Labor Does More With Less on Streaming Data


The agency has a single Application Programming Interface that connects to 175 data sets.

When the White House directed agencies to make at least two data services available through a streaming process known as application programming interfaces in 2012, most tech and transparency savvy agencies focused on releasing as many APIs as they could.

The Health and Human Services Department, for example, released 61 APIs connecting to 61 datasets.

The Labor Department took a different tack, releasing 175 datasets through a single API.

An API is a method for one computer to continuously receive updated information from another, such as the Twitter streams and today’s weather icons embedded in some news websites. The API mandate was part of the White House’s Digital Government Strategy.

The single API method has several advantages, the Labor Department’s lead information technology specialist Mike Pulsifer said.

First, a single API means developers only need to learn one system to access any Labor data sets they’d like, he said. Second, if the department wants to release new data sets, it can simply tack them onto the existing API rather than build new digital infrastructure.

Finally, if department officials need to fix something for security or some other reason, they can do so with one patch rather than dozens or hundreds, he said.

Pulsifer was inspired by hobby work he’d done with the photo sharing site Flickr, which similarly offers a single API to reach all its services.

Pulsifer had an additional advantage when he began work on his API in 2011. Numerous Labor divisions had released data sets in downloadable formats but none had ventured into API territory. That meant, with a little persuasion, he’d be able to build a departmentwide streaming open data system from the ground up rather than herding many existing efforts into one.

The Health and Human Services Department, with its many data savvy divisions, on the other hand, had numerous legacy APIs.

Several large private sector tech companies launched multiple APIs in the early days of the technology but then slowly pared down to a single API later, said Gray Brooks, who is managing the government API initiative from inside the General Services Administration.

“What we did is started really small with three data sets that were of admittedly limited usefulness,” Pulsifer said. “We didn’t want to jump out with something that would be so popular that it would get hammered right out of the gate and we’d be stuck trying to figure out how to keep it running.”

Once the API had proved its usefulness, other Labor divisions, including the Bureau of Labor Statistics, agreed to link their data onto it. By the time the White House API deadline passed on May 23, the nearly two-year-old Labor API connected to 175 datasets.

Labor attached a login function to its API so it could measure which data sets are the most popular and put additional resources toward making them available.

The agency isn’t tracking how individual developers use its data sets, Pulsifer said, but would be open to advice from particular developers about data they’d like to access more easily. 

The department has also published a series of software development kits -- essentially toolkits to build applications in various computer languages or on different mobile platforms -- that other agencies could easily piggyback off, Brooks said.  

“We’ve got a tremendous amount of data that we’d love for app developers out there to turn into information,” Pulsifer said. “The stories that can be told from this data, that’s what we’re really hoping the public can produce out of this.”

(Image via Haver/