InsideGov

insidegov.fyi · the U.S. government, by the numbers: public records from all three branches, turned into charts you can explore. The presidency, Congress (1789 to today) and the Supreme Court, without reading the raw files.

The site is organised by branch. Each branch has several pages, each built around one question.

The presidency (/presidency/…)

• Executive orders — how many each president has signed since 1994, and on what topics.
• Economy — jobs, inflation, mortgage rates and more, laid against administrations.
• Energy — oil, gas, electricity and the Strategic Petroleum Reserve, with the policy actions marked.
• Trade — U.S. trade with each country, plus tariffs and duties collected.
• Immigration — green cards and naturalizations (FY1977 on), refugee admissions against each year's ceiling (FY1980 on) and ICE removals (FY2003 on), by fiscal year and country.
• Troops abroad — where U.S. troops are stationed abroad, and the bases, over time.
• Foreign aid — where U.S. foreign assistance goes, on a world map.

Congress (/congress/…)

• Ideology explorer — every member's voting record reduced to a point in a two-dimensional space (DW-NOMINATE), from the 1st Congress (1789) to the 119th (2025–27). Scrub through 236 years, compare a state's delegation, and read any current member's trajectory against their party's mean.
• Laws (/congress/laws) — every public law since the 93rd Congress (1973), by policy area and by how close the final-passage vote was, with David Mayhew's major laws marked and a page for each law showing who voted for it.
• Who serves — age, gender and length of service, for every Congress since 1933.
• Net worth (/congress/wealth) — first vs. latest net worth on a shared signed-log scale, highest and lowest lists, and the actual assets and liabilities on every current member's profile, estimated from their annual financial disclosures.
• Committees and subcommittees — every House, Senate and joint committee of the 119th Congress blended to a point, with rosters and a page for each.
• Member profiles — a page for every current representative and senator.

The Supreme Court (/supreme-court/…)

• Ideology explorer — where the Court's middle sits, term by term, from Martin–Quinn scores for every justice since 1937.
• Decisions — how many argued cases the Court decides each term and how often it splits, by issue area, since 1946, with a case list and landmark cases.
• Justice profiles — a page for every justice, with a bio, a portrait and their voting record.

Data

Branch	Source	Used for
Congress	Voteview (Lewis, Poole, Rosenthal, Boche, Rudkin & Sonnet)	DW-NOMINATE ideal points — the static career score and the per-Congress (Nokken–Poole) score
Congress	@unitedstates/congress-legislators	Names, states, parties, terms, leadership, the icpsr ↔ bioguide_id crosswalk, and current committees and rosters
Congress	@unitedstates/images	Official member portraits (current members only), committed under public/images/members/
Congress	House Clerk (disclosures-clerk.house.gov) and Senate eFD (efdsearch.senate.gov)	Annual financial disclosures behind /congress/wealth and each profile's net worth card
Congress	GovInfo Bill Status (108th on), the Congress.gov API (93rd–107th) and Voteview roll calls	Public laws, sponsors and cosponsors, policy areas and final-passage votes behind /congress/laws
Court	Martin–Quinn scores	Per-term ideal points for every justice
Court	Supreme Court Database (Washington University)

The full list, with notes on what each source covers, is on /methodology.

Lewis, Jeffrey B., Keith Poole, Howard Rosenthal, Adam Boche, Aaron Rudkin, and Luke Sonnet (2026). Voteview: Congressional Roll-Call Votes Database. voteview.com

Raw snapshots are committed under pipeline/raw/, so builds are reproducible and don't touch the network. Scheduled GitHub Actions (.github/workflows/) re-fetch each source on its own cadence and open a PR only when something material changed; the PR merges itself only after the same checks CI runs have passed. Hand-curated data (tariff and energy actions, ICE figures) has review reminders instead. Data conventions (why each track has its own join key, why the two DW-NOMINATE scores must not be conflated) are written down in docs/DATA_CONVENTIONS.md, with a methodology doc per topic alongside it.

How it's built

• Next.js 16 (App Router) + React 19 + TypeScript, Tailwind v4 themed from a small set of CSS custom properties (light/dark).
• A build-time data pipeline, not a database. pipeline/ fetches the raw files, validates every row against a Zod schema, and transforms them into normalized JSON, one set of files per track (Congress, the Court, executive orders, indicators, energy, trade, immigration, foreign aid, troops, court decisions). The app reads those at build time and statically prerenders every page — there is no runtime data fetching and nothing to operate. Financial disclosures are a separate Python sidecar (pipeline/financial_disclosures/, PDF/HTML parsing is easier there) producing financial_disclosures.json (one row per member per reporting year, band-count grain) and line-items/<year>.json (the verbatim per-asset/per-liability grain, sharded by year) — both validated the same way (Zod on the TS side, a dependency-free schema check on the Python side) before the app reads them.
• D3 for the maths only (d3-scale, d3-shape, d3-force) — scales, path strings, and the beeswarm collision layout. Every <circle>, <path> and <line> is plain JSX, so React owns the DOM. Low-level primitives (ChartFrame, Axis, Tooltip) plus the ScatterPlot and SwarmRows chart bodies back every view — the member and committee compasses are the same ScatterPlot with different accessors, not separate charts.
• CI (.github/workflows/ci.yml) type-checks, lints, runs the unit tests, re-runs the pipeline and fails if the committed output drifts, then builds.

app/          routes — the hub (/), the explorer (/congress/ideology), the Court (/supreme-court/ideology), executive orders (/presidency/executive-orders), foreign aid (/presidency/foreign-aid), member profiles, committee pages, /congress/wealth
components/    charts/ (primitives), senate/ (the ideology views), profile/, committee/, wealth/
lib/           the build-time data layer + shared helpers
pipeline/      fetch → validate → transform → pipeline/output/*.json
                financial_disclosures/ — Python sidecar (PDF/HTML parsing) for
                financial_disclosures.json + line-items/<year>.json
public/        static assets, incl. committed member photos (images/members/)
docs/          DATA_CONVENTIONS.md, CREDITS.md, and a *_METHODOLOGY.md per topic

See ARCHITECTURE_MAP.md for the data-layer, route, and shared-component maps.

Running locally

pnpm install
pnpm dev          # http://localhost:3000

Node ≥ 20.9. Other commands:

pnpm build          # production build
pnpm typecheck      # next typegen && tsc --noEmit
pnpm test           # vitest
pnpm pipeline       # fetch + validate + transform (re-fetches from source)
pnpm pipeline:check # validate + transform + assert pipeline/output is unchanged

Scope

• Congress: the House and the Senate, every Congress from the 1st (1789) to the 119th, with a profile for every current member and a page for every committee and subcommittee of the 119th. Committee membership is only tracked for the current Congress (there is no historical roster file upstream), so those views are pinned to the 119th.
• The Court: every justice with a Martin–Quinn score (1937 on), and decision counts from the Supreme Court Database from 1946.
• The presidency: executive orders from 1994; economic and energy series with a display window starting in the modern era; trade and duties from 1991 and 1993; green cards and naturalizations FY1977 on, refugee admissions FY1980 on, ICE removals FY2003 on; foreign aid, troops and bases over their source's span. Each page's "Data notes" give the exact window.

Still future work: a per-member record of every roll-call vote (profiles list the laws a member sponsored or cosponsored, not their votes).

Licence

Personal project — the code isn't licensed for reuse yet. The underlying data belongs to its sources: Voteview data is free for scholarly and public use with attribution; congress-legislators is public-domain (CC0).