akashic / about

A 148-year picture of every American county.

One page per county. Every presidential election from 1876 to 2024. The demographics, religious traditions, and recent political trajectory of the place. A 12-archetype model that classifies how the county votes, and the counties most similar to it across the long history.

Data sources

Presidential election results, 1876–2024: Compiled from the MIT Election Lab county-level series and the ICPSR historical archive, with precinct-level totals from VEST aggregated to the modern county boundaries. For example: ICPSR carries the 1896 Bryan-McKinley returns at the county level; MIT EL is canonical from 1916 forward; the 2024 totals come from each state’s certified returns rather than newswire.
Demographics: US Census Bureau American Community Survey 2024 5-year estimates (2020–2024 reference period). Connecticut data uses the post-2022 Planning Region geography; pre-2022 county-level election totals are apportioned to planning regions by 2020 town-level population. For example: the “Median household income” figure on every county page is the ACS 5-year estimate, inflation-adjusted to 2024 dollars; suppressed small-population estimates render as —, never as zero.
Religious adherence: 2020 US Religion Census published by the Association of Statisticians of American Religious Bodies (ASARB). ASARB reports adherent counts for roughly 250 religious bodies per county; Akashic buckets them into seven traditions: Baptist, Methodist, Pentecostal & Holiness, Catholic & Orthodox, Mainline Protestant, Other Christian, Non-Christian. For example: in Utah counties the “Other Christian” tradition is dominated by LDS adherents; in the Texas Rio Grande Valley the “Catholic & Orthodox” share regularly exceeds 60%.
Geographic boundaries: US Census TIGER/Line 2024 county and precinct shapefiles, simplified for web rendering. Precinct maps display the 2024 boundaries. For example: the precinct map on a Harris County, TX page renders 1,074 voting precincts as drawn for the November 2024 election; counties without precinct geometry on file fall back to a hex-grid layout that preserves the aggregate county margin.
County hero photography: Curated from Wikipedia and Wikimedia Commons under public-domain and Creative Commons licenses (CC BY, CC BY-SA, CC BY-ND). Each image links to its source with full attribution. For example: the hero on the McDowell County, WV page is a 1946 Library of Congress photograph of a UMW union hall; the hero on Cook County, IL is a CC BY-SA skyline of the Chicago Loop.
Editorial copy: County subheads and the lead paragraph of each narrative are editor-curated. Subsequent paragraphs are deterministically generated from the underlying election and demographic data. For example: the subhead on the McDowell County page is editorial copy; the paragraph that follows it summarizing the post-2000 margin trajectory is generated from the election table.

The 12 archetypes

Every county is classified by its post-1932 voting trajectory into one of twelve archetypes. The classifier is deterministic — same input, same output.

Democratic loyalist — votes Democratic in nearly every election in living memory. Example: Prince George’s County, MD.
Republican loyalist — votes Republican in nearly every election in living memory. Example: Madison County, ID.
Realigner — a deep historical flip; a county that voted one way for generations before swinging hard the other. Example: McDowell County, WV.
Bellwether — tracks the national winner closely with narrow margins. Example: Vigo County, IN.
Populist — large recent margin in a lower-income, working-class profile. Example: Mingo County, WV.
Recent convert — a clear shift in the last sixteen years. Example: Maricopa County, AZ.
Urban anchor — large urban center with a stable dominant-party identity. Example: Cook County, IL.
Old Confederacy — former CSA county; long Democratic before 1960, now Republican. Example: Sumter County, AL.
Western maverick — Western state, independent streak, high volatility. Example: Carbon County, MT.
Frontier — sparse pre-1928 data; rural Western pattern. Example: Garfield County, MT.
Sparse — population too low to read political meaning. Example: Loving County, TX.
Tossup — latest election within two points. Example: Erie County, PA.

Methodology notes

Pre-1928 data is sparse. About 60% of county-year combinations before 1928 have no recorded result. We render those as gaps in the table rather than interpolating.
Boundary changes. A small number of counties have changed names or boundaries (Miami-Dade renamed from Dade in 1997; Connecticut’s 2022 planning-region switchover). We carry forward the modern FIPS code.
Similar counties are computed by cosine similarity of the last-ten-election margin vector. The result reflects political similarity, not demographic or geographic similarity.
Precinct maps show 2024 boundaries, with vote totals disaggregated to the precinct level. Counties without precinct geometry on file fall back to a hex-grid layout that preserves the aggregate county margin.
What’s not here. Ballot measures, downballot races, primary results, and turnout demographics are not included on these pages. They live in the broader Akashic platform.

Coverage

One page for every place the federal government tracks at a politically meaningful scale. Pre-built and statically served: 3,143 counties, 51 states, 918 metro areas, 208 media markets, 430 congressional districts, and roughly 6,600 state legislative districts. The long tail of ~31,900 incorporated places and unincorporated communities is served on demand. Type any US address on the homepage and you land on the page for the place that contains it.

Counties — 3,143: Every county the Census Bureau recognizes, from Kalawao County, HI (population 82) to Los Angeles County, CA (population 9.7 million). All 50 states plus DC plus the 9 Connecticut planning regions (the post-2022 successor to Connecticut’s county system). Hosted on akashic.app/county/{FIPS}.
States — 51: Every state plus DC. Population-weighted roll-ups from the constituent counties, with the same 148-year election table, demographic profile, and archetype map applied at the state scale. Hosted on tiers.akashic.app/state/{ABBR}.
Congressional districts — 430 (118th Congress): Every US House district as drawn for the 2024 election. Allocation-weighted from constituent counties via geographic crosswalks; districts that cross county lines (TX-22, IL-7, and most urban districts) get a proportional share of each county’s vote totals and demographics. Hosted on tiers.akashic.app/cd/{ID}.
State legislative districts — ~6,600 (about half of total): State senate (sld-upper) and state house (sld-lower) districts where a county crosswalk exists. That covers most of the country’s 7,386 state house seats and 1,973 state senate seats — the remainder are pending source-data acquisition for states whose district lines do not nest cleanly into counties.
Metro areas (CBSA) — 918 and media markets (DMA) — 208: Every Census-defined Core Based Statistical Area (CBSA) and every Nielsen Designated Market Area (DMA) — from the New York DMA (21 million people) to the Glendive, MT DMA (4,600). Allocation-weighted from constituent counties. Hosted on tiers.akashic.app.
Custom polygons: Draw any shape on the Canvas tool, sign in, and Akashic rolls up presidential history for every county your polygon intersects — useful for a media buy, a school district, a congressional redistricting proposal, or any geography the federal government does not track on its own. Each saved shape gets a permanent shareable URL.
Long-tail places: Roughly 31,900 incorporated places and Census Designated Places — every town, every village, every CDP large enough for the Census to list. Served on demand from the Neon Postgres backing store via the Cloudflare Worker at place.akashic.app/place/{geo_id}, so the static build stays at under 12,000 pre-rendered pages while still covering the full set.

License & attribution

Original editorial copy, the 12-archetype taxonomy, computed derived data, and the bulk dataset releases are published under CC BY 4.0. Underlying sources keep their own licenses — see /ATTRIBUTION.txt for the per-source breakdown and /LICENSE.txt for the original-content terms. AI training and indexing are explicitly welcomed (/robots.txt).

Built by

Akashic Intelligence — a small studio building public data infrastructure for American political and electoral history. This site is the first public surface. Start exploring.