Open data holds the promise of connecting legislation in new ways, bringing federal, state and local laws together and automating compliance. But teaching machines to read law –- or lawmakers to read machines -- is no simple task.
(Artem Samokhvalov / Shutterstock)
Promising to disrupt the legal, accounting and consulting fields in front of an audience of consultants?
That’s the Data Transparency Coalition.
At a July 28 breakfast, the coalition and its partners championed the transformation of the legislative process with open data – and teased new “legislation about legislation” that could help speed that transformation.
There are several open data avenues being pursued by legislative and regulatory bodies worldwide, including UN favorite Akoma Ntoso and U.S. Legislative Markup, the schema for producing the U.S. Code in Extensible Markup Language (XML).
Hudson Hollister, executive director of the Data Transparency Coalition, was bullish on the promise of machine-readable laws.
“We believe that if the government begins to embrace the notion of law and regulation as data, eventually that will lead to the automation of compliance,” he said, positing that existing financial software could easily be linked to open data legislation for automatic updates upon enactment of new laws.
Other open data pros: automatically linking new laws to the old laws they change, connecting federal laws to state and local laws, and possibly even rooting out mistakes in old legislation.
Dave Zvenyach, an 18F director who served as D.C.’s general counsel until very recently, called for a “pipeline” between different layers of government “so that the moment Congress passes a law, we’ll have knowledge of what it affects [at the state and local level].”
Zvenyach recalled tiptoeing into open data and discovering, in one instance, that an old law had two sections titled “D” and included references to another section that didn’t exist.
Hollister touted that an “ally” of open data would introduce “legislation about [XML-based] legislation” soon. (A Data Coalition spokeswoman told FCW that the group had hoped to announce the legislation at the July 28 session, but was waiting for the OK from the lawmaker’s office. She confirmed that the legislation would be introduced in the House, likely this week.)
Bridging the divide
The march to open data legislation has not been without hurdles.
There’s a “philosophical divide,” Zvenyach said, in approaches to the process: Do you start with XML and rich tagging, or do you infer meaning from context, using natural language processing “in the same way that humans look at a document”?
Essentially, do we force lawmakers to write what computers can read, or do we teach computers to read more like lawmakers?
“They’re not incompatible strategies,” Zvenyach noted. “They may be able to bridge the divide between rich semantic documents and documents that are unstructured.”
Other challenges include the fact that many members of Congress don’t much seem to care – Zvenyach said lawmakers tend to be “agnostic” on open data, while House Committee on Administration Director of Technology Policy Reynold Schweickhardt blamed the “inertia” of a hundreds-of-years-old congressional culture – and presentation kinks still need to be worked out.
Schweickhardt and Kirsten Gullickson, senior systems analyst for the House, have been working since 1997 to put laws online, open up the legislative process and connect laws with rich tagging.
They said some of the weirdest hang ups have come from the Government Publishing Office, which tends to format laws to look good on the page instead of using the right machine-readable tags.
The difference between machine- and human-friendly even reaches the indentation level.
“To lawyers, indents are really important to how [a law] looks and how they read it,” noted Gullickson.
But challenges aside, open data legislation is on a forward march, and it’ll take over government “sooner than you think,” said Hollister.