Earlier this year, the tech company Novo Dia Group announced it would not continue as a vendor with the US Department of Agriculture, due to a switch in federal contractors. What seemed a run-of-the-mill business decision threw a very real wrench into the availability of locally-grown foods for low-income Americans.
The problem was that Novo Dia held the only keys to a USDA program dedicated to Supplemental Nutrition Assistance Program processing software and equipment for 1,700 farmers’ markets nationwide. Without Novo Dia providing this service, markets would have no way to accept SNAP — a disruption that would cost farmers income and SNAP recipients food.
If you’ve ever attempted to switch your cell phone provider but keep your actual device, you might be able to relate: Farmers’ markets had perfectly functional and expensive equipment that simply would not work with any other SNAP processing software. It’s the government equivalent of trying to keep your iPhone when you move from Verizon to AT&T.
This episode raised a lot of questions about the government’s relationship with tech companies tasked with administering public programs: How does it choose who to hire? How does it hold those companies accountable? And how do those decisions affect the daily lives of low-income Americans who rely on being able to access their benefits?
The answers are vitally important: Governments are increasingly relying on new technologies to sort applications, manage caseloads, and distribute benefits. How such technology is contracted, developed, and deployed will have real impacts on millions of low-income Americans.
Take, for instance, what happened in Washington, DC. In the fall of 2016, the city’s Department of Human Services, along with the contractor Infosys Public Services Inc., replaced a computer system the District had been using since the early 1990s to enroll low-income residents in SNAP and cash assistance programs. The Food and Nutrition Service, the USDA agency responsible for administering the country’s nutrition assistance programs, issued a letter to the DC Department of Human Services, warning against launching the new system without having done adequate testing.
But two months later, DC rolled out the system anyway — to repeated outages and glitches, including benefits not being loaded onto Electronic Benefit Transfer cards.
Frustrations between agency employees and customers ran so high that there were physical altercations in some enrollment offices, causing the union representing the workers to issue a formal grievance. The union asked that the agency return to using the previous technology or distribute hazard pay to employees.
Rhode Island, meanwhile, has been struggling to serve its SNAP recipients since it rolled out a new $364 million computer system in 2016 — known as the Unified Health Infrastructure Project — causing delays in distribution by the thousands. Recently, the Food and Nutrition Service threatened to withhold more than $900,000 in federal reimbursements due to Rhode Island’s continued failure to address a list of nearly 30 items related to system functionality, issuance of benefits, backlogs, certification, and more.
In turn, state Department of Nutrition Services Director Courtney Hawkins blamed Deloitte Consulting, the company contracted to build the computer system saying, “This formal warning underscores the fact that Deloitte has not yet delivered a fully functioning system that works on behalf of Rhode Islanders.” In April, the company apologized for its disastrous roll-out.
To date, two federal class action lawsuits have been filed against Rhode Island over its SNAP program. Recently, it was reported that the total cost of its new computer system had reached “$647.7 million through the 2019-20 federal fiscal year, with $138 million of that amount to be covered by state taxpayers and the rest by the federal government.”
Part of the problem in developing these systems is how the government chooses which companies to hire, said Dave Guarino, director of GetCalFresh, a project of Code for America. He notes that there are only “a small number of vendors who know how to navigate the procurement process, and they’re the ones who get the contracts.”
Thus, the proposal and bidding process limits the amount of competition and creates stagnancy in the technology developed for government programs. It also leaves out newer, smaller, and more innovative companies.
In theory, this is because the government process is designed to decrease risk, given the high amount of sensitive and confidential information managed by these systems, so it’s the well-known contractors with a track record of managing large projects who ultimately get the gig.
But Guarino says that government technology crises, such as IT disasters for SNAP recipients, highlight the need for a true shift in thinking about risk and agility. “We should be demanding better software and better experiences,” said Guarino. “But if we want government to be able to act more nimbly and quickly, we also need to be able to say that government can take more risks.”
While risk-taking can have downsides, Guarino said the best practice is to test new ideas “on a really small scale in a way that minimizes risk, but maximizes learning” — a concept that could have helped to prevent harm caused by the failure of the DC system roll-out, as problems could have been spotted and fixed with a relatively small control group.
Guarino also noted the importance of working with a broad range of partners to develop and administer technology, as well as dividing up tasks to “the best firm for each job.”
His own project, GetCalFresh, is one such successful model. GetCalFresh offers online SNAP applications for 36 counties in California, and its technology was developed to measure and remove barriers that often prevent low-income people from accessing their benefits. Users can easily submit SNAP applications by mobile phone or computer, often in fewer than 10 minutes, and can also send verification documents securely via their phones. And by working with a wide range of partners, including Code For America, state and county agencies, and organizations, Guarino said the project is more successful than it would be with a single entity at the helm.
“These things often aren’t talked about as dimensions of why poverty persists and why some poverty solutions don’t reach everybody they could,” said Guarino, “But they’re a really huge deal.”
The thousands of farmers and customers affected by the Novo Dia debacle would likely agree. And as DC, Rhode Island and surely other places have proven, short-sighted decisions and worse implementation of new government tech can adversely impact scores of people. Indeed, if we want innovative, effective poverty solutions in today’s digital landscape, we need to think hard about tech.