Voters cannot hold lawmakers accountable for behavior they cannot see. Disclosure laws produce mountains of data that almost nobody reads. The Swamp Score reads it for them.
The Swamp Score is a composite measure of self-dealing, donor capture, and ethical misconduct by U.S. federal lawmakers. It is built from publicly available behavioral and outcome data, scored on a transparent rubric, sourced down to the underlying document, and applied identically to lawmakers of both parties.
A measurement, not an accusation.
This distinction is doctrinal, not cosmetic. Every public-facing number reinforces it.
- A measurement of behavior on the public record — filings, disclosures, votes, court records, ethics actions.
- A composite of eight component scores, each independently published and defensible.
- Applied to a defined population (535 federal lawmakers) using the same rubric.
- Updated quarterly.
- An accusation of criminal conduct. The score spans legal-but-unseemly to felonious; it does not collapse that spectrum into a single legal claim.
- A judgment of a lawmaker's character. It scores what is on the record, not who they are.
- A predictor of future behavior. It is a backward-looking accounting of disclosed and discovered facts.
Eight scores, one Swamp Score.
Each lawmaker receives a Swamp Score from 0 to 100, where higher means more concerning conduct on the record. It is built from eight components, each also scaled 0–100, in two equal halves:
- Half is the weighted average of all eight components, using the weights below. This half rewards breadth: problems across many parts of a record add up.
- Half is the single highest component. This half keeps one serious matter from being diluted by clean marks elsewhere. A lawmaker with a formal ethics finding and nothing else on the record should not land near the bottom of the table.
Swamp Score = ½ × weighted average + ½ × highest component, rounded half-up. Two lawmakers with the same worst matter are separated by the rest of their records; two with the same average are separated by how serious their worst matter is.
| Component | What it measures | Weight |
|---|---|---|
| Trading | STOCK Act trades by member and household; private-equity and other non-public vehicles tied to donors or to sectors under the member's jurisdiction; timing against hearings, votes, and non-public information windows; performance vs. benchmarks; disclosure compliance. | 15% |
| Donor Capture | Concentration of donor base; sector–jurisdiction overlap; vote-contribution timing patterns. | 15% |
| Family Enrichment | Spouse/dependent employment by regulated entities; household lobbying registrations; net-worth growth attributable to family business; documented extended-family flows. | 12% |
| Earmarks & Member-Directed Spending | Funding directed to donors, donor-affiliated entities, or family-connected entities; ratio of directed spending to constituent need. Measured on member-directed requests from FY2022 onward, when earmarks returned; pre-moratorium earmarks are context unless tied to a donor or family interest, and a member with no requests on record scores at the floor. | 12% |
| Ethics Actions | OCE referrals, House/Senate Ethics matters, reprimands, expulsions — separated by stage (opened / referred / sanctioned). | 13% |
| Legal Actions | DOJ, SEC, IRS, and state-AG investigations, charges, convictions, settlements — separated by stage. | 13% |
| Revolving Door | Pre-office employment by regulated entities; post-office lobbying or board placements; staff revolving-door density. | 10% |
| Travel, Gifts & Disclosure | Privately funded travel volume and sponsors; gift receipts; late, amended, or missing filings. A periodic transaction report is late if filed more than 30 days after the member was notified of the transaction or more than 45 days after it, whichever comes first (the STOCK Act standard). Senate reports record no notification date, so senators’ reports are tested against the 45-day limit only. | 10% |
Behavior vs. outcomes
Each component blends two signals. Behavior — discrete acts on the record (a trade made, an earmark sponsored, a gift accepted, a filing missed) — carries the higher weight (~65% within each component). Outcomes — patterns and results such as net-worth growth or donor-concentration ratios — carry the rest (~35%) as supporting signal.
Lifetime and annualized.
Every score is published two ways. The Lifetime score is total points accumulated across a member's full federal tenure. The Annualized score divides that by years served. The lifetime view favors deep surface area; the annualized view favors concentrated misconduct per year served.
Three counters, not one.
Prosecutorial and ethics activity is decomposed into three separate signals, weighted by stage. This separates the cover-up problem — smoke without fire still counts, just less — from the presumption-of-innocence problem, where full weight attaches only after due process.
| Stage | Legal Actions | Ethics Actions |
|---|---|---|
| Outside complaint filed, no proceeding opened | floor | 6–10 pts |
| Investigation opened / referral filed | 0.2× | 0.3× |
| Charges filed / sanction recommended | 0.5× | 0.6× |
| Conviction / settlement / sanction imposed | 1.0× | 1.0× |
Edge cases
- Complaints that go nowhere. Advocacy groups on every side file ethics complaints cheaply. A complaint that no official body has acted on scores at the complaint level — 6 points for one, up to 10 for several — below any opened proceeding. A proceeding closed by formal dismissal steps down one stage; one dropped without explanation stays where it is.
