Official sources
5
Congress, House, Senate, FEC, Census.
Methodology
Evidence first, interpretation second
The MVP separates official data, reviewed classifications, source-backed summaries, and user-defined alignment math.
Method
Confidence labels
4
High, medium, low, not scored.
Reviewed scoring
Only
Unreviewed votes are excluded.
Corrections
Public
Accepted changes appear in the changelog.
Pipeline
Step 1
Raw source snapshots
Every fetch job stores source URL, timestamp, status, raw hash, parser version, and storage path before parsing.
Step 2
Validation
Vote totals, required fields, representative mappings, source URLs, and campaign finance identifiers must validate before publication.
Step 3
Issue mapping
AI can suggest issue tags, but public alignment uses reviewed classifications or high-confidence official metadata mappings.
Step 4
Scoring
Each reviewed yea or nay vote is compared with the user's selected stance and weighted by importance from 1 to 5.
Known Limitations
This first build is federal-only and uses sample records until API keys, Postgres storage, and ETL jobs are connected. It does not score procedural votes, quorum calls, ambiguous bundled legislation, or classifications still awaiting review.
Sources
Sources5 refs
Congress.gov API referenceCongress.gov · Official · fetched 2026-05-15House Clerk roll call votesU.S. House Clerk · Official · fetched 2026-05-15Senate roll call votesU.S. Senate · Official · fetched 2026-05-15FEC campaign finance dataFederal Election Commission · Official · fetched 2026-05-15Census geocoderU.S. Census Bureau · Official · fetched 2026-05-15