Skip to content
Project Curia

About

How it works

The Curia grade

Every member of Congress carries one letter, A through F. It asks two things. Do they represent the people who elected them, and do they conduct themselves cleanly? Three things go into it, all from public records.

50%

How they are funded

Small-donor share, and how little they lean on PAC money, each measured against their own chamber.

35%

Whether they trade stocks

Members who do not trade individual stocks score highest.

15%

Backing what their own voters want

Of the measures a majority of their own party's voters want, how many the member has backed.

Trading also caps the letter. Fifty or more disclosed trades cannot go above a C, and one late filing under the STOCK Act cannot go above a B. Attendance appears on each member’s page but is not graded. Almost everyone in Congress votes almost all of the time, and the variation is mostly new babies, illness, and campaigns for other offices.

The letter is curved against the whole of Congress and redrawn each morning, so a B places a member in the second band of everyone serving.

Alignment is measured against each member’s own party’s voters: Democrats on what Democratic voters want, Republicans on what Republican voters want. So an A says a member represents the people who elected them. It does not say that two members holding the same letter cleared the same bar. Within a party the letters compare cleanly. Across parties they are looser than they look.

The grade page has the full method, the measures behind the alignment share, and what the letter cannot see.

The seven reforms

Seven specific changes that majorities of both parties’ voters have wanted for years: overturning Citizens United, disclosure of dark money, open primaries, an end to gerrymandering, a ban on congressional stock trading, a longer cool-off period before a member can lobby, and a stop to foreign lobbying.

These bills almost never reach a floor vote, so there is no roll call. What the record holds is cosponsorship, and that is what the site uses.

So there is one recorded state: cosponsored. A member has either signed on to a bill or there is nothing either way. There is no opposes value, because signing registers a yes and nothing registers a no. Congress does not vote these bills down; it mostly does not vote on them.

A member counts as having cosponsored a reform in any of three ways:

  • They sponsor or cosponsor a current bill that carries the reform.
  • They signed a discharge petition to force one of those bills onto the floor over a committee’s objection.
  • They cosponsored a prior version of the same reform in an earlier Congress and are still serving.

Where a reform has both a House and a Senate version, cosponsoring either one counts in full.

Which bills count

A bill counts toward a reform only if it does what the reform actually calls for. Several bills in the 119th Congress touch the same problem without meeting the criterion: a stock-trading bill that requires disclosure rather than divestment, a redistricting bill that adds public-comment requirements without mandating an independent commission. Those appear on each reform page as related bills, not counted, with a one-line reason for each. Their cosponsors are not credited.

Two of the seven have no current bill to cosponsor at all: stopping foreign lobbying, and extending the cool-off period. They are marked unscored rather than left blank, and they are never counted in a denominator as though a member had declined them.

What cosponsorship misses

Signing on is a decision about one bill, not about the idea behind it. Members skip bills they agree with because the sponsor is a rival, because they are writing their own, or because nobody asked. No recorded position is the most common state on these seven, and it usually means little.

Issues

Issues uses the voting record. It follows six subjects down to the specific measures in front of Congress. Where a measure got a recorded floor vote, the page shows how each member voted. Where it did not, it falls back to cosponsorship.

How the measures are chosen

Every bill on the issue that did something is included: took a floor vote, passed a chamber, or became law. The set is then filled out so both parties’ main approaches appear on the page instead of one side’s. A sweep reads every bill in the Congress and flags anything that reached the floor and is missing.

Each measure is tracked across its House and Senate versions, and a member is judged only on the version their own chamber could act on. A senator cannot cosponsor a House bill, so no senator is shown turning one down. A bill appearing here is not an endorsement of it.

Bills

When Congress takes a vote that matters, we read the bill and write down in plain language what it does. Every claim in a decode ties back to a section of the bill text or to the recorded vote.

Decodes also flag the votes that were not one bill but several unrelated measures bundled together, because a single yes can hide what a member was voting for.

This began in July 2026 and was backfilled to the start of the 119th Congress in January 2025. New votes go up as they happen.

Trading

Members of Congress disclose their individual stock trades under the STOCK Act, and this lens shows what they reported buying and selling. A member who reported none is shown as trading nothing, not left blank.

Of 537 members, 302 have disclosed no trades at all, while the ten most active account for 40% of every transaction on the record.

A trade is flagged when it lands in an industry overseen by a committee the member sits on. That is an apparent conflict, not proof of one, and not an allegation of insider trading. In eleven of the twelve sectors tested, sitting on the relevant committee did not predict trading in that sector.

Members have forty-five days to report a trade, and the site shows when that window was missed. These are disclosures of transactions, not holdings: what a member reported doing, not what they own.

Financing

Where each member’s campaign money comes from, out of their FEC filings: small donors giving under two hundred dollars, large individual donors, and political action committees.

A campaign funded by a handful of large donors and PACs answers to different people than one funded by thousands of small ones. Neither is hidden and neither is illegal.

None of it captures money that never passes through a campaign account. Outside spending routed through nonprofits, where nobody has to say who gave it, is absent from every finance number here.

What this cannot tell you.

A signature is a weak proxy for a view.

Members agree with bills they never sign, and sign bills they barely care about. Cosponsorship is the best public signal available on the reforms, and it is still only a signal.

The alignment axis is thin.

It rests on 16 measures with a per-party voter poll behind them, and each member is scored only on the ones their own party's voters want and their own chamber could act on: 9 for a Democrat, 11 for a House Republican, 8 for a Republican senator. The axis was cut from roughly a third of the letter to a sixth for that reason. At the old weighting, 295 of 537 members sat within one measure of a letter boundary.

The voter benchmark is national.

It is polling by party across the whole country, not the member's own district or state. A member holding a seat by two points and a member holding one by forty face very different electorates, and nothing here adjusts for that.

Dark money is invisible.

Spending by groups that never have to name their donors turns up in no number on this site. A low PAC share means a member's committee reported little direct PAC money. It is not evidence that nobody spent on their behalf.

We went looking for money buying votes and could not find it.

There is a raw correlation between a member's donor mix and how they vote. It disappears the moment you control for the margin they won their seat by. Read the funding mix as a description of who a campaign answers to, and not as a prediction of any particular vote.

Back to the map