Why do Americans trust Congress less than at any point in fifty years?
Roughly 1 in 10 Americans approve of the job Congress is doing. In the 1980s it was closer to 3 in 10. This site exists to answer that question with evidence rather than opinion — using the same public records Congress keeps on itself.
What this is
Congress produces an enormous amount of public information about itself — every vote, every bill, every committee referral, every signature. Almost none of it is readable by a person who does not already work in Washington. It arrives as bulk XML, fixed-width vote matrices, and codes like [R-TX-11].
Congress Atlas takes those records and turns them into plain answers: how divided the two chambers are, whether they are passing anything, who is actually getting things done, and how the public feels about all of it. Every number on this site is traceable back to a government record or a published poll, and every calculation we invented is shown as arithmetic you can check line by line.
What you can do here
Look up any of the 12,249 people who have ever served and see how they voted, what they wrote, and which of their bills became law. Compare the House and the Senate as separate institutions. Watch two centuries of political division play out on a single scale. Read what the public has thought about all of it since 1974.
What this is not
It is not a scorecard for one side. The measures here are symmetric: they treat both parties identically, and where a number reflects a judgement rather than a fact in the record, we say so and show the working. It is not a prediction engine, and it does not tell you how to vote.
Where the data comes from
Four public sources, all free, none of them ours. Nothing is scraped from a paywall, and nothing is estimated where a record exists.
| Source | What we take from it | Covers | Access |
|---|---|---|---|
| Voteview — UCLA political science project | Every recorded vote by every member, and the ideology scores derived from them. | 1789 – present | Free bulk CSV, no key |
| GovInfo BILLSTATUS — U.S. Government Publishing Office | Bills, who wrote them, co-sponsors, committee referrals, and public law numbers. | 2013 – present | Free bulk ZIP, no key |
| congress-legislators — open-source civic data project | Names, states, districts, biographical IDs and current committee rosters. | 1789 – present | Free JSON, no key |
| Gallup — polling organisation | Public approval of Congress, from the identical question asked monthly. | 1974 – present | Published tables, cited |
The pipeline downloads these sources, reduces them, and computes the derived measures. The heaviest single file is a 700 MB matrix of 26.4 million individual votes cast — every member on every roll call since 1789. It is collapsed into per-vote party splits and per-member records so it can be queried instantly.
The one honest gap you should know about
Complete bill-level records begin with the 113th Congress in 2013. That is where the Library of Congress's machine-readable bulk data starts. Questions like “was Congress more productive in the 1960s?” cannot be answered here yet, and we do not pretend otherwise. Voting and ideology data go back to 1789 and carry no such limit.
The gap is closable: Congress.gov's API covers bills back to 1973, and the pipeline includes a backfill command for it.
Why this matters
Division is measurable, not a vibe
“Congress is more polarised than ever” is said constantly and rarely evidenced. It is true, it is measurable, and the interesting part is how: the party averages moved apart, but the bigger change is that the members in between stopped existing entirely. Those were the people who made cross-party deals arithmetically possible.
Output is not what people assume
Roughly 30% of the laws Congress passed last session renamed a post office, awarded a medal, or declared a day of recognition. Counting raw laws makes Congress look busier than it is. We separate the two everywhere.
The Senate has quietly changed jobs
It now takes 19 votes a Congress on passing bills and 219 on confirming appointments. In 1989 those numbers were 83 and 24. That is a different institution than the one described in civics class, and almost nobody is told about it.
Accountability needs specifics
“Do something” is not a demand anyone can act on. “Your representative wrote 125 bills and 3 became law” is. The point of making this legible is to make the asks specific.
So what would actually help?
Proposals that people who study Congress have put forward, each tied to a specific number on this site — each with the strongest argument against it.
Stop letting ceremonial bills stand in for legislating
What the data shows. 81 of the 274 laws passed in the 118th Congress — 30% — were post office renamings, medals, or days of recognition.
The proposal
Move naming and commemorative measures onto a consent calendar handled in bulk, as several procedural reform groups have urged, so that “laws passed” stops being a number that can be inflated with ceremony. Some have gone further and suggested delegating facility naming to the agencies that own the buildings.
The case against it
They are genuinely popular in the districts that get them, they cost almost no floor time as it is, and the real bottleneck is committee and leadership scheduling rather than the calendar. Removing them would improve the statistic without changing the underlying output.
