How government data is at risk

Open data advocates and former government employees have concerns about the fate of open government data, given budget and staffing cuts and the politicization of science and health care policy.

At a Feb. 27 event hosted by the nonprofit Information Technology and Innovation Foundation's Center for Data Innovation, Paul Farber, the managing director of the University of Pennsylvania's Program in the Environmental Humanities, characterized the federal stances on data to date as "a mixed bag."

"In one sense, open data initiatives still remain," he said. "But a number of the administration's tactics around data writ large have to do with data that's proprietary" and surrounding "things that are tied to contentious issues."

The Department of Agriculture, Department of Health and Human Services and Environmental Protection Agency, among other agencies, have changed the way they collect and publish data.

"Access to data is access to power," Farber added. "It's the ability to understand what you've inherited and what you are shaping."

Denice Ross, a former Presidential Innovation Fellow who is now a Public Interest Technology Fellow at the nonprofit New America, said the current danger surrounding open data is "that it's not an active priority."

"If the will's not there, the data's not going to flow as freely," she said.

On the whole, "the toothpaste is out of the tube with open data," she said, "but there's still risks and still many ways to undermine open data."

For example, Ross said officials can set up "bureaucratic barriers" -- such as increased involvement of communications teams or requiring the added step of making a FOIA request before sending data out. She also noted there are ways to "slow walk" data releases -- such as saying more approval is needed before releasing or even by changing the data schema or aggregating data dumps so they're more difficult to use.

In these cases, Ross said, "technically the data are still available, but it's much harder for regular people to access."

Ross specifically pointed to the FBI's publishing about 70 percent fewer data tables in the 2016 Crime in the United States report, the first released under the Trump administration, compared to the 2015 one. An FBI spokesperson justified withholding data tables by reviewing "the number of times a user actually viewed the tables on the internet," according to a FiveThirtyEight report.

"That's not a great way to assess data value, but that seems to be where we are," Ross said.

While ITIF Vice President Daniel Castro said there have been instances of the "intentional undermining" of dataset, he said "the biggest risk is in data quality."

Staffing and budget cuts to statistical agencies and offices also make it harder for government to ensure the timely and complete release of quality open data sets.

"The big deal is that the agencies are all getting squeeze tremendously," said former Census director John Thompson, who now serves as executive director of the Council of Professional Associations on Federal Statistics.

With fewer staff and resources, Ross said, there "may be less quality control" over the data supply chain.

She pointed to the importance of human input, from the collection and entering of data, to its cleaning and eventual publishing, in making sure data is put out in a useful way.

"Any failure in the supply chain weakens it," she said.

In Congress, there are some open data legislative efforts. The Open, Public, Electronic and Necessary Government Data Act, which would codify Obama's 2013 executive order to require federal agencies to publish their information to Data.gov in a non-proprietary, machine-readable form, passed the House in November 2017 as part of Speaker Paul Ryan's (R-Wis.) bill on evidence-based policymaking.

The bill passed the Senate before the close of the last session of Congress, but must pass the Senate again this Congress.