Every score on Politicount is computed from cited public sources, and the evidence rows sit one tap away under "Why this score." No evidence → no score: the meter reads "Not yet weighed." Scores carry the period they cover and a "Data through {date}" stamp. There is no single composite grade — one number would hide exactly the trade-offs this site exists to show.
The gap between how an official presents themselves and what the record shows. The Face = documented public statements only — quote, date, source. The Reality = official records only — votes, filings, actions. The page shows paired evidence rows (statement beside record) plus an Alignment % of checkable claims whose record matches. Statements too vague to check are logged as uncheckable and excluded — and we say so.
Where they sit left–right inside their own chamber, from roll calls only. Federal: DW-NOMINATE (Voteview, UCLA) — the academic standard. State: party-unity and agreement-vs-chamber-median scores, labeled plainly as agreement scores, never dressed up as ideology scores.
A 0–100 weight from coded votes and authored bills across a published codebook: labor & wages, healthcare costs, housing, consumer protection, corporate taxation, small business, deregulation, trade, antitrust. Every coded vote shows its rationale. Donor money never feeds this meter — that would make it circular.
Every specific, checkable campaign promise is stored verbatim with its source before scoring begins, then tracked: Kept / In progress / Stalled / Broken / Not yet started. The delivery ratio counts resolved promises only; the unresolved ones are shown as counts, never hidden.
Are they using the office? Banded activity (Quiet / Working / Driving) per category: bills authored and advanced, amendments, oversight actions, executive actions for executives. Press releases claiming action with no record behind them are Face items, not power use.
Named by Raphael's explicit ruling (Nov 7, 2026) for the February 2021 episode defined below. It is applied to every official by the same formula — including Senator Cruz's own page, which is computed from the same record as everyone else's.
The scale: ✈️ low end (left when it hit the fan — the label links out to Wikipedia's account of the February 2021 episode) · green stop-sign high end. Between the two: the dated event rows, sourced.
Trigger family 1 — crises: a declared emergency or disaster in the official's own jurisdiction (freeze, hurricane, flood, blackout, mass-casualty event). Trigger family 2 — tension: moments of unresolved political weight — shutdowns, emergency funding fights, major pending votes, impeachment- or war-grade moments.
For each event, one dated, sourced row records the official as: present & acting / present / absent-explained / left the jurisdiction / unreachable — built only from emergency declarations, official schedules and statements, credible reporting with links, and votes missed inside the window. It also records return timing against resolution, and it records the positive rows: officials who stayed get "present through it, working in-district" where sourced.
The honesty fence: a scheduled recess alone is not abandonment — a row must show the unresolved crisis or tension and the absence together; the context is the evidence. No event on record → "No crisis on record yet." Why the name: in February 2021, as Texas froze during Winter Storm Uri, Senator Ted Cruz was photographed boarding a flight to Cancún — documented at the time by multiple national outlets and addressed in the senator's own subsequent statements. The meter asks, for every official and every crisis since: were you here, or were you gone?
Attendance — share of roll calls voted, shown as a plain stat. Accessibility & responsiveness — town halls, public meetings, office presence; self-reported items labeled as such. Transparency habits — disclosures filed on time, published schedules and meeting logs where the official publishes them. Independence — votes against their own party's majority, shown as a sourced stat, never labeled "courage."
Money trail: who funds them (small donors, large donors, PACs, self-funding; top industries) — labeled "Context — not a score." Money raises questions; it is never asserted to have caused a vote. District outcomes: Census indicators for their district, shown as context. Ethics flags: official findings only — ethics committees, courts, inspectors general. Never allegations, never rumors.
Officials' pages carry a link out to the official's own campaign page (we never touch donation money), and citizens can gift citizens for their own work. None of that can move a score — by structure, not by promise: scores are formula-only, computed from the public record. Support flows never feed any meter, ranking, or display order. No official gains or loses a point because money moved.
Congress.gov API and Voteview (federal votes and ideology) · OpenStates (state legislators, bills, votes) · OpenFEC (federal money) · Texas Ethics Commission and capitol.texas.gov (Texas) · GDELT (news headlines, linked out — never republished) · officials' own channels (the Face) · Census (district context and ZIP-to-district mapping). Every ingested record keeps its source, fetch date, and source URL — that stamp is what makes "Why this score" possible.
Methodology version 1 · Nov 7, 2026 · Changes are versioned and announced; when a formula changes, affected scores are recomputed or the change is disclosed on the page.