Census Has Big Tech Plans for the 2020 Count. Will They Stay on Track?

In four-and-a-half years, if you’re one of the many Americans who don’t mail in your Census forms quickly enough, you might get a knock on your door from a representative of the federal government.

And if everything goes according to the Census Bureau’s plans, the agency’s 200,000 field workers might look a little different than they did back in 2010: Rather than furiously jotting down your information using the age-old pen and paper, they’ll be equipped with mobile devices.

That’s not the only planned upgrade to the nearly 230-year-old, once-every-10-years survey designed to provide a snapshot of the American populace. For the first time nationwide, Census will allow people to fill out the decennial survey using the Internet, meaning many respondents won’t see the agency’s typical fill-in sheets at all. Officials predict the online option will cut costs and be easier to complete for many users.

While the upcoming decennial count may seem like a long way off, Census is entering crunch time if the agency wants to stay on track with its ambitious tech plans for 2020. Already a number of hurdles stand in the way.

At the same time the bureau is preparing for the count, it’s also hunting for a new chief information officer. After serving as Census’ top tech official since 2009, Brian McGrath announced in July he’s moving to a new post in the Justice Department.

In addition, Census officials are searching for cost-cutting measures; Congress wants the agency to limit the count’s price tag. Meanwhile, a data-collection effort characterized as the backbone of the agency’s 2020 plans, has garnered negative attention from government auditors as a “high-risk” federal IT investment.

Nextgov requested interviews with Census IT staff to get a better read on the agency’s tech plans. However, agency officials declined to make officials available for an interview. Instead, the agency agreed to provide email responses from the bureau’s deputy director.

Key Deadline This Month

One of the first big deadlines on the road to 2020 is approaching quickly.

This month, Census must submit to Congress a detailed design outline setting out how the agency plans to carry out the 2020 census.

The bureau tried to incorporate mobile devices into the 2010 survey, but those plans were ultimately scrapped long before the count actually got underway after they fell too far behind schedule early on.  

This time around, Census is mulling a few different strategies for using mobile devices, according to Carol Cha, director of information technology issues at the Government Accountability Office.

The option that appears to be most popular is a bring-your-own-device model. Census field workers would likely download a Census-developed app onto their personal smartphones that they would use to collect responses from households.

“Bring-your-own-device is being actively explored as an opportunity to reduce reliance on government furnished equipment and reduce equipment and associated support costs,” said Nancy Potok, Census’ deputy director and chief operating officer, in an email to Nextgov.

But even if Census meets the design deadline, the bureau has already fallen behind schedule.   The agency’s original deadline was actually last September, according to Cha, but that was postponed because Census had not yet completed the necessary planning for research and testing.

Meeting this month’s deadline is key for the agency.

“This decision cannot afford to slip any further beyond that September 2015 date because they have to start acquiring IT contracts, start building these systems and also have them adequately tested before they start the end-to-end testing in October of 2018,” Cha told Nextgov.

Learning from the 2010 Tech Bust

While missed deadlines also sank Census’ last attempt to use mobile devices, it’s worth noting the agency’s approach to using mobile devices is far different than the failed attempt five years ago.

The ambitious strategy back then -- in an era before the first iPhone had rolled off the production line and the term “mobile device” had barely entered the popular lexicon -- was to create new “handheld computers” from scratch to be used by Census field workers. In 2006, Census awarded a $595 million contract to Harris Corporation to develop the devices.

Because of a combination of technical issues, insufficient management and ineffective oversight, the agency started to fall behind schedule, auditors say. About two years before the start of the 2010 Census, officials realized they had fallen too far behind and would not be able to complete end-to-end testing, a crucial step in determining the actual performance of a system in production.

Faced with the slipping deadlines, Census pulled the plug on its mobile device plan in favor of traditional paper-based forms.

The cost of reverting back to pen and paper increased the total price tag of the 2010 census by $3 billion and seeded doubts in the minds of many lawmakers about the agency’s ability to embrace new technology.