- Scope. Conduct unrelated to self-dealing, donor capture, disclosure, or misuse of office — personal-conduct offenses, political speech and votes, résumé claims, private family history, private business debt — is surfaced as context and not scored. A matter that mixes in-scope and out-of-scope allegations is scored on the in-scope part alone.
- One episode, one component. Each episode is scored in the component it fits best; a second component may score it only for genuinely distinct facts. Ethics and Legal may each score the same episode because they track different bodies, and undisclosed bribery may also surface under Disclosure (§08). A proceeding against a household member is scored once, under Family Enrichment.
- Matters where the lawmaker was a witness or victim are excluded.
- Investigations dropped without explanation are kept at the lowest weight — the drop is itself the cover-up signal.
- Pardon before adjudication. Charges ended by executive pardon before trial are weighted at the charges-filed tier (0.5×). A pardon is an executive act, not a judicial finding; it forecloses the verdict rather than resolving the facts, so it is neither a conviction nor an acquittal. The pardon is surfaced as the reason no adjudication occurred. The rule applies identically regardless of the party of the pardoner or the pardoned.
- Pre-federal findings. A formal, adjudicated finding against the lawmaker from a prior public office — a state ethics commission ruling, a campaign-finance board order, a repaid tax irregularity — is scored at reduced weight within the relevant component. Allegations, lawsuits, campaign attacks, and lawful but unflattering conduct from that period that produced no finding are surfaced as context and not scored, whether the underlying facts are disputed or not. Private-sector litigation before any public office is scored at reduced weight only where a court entered a finding against the lawmaker personally.
- Politically-flavored investigations from aligned prosecutors are kept; a methodology fight is preferable to selective omission.
- Sealed matters are excluded until unsealed, flagged as "known sealed matters" with no scoring impact.
- A lawmaker acting as accuser or complainant against others is not scored — the score measures conduct against the member, not advocacy regarding others.
A position, not a second score.
The absolute Swamp Score is the primary measurement — it answers, "how bad is this lawmaker against a fixed rubric?" and is stable over time. Alongside it, every member shows a percentile answering, "where does this lawmaker fall among everyone scored so far?"
The percentile is the share of other scored lawmakers whose Swamp Score is lower, with ties counted as half, rounded and shown from 1 to 99. A percentile of 80 means the lawmaker scores higher than about 80% of the members scored so far.
Substance over form. Wealth is not misconduct.
Trading scores the substance of an investment — who controls it, what sector it touches, and what relationship the member has to it — not its form. That closes the structural loophole by which investments routed through LLCs, private funds, or other non-public vehicles would otherwise evade scoring. A private-equity or hedge-fund position counts when the underlying investment is in an entity tied to the member or a donor, or in a sector under the member's committee jurisdiction.
- Individual-stock and household trades, with timing against hearings, votes, and information windows.
- Non-public vehicles when the underlying holding is donor-affiliated or jurisdictionally relevant.
- Holdings in sectors under the member's committee jurisdiction; disclosure timeliness and accuracy.
- Diversified mutual funds, ETFs, and index funds — no targeted exposure.
- Standard 401(k)/IRA retirement vehicles.
- Direct real estate unrelated to the member's jurisdiction.
- Inheritance-derived holdings clearly predating tenure. Broad market exposure and personal wealth, by themselves, are not a conflict.
Separating normal fundraising from capture.
The line between ordinary political fundraising and capture requires explicit, defensible criteria. Three measures do the work:
Sector–jurisdiction overlap
The share of campaign and leadership-PAC money coming from industries directly regulated by the member's committees. Below 15% is a low signal; 15–35% medium; 35%+ high.
Donor concentration
A concentration index of the top donors and donor-affiliated PACs as a share of total fundraising. A broad small-donor base scores low; a concentrated industry-PAC base scores high.
Vote-contribution timing
Contribution spikes from affected interests in the 90 days before and after relevant votes, and the share of post-sponsorship money from benefited sectors. None of these prove a quid pro quo; they flag patterns consistent with one. One coincidence is noise. A twenty-year career of coincidences is a signal.
The asymmetric household rule.
Family conduct is scored asymmetrically, in a deliberate and defensible direction.
- The lawmaker, spouse or domestic partner, and dependent children.
- Trusts, LLCs, and partnerships where any of them is a beneficial owner or has signing authority.
- This is the standard household definition under the Ethics in Government Act, already reflected in Personal Financial Disclosures.
- Adult independent children, parents, siblings, in-laws, and other persons or entities.