A specific thing to ask for. Ask your member whether they would support reporting ceremonial and substantive legislation as separate counts in official summaries.How contested: Low — mostly a measurement fix
Give bills with broad support a path out of committee
What the data shows. 10,361 bills in the 118th Congress — 65% of everything filed — died in the committee they were referred to, without ever receiving a vote.
The proposal
Lower the bar for forcing a committee to act. Proposals range from reducing the discharge petition threshold, to an automatic markup requirement once a bill passes a set number of co-sponsors from both parties, to guaranteed committee votes on bipartisan measures.
The case against it
Committees exist precisely to filter. A forced-vote rule rewards bills built to collect signatures rather than bills built to work, hands the minority a tool to force show votes, and would likely increase the number of messaging bills without increasing the number of laws.
A specific thing to ask for. Ask whether they have signed any discharge petition, and what their standard is for doing so.How contested: Moderate — procedural, but advantages shift with the majority
Unclog the Senate's confirmation queue
What the data shows. The Senate took 219 votes on nominations and only 19 on passing bills this Congress. In -1586 it was 24 and 83.
The proposal
Confirm routine sub-cabinet and non-judicial nominees in blocks rather than one at a time — an approach the Senate has adopted temporarily in both 2011 and more recent sessions, and which scholars at both Brookings and AEI have argued should be permanent. The stated aim is to return floor time to legislation.
The case against it
Individual confirmation votes are one of the few remaining tools an individual senator has to extract information or concessions from an administration, and batching hands the executive branch a freer hand. Scrutiny of appointees is a real constitutional function, not queue congestion.
A specific thing to ask for. Ask your senators whether they would support en-bloc confirmation for non-judicial, sub-cabinet posts.How contested: High — cuts directly at individual senators' leverage
Decide honestly what the 60-vote threshold is for
What the data shows. Cloture motions — the procedure for ending a filibuster — went from 24 in -1586 to 284 today.
The proposal
The options on the table are well worn: return to a “talking filibuster” that requires continuous floor presence, lower the threshold on a sliding scale as debate continues, carve out particular categories of legislation, or abolish the 60-vote requirement for legislation as it has already been abolished for nominations.
The case against it
This is the most contested item on this page and the argument does not divide neatly. Whichever party is in the minority values the threshold, and both parties have defended and attacked it depending on where they sat. Serious people argue that forcing broad coalitions produces more durable law, and that a bare-majority Senate would swing policy violently every few years.
A specific thing to ask for. Ask your senators to state a consistent position that does not depend on which party holds the majority.How contested: Very high — no consensus exists, in either party or the literature
Look at how the people in the middle disappeared
What the data shows. 0 House members now sit in territory the other party also occupies. In 1969 there were 203.
The proposal
Most proposals target the primary, on the theory that in a safe seat the primary electorate is the only one that matters. The main candidates are top-two or top-four primaries, ranked-choice general elections, and independent redistricting commissions. Alaska, California, Maine and several states have adopted versions.
The case against it
The evidence that these reforms actually moderate who gets elected is mixed and contested. California's top-two primary has not obviously produced a more moderate delegation; studies of ranked-choice reach conflicting conclusions; and redistricting reform cannot help the Senate, where the districts are states. Some researchers argue national media and party fundraising, not primary rules, are doing the work — in which case election reform would change little.
A specific thing to ask for. Ask what your member believes caused the disappearance of the middle, and whether they think any election rule would change it.How contested: High — the diagnosis itself is disputed, not just the remedy
Make the record legible by default
What the data shows. Complete machine-readable bill records begin only in 2013, which is why this site cannot compare today's output with the 1960s. Voting records go back to 1789 only because academics rebuilt them.
The proposal
Extend official bulk data backwards, publish committee proceedings in the same structured format as floor votes, and require that amendment and markup activity be released as data rather than PDFs. The House's own modernisation committee recommended much of this.
The case against it
Genuinely uncontroversial, which is also why it is unglamorous and keeps losing to other priorities in appropriations. It is a resourcing question more than a political one.
A specific thing to ask for. Ask whether your member supports funding the Clerk and GPO to digitise pre-2013 legislative records.How contested: Low — the obstacle is money and attention, not disagreement
Sources for the proposals above include the House Select Committee on the Modernization of Congress, the Bipartisan Policy Center, and published work from Brookings and the American Enterprise Institute. Listing a proposal here is not an endorsement of it.