclaves remain Green (Net Gain).
- Low-Income (Bottom 20%): Directly reveals the -$300 to -$1,400/yr losses suffered by low-income families due to Medicaid and SNAP reductions.
- Relative Quantile Scaling: Colors counties into 7 percentile tiers relative to the national median (+$1,450/yr) so you can immediately distinguish counties gaining only +$400 from wealthy counties gaining +$4,500.
Between the 2024 and 2026 elections, an unprecedented wave of mid-decade redistricting altered congressional boundaries in 9 states: Texas (TX), California (CA), North Carolina (NC), Florida (FL), Ohio (OH), Tennessee (TN), Louisiana (LA), Utah (UT), and Alabama (AL). We load both boundary sets simultaneously so you can overlay the old 2024 lines (dashed amber) directly over the new 2026 midterm results.
Combines three distinct enforcement datasets: (1) 455 geocoded community ICE raid reports with timestamps and incident descriptions; (2) 25 ICE Area of Responsibility (AOR) Field Office hubs tracking 782,559 arrests and 279,318 removals disaggregated by Criminal Conviction, Immigration Violation, and Pending Charges; and (3) 1,000 official ICE Detention Facility records.
Comprehensive spatial inventory of 1,757 US data center campuses (~159+ GW of combined tracked power capacity), including facility operator (AWS, Microsoft, Meta, Google, Oracle, CoreWeave, Stak, etc.), build status (Operating, Under Construction, Proposed, Approved, Suspended), Megawatt (MW) power demand, power/cooling sources, and local community resistance/pushback status.
Live daily retail pump averages across all 50 US states + DC (Regular Unleaded, Mid-Grade, Premium, and Diesel) combined with county-level retail gas prices across 3,143+ US counties from AAA's state county-map feeds. Includes price spreads relative to the US national average and estimated annual household fuel expenditures (~1,150 gal/yr).
Provides county-level 2024 Presidential general election returns across 3,160 US counties (Trump vs. Harris vote totals, percentages, and margins) so you can compare county-level OBBBA economic impacts and ICE/data center locations against local partisan lean. Rendered over keyless OpenStreetMap / CARTO Dark Matter WebGL vector tiles.