Read the Bill 2.0


People have become increasingly aware that their elected representatives often do not read the legislation they vote upon. Sometimes there’s not enough time between the introduction and adoption of legislation for anyone but the bill’s sponsor to grasp the contents. Other times so many amendments are added that few people — if any — fully understand the final document. In the haste to build support and pass bills, provisions are inserted unnoticed that contain unrecognized flaws and political pay-offs. The public and legislators alike are disconnected from the legislative process intended to serve their needs

Congress and the general public need tools to access and analyze legislation. They need Read the Bill 2.0.

Many politicians and politically-aware citizens seek to reconnect the legislative process through their support of the “read the bill” reform. Its central premise is the idea that legislators and members of the public should have enough time to read legislation before it is voted upon. The Sunlight Foundation, one of the primary advocates of this idea, has long recommended that non-emergency legislation and conference reports be posted on the Internet for 72 hours before final consideration.

In the expiring Congress, this idea gained enough political currency that some of its major legislation — including health care and financial reform — was available online 3 days prior to final consideration per  Speaker Pelosi’s promise to do so.  However, no legislative rule imposed this requirement, it was not always imposed uniformly, and often times you had to (1) know where to look to find the bill and (2) be willing to thumb through PDFs containing up to 1,000 pages. Even so, it was a commendable step forward.

Reps. Baird and Culberson introduced legislation (H. Res 554) that would have gone one step further and formalized the “read the bill” requirement by amending the House rules; it garnered 217 co-sponsors. Presumptive-Speaker John Boehner has pledged to impose a “read the bill” requirement for all legislation, but it is not yet clear whether he will update the House rules to do so. We think he should.

Anyone who has tried to read legislation soon realizes that doing so can be complicated and difficult. The text is often written in a shorthand that can be unlocked only with patience and frequent reference to the U.S. Code and Statutes at Large. It’s difficult to see how draft language changes over time because there is no redlined version of the legislation, and sometimes you need to compare two lengthy PDFs against one another. The bills themselves don’t always show up on THOMAS, the official online legislative database, in a timely manner, and they are difficult to find when published elsewhere.

Congress and the general public need tools to access and analyze legislation. They need Read the Bill 2.0. There are three main technological obstacles.

Legislation should be available online in real time, in machine-readable format, and accessible in bulk.

First, legislation that’s supposed to be available on THOMAS isn’t always there on time. The Senate version of the DISCLOSE Act, for example, took more than 4 days to appear on THOMAS after it was introduced. When legislators bypass THOMAS and directly publish legislation online to satisfy the 72-hour rule, it’s often difficult to find. THOMAS must be comprehensive and timely.

Second, all legislation should be published in a machine-readable format. PDFs aren’t good enough. Unless the data is in a machine-readable format, you cannot easily perform a redline comparison between two bills, like I did here, which showed how the DISCLOSE Act changed over time. Fortunately, 97% of the legislation published on THOMAS this past Congress is available in XML, which is an incredibly useful machine-readable format that provides a helpful structure to the data. But 3% of bills are available in PDF format only, and those bills are usually the ones that are most important, lengthy, and time-sensitive.

Third, the THOMAS database should be available to the public “in bulk,” which would allow users to download large amounts of information at one time. Currently, third parties write programs that extract human-readable information from THOMAS in a time-consuming and error-prone process known as scraping. They then use this information to build citizen-empowering tools like OpenCongress and GovTrack. This kind of innovation — allowing people to use congressional information to better understand what Congress does — should be encouraged.

Indeed, in 2009 the Library of Congress and GPO were directed to form a task force to report on the feasibility of bulk access to THOMAS, but no report has materialized. Rep. Foster recently introduced legislation (HR 6289) to partially implement bulk access to THOMAS data and jump-start the report. The incoming Congress should clear the pipes and let public legislative data flow to the public.

Real-time redlining of the US Code become possible once you combine bulk access to information with publishing legislation in a machine-readable structured-data format. Wouldn’t it be great if you can see at a glance how a bill would change the law? Or how an amendment would change the text of a bill? Congress could build this tool, and others, or could require that the legislative text be structured so that outside parties could do so. It could also take steps to improve THOMAS generally. Regardless, nothing will likely happen without congressional action.

The goal isn’t merely to allow legislators and citizens to be able to read the bills, but to understand them as well.

For “read the bill” to have bite, people need better tools to make use of legislative information. Publishing PDFs of bills online 3 days prior to a final vote is necessary, but insufficient, for legislative transparency. The kinds of tools we need can only arise with faster, better access to raw legislative information. We need Read the Bill 2.0. After all, the goal isn’t merely to allow legislators and citizens to be able to read the bills, but to understand them as well.