- Scored only when documented by at least two independent reputable outlets with overlapping primary-document sourcing.
- Higher confidence applies when reporting has survived a defamation challenge, appeared in a court filing, or been confirmed by official action.
Applied identically, and audited for it.
No party variable. Party affiliation is metadata; it appears on the page but never enters a scoring formula.
Matched-pairs audit. After scoring, any meaningful gap between similarly situated lawmakers of opposite parties — same committee, same tenure bracket — must come with a sourced explanation. If it can't, the methodology is wrong, not the lawmakers.
Editorial decoupling. Commentary can analyze the rankings; it cannot produce them. The rubric does that.
What the pilot found
| Matched pair | Composite gap |
|---|---|
| Fallon (R, 47) vs. Suozzi (D, 48) — ethics referrals over late stock-trade reports, both dismissed | 1 pt |
| Guthrie (R, 46) vs. Pallone (D, 40) — chair and ranking member of Energy and Commerce | 6 pts |
| AOC (D, 31) vs. Luna (R, 25) — recent tenure, low surface area | 6 pts |
| Rick Scott (R, 24) vs. Warner (D, 32) — nine-figure Senate fortunes, no blind trust | 8 pts |
| Menendez (D, 78) vs. Santos (R, 69) — both convicted | 9 pts |
Among lawmakers with similar conduct profiles, partisan gaps run from one to eight points, and each comes with a sourced explanation on the members’ pages. The larger gaps come from each record’s single worst matter, which carries half the score (§03). Warner’s worst component is the revolving door: 22 former staffers registered as lobbyists. Scott’s record is spread across several mid-range components with no single large one. AOC’s is a House Ethics Committee finding over the 2021 Met Gala; Luna’s highest component is lower. Guthrie’s pharmaceutical and health-industry donor concentration is heavier than Pallone’s. Menendez and Santos share the maximum on their worst component; the gap reflects three decades of documented conduct against eleven months, and the fact that defrauding donors is scored as fraud, not as donor capture (§08). The components move in both directions on the facts: AOC scored higher than Luna on Ethics; Tuberville scored higher than Pelosi on Disclosure; Luna scored higher than AOC on Trading; Pelosi scored higher than Tuberville on Trading. Different signs and magnitudes for different lawmakers, based on what they actually did.
Source-linked. Answerable. Versioned.
Source-linking everything. Every numerical input on every lawmaker page links to the underlying public document. Truth is a defense; documented public records are truth.
Right of Reply. Every page carries a permanent response section. Any subject may submit up to 1,000 words, published verbatim. A refusal to respond becomes its own signal.
Methodology transparency. The full methodology lives here and is versioned. When weights shift, prior versions remain published.
Honest about what we cannot see.
Every published profile carries a Coverage Note disclosing that the score reflects publicly documented episodes as of the profile date; that coverage is non-exhaustive, especially for long-tenured members; that readers may submit additional documented inputs; and that submissions are evaluated against the family-scope sourcing standard before being incorporated.
Coverage will never be complete. Transparent uncertainty is the project's competitive advantage — readers trust it more than confident incompleteness.
Published, not hidden.
- Lawmakers who entered office rich show less "growth-in-office" signal than those who get rich in office.
- Sophisticated misconduct is, by definition, less visible than crude misconduct; members skilled at staying off the record score lower than their conduct warrants.
- State-level investigations, foreign-government dealings, and non-public matters are under-represented.
- Family-enrichment inputs depend on journalism, which introduces coverage bias: more-covered lawmakers surface more inputs.
- The weights are a defensible opening position, not a derived truth; weight-sensitivity discussion is explicitly invited.
- The percentile describes position among the lawmakers scored so far, not all of Congress, and will shift as coverage grows. The Swamp Score is the historically comparable number.
- Counting the highest component at half weight means a single severe matter dominates a score by design. A lawmaker with one finding and an otherwise clean record can outscore one with several smaller problems.
Where the inputs come from.
| Source | Feeds |
|---|---|
| House & Senate STOCK Act filings; annual Personal Financial Disclosures | Trading, Family Enrichment |
| FEC campaign-finance data | Donor Capture, Earmarks |
| House & Senate LDA lobbying filings | Revolving Door, Family Enrichment |
| OCE and House/Senate Ethics Committee records | Ethics Actions |
| DOJ, PACER, SEC enforcement, state-AG actions | Legal Actions |
| Congressional earmark / member-directed spending records | Earmarks |
| Foreign and privately funded travel disclosures | Travel, Gifts & Disclosure |
| Congressional-trading aggregators; net-worth estimates; investigative journalism (per family-scope standard) | Trading, Family Enrichment |