“... With all the modern scientific improvements and technological advancements that have been made over the years, the framework for conducting the 2010 Census was based off of a model we first used in the 1970s,” said Sen. Tom Carper, D-Del., in a 2012 hearing -- one of many -- on lessons learned from the failed effort.

Too Much on Census’ Plate?

The bureau’s 2020 tech plans primarily fall under two large-scale projects: Census Enterprise Data Collection and Processing, or CEDCAP, and Reorganize Census with Integrated Technology, or ROCkIT. (No, this isn’t a typo).

These are only two of five massive initiatives Census is pursuing to revamp how it collects and uses data, according to a March report by National Academy of Public Administration, which evaluated the agency’s overhaul efforts.

CEDCAP, the program aimed at revolutionizing the agency’s approach to data collection, is made up of 14 separate IT projects, only a handful of which are related to the 2020 count. Other pieces of the plan aren’t scheduled to be completed until 2022. The whole project wound up on the latest update to GAO’s “High Risk List,” which tracks problematic federal programs.

Is there simply too much on Census’ plate, especially as it preps for crunch time?

“We do think that simplifying the solution, perhaps maybe taking on the top two or three initiatives versus this large swath of initiatives. . . would be in their best interest,” Cha said.

But Roger Kodat, the project director of the NAPA study, does not agree. He, along with a panel of academy fellows, interviewed more than three dozen Census and Commerce Department officials, including several from Census’ Information Technology Directorate.

“We saw how the five initiatives are linked to each other,” Kodat said. If the agency instead focused on only a few of them at a time, “you’re missing an opportunity for synergy,” he told Nextgov.

New Tech Key to Cost Savings

Along with yanking Census out of the 1970s, the new tech options are designed to save the agency some cash.

The cost of the decennial census has increased dramatically since the 1970s. In 2010, alone, the overall cost was $13 billion -- 56 percent more than in 2000 and 500 percent more than the survey completed in 1970, according to GAO.

If the 2020 count were to experience the same proportional increase, that would put the total price tag at about $25 billion, GAO predicted.

Census planners have pledged to keep costs down -- promising as much as $5 billion in savings -- in part through the use of new technology.

For one, Census officials expect to save as much as $548 million by cutting down on door-to-door visits, including allowing respondents to use the Internet to fill out forms.

“By allowing respondents to easily answer the questionnaire online, we can save millions on the costs of mailing out, getting back, scanning and hand-keying the information from paper forms into our system,” said Potok, Census’ deputy director and chief operating officer, told Nextgov in an email.

Equipping field workers with mobile devices could net even more savings. Successful implementation of smartphones or tablets to manage field operations could save the agency as much as $2.3 billion, according to GAO estimates.

“If that Date Slips...’

2020 may seem like a far-off finish line. But the deadlines are already piling up -- and Congress is getting concerned.

In an Aug. 24 letter to Comptroller General Gene Dodaro, members of the House Oversight and Government Reform Committee highlighted the bureau’s unreliable track record when it comes to acquiring and implementing new IT solutions and asked auditors to beef up oversight of the agency as it preps for the 2020 Census.

Census is planning a 2016 test run to make sure the decennial count goes according to plan. The test, which will take place in Houston and Los Angeles, is expected to allow the bureau to assess both the Internet response option and the feasibility of equipping field workers with mobile devices. The test run will include both new Census-owned devices as well as a “bring your own device” option and will help the agency finally settle on a plan of action.

By 2018, Census plans to complete end-to-end testing on whatever mobile devices it ultimately selects, according to Cha, the GAO official.

“That date sounds pretty good from an IT implementation perspective because that will give them enough time to work out the kinks, debug the end-to-end issues in time for 2020,” she said. “But if that date slips to the right then we have some issues.”

Missed deadlines some eight years ago left too little time for end-to-end testing and ultimately halted the mobile device plans. This go-around, managing the deployment to make sure there’s enough time for end-to-end testing is just as crucial, auditors say.

“End-to-end system integration testing is critical,” Cha said. “When you compress that or forego that altogether, you have HealthCare.gov.”

(Image via Everett Historical/ Shutterstock.com)