Skip to main content

What really matters

In a world with too much noise and too little context, Vox helps you make sense of the news. We don’t flood you with panic-inducing headlines or race to be first. We focus on being useful to you — breaking down the news in ways that inform, not overwhelm.

We rely on readers like you to fund our journalism. Will you support our work and become a Vox Member today?

Join today

How big data is helping states kick poor people off welfare

“These systems make our values visible to us in a way that calls us to a moral reckoning.”

Shutterstock
Sean Illing
Sean Illing is the host of The Gray Area podcast.

Technology is being used to target and punish poor people in America, according to a new book by Virginia Eubanks, a professor of political science at the University of Albany, SUNY.

Here’s one example: In 2014, Maine Gov. Paul LePage released data to the public detailing over 3,000 transactions from welfare recipients using EBT cards in the state. (EBT cards are like state-issued debit cards, and can be used to disperse benefits like food stamps.)

LePage created a list of every time this money had been used in a strip club, liquor store, or bar, and used it to push his political agenda of limiting access to state benefits. LePage’s list represents a tiny fraction of overall EBT withdrawals, but it effectively reinforced negative stereotypes and narratives about who relies on welfare benefits and why.

I spoke with Eubanks recently about her new book, and why she believes automated technologies are being used to rig the welfare system against the people who need it the most.

A lightly edited transcript of our conversation follows.


Sean Illing

What’s the thesis of your book?

Virginia Eubanks

There’s a collision of political forces and technical innovations that are devastating poor and working-class families in America.

The political forces I’m talking about are a backlash against racial justice victories and the scapegoating of immigrants. On the technical side, we have this explosion of technologies that go under the rubric of algorithmic decision-making or predictive modeling. These technologies increasingly decide who gets public services, who is denied public services, and how we monitor and police the most marginalized people in our society.

Sean Illing

This feels so abstract when you lay it out like that. It’s hard to connect these technological innovations to what’s happening on the ground.

Virginia Eubanks

It does sound abstract, and it is hard to see the connections — but they’re there. In the book, I try to tell stories about systems across the country and link them to concrete results in people’s lives. I talk about an automated eligibility system built in Indiana in 2006 that was supposed to automate the eligibility processes for all of their welfare services, but instead brought poor people under immense scrutiny. I talk about an electronic registry for the homeless in Los Angeles that has had similarly awful consequences. And on, and on, and on.

What’s different about this book is it’s told from the point of view of the people who are targets of these systems. I talked to designers and administrators, to folks who built the systems, but I also think it’s really important to hear the voices of those who are impacted directly. We don’t often hear these voices in stories about technology and political change.

Sean Illing

So tell me how data-based technologies have changed the way public services are managed.

Virginia Eubanks

I don’t actually see the systems changing the way we think about public services or provide public services so much as I see them intensifying the way that we’re already providing services. They’re a really great diagnostic tool for when we have a social justice problem that we’re not addressing directly.

Sean Illing

That’s interesting. Give me an example.

Virginia Eubanks

The 2006 Indiana system was built on two assumptions.

One was that most people who were in the public service system didn’t need to be there, and that if they just had a little push, they wouldn’t collect the resources that they were entitled to by law. The second assumption was that all case workers do is process paperwork and collude with folks who are requesting resources in order to defraud the system.

Basically, it was built on the assumption that the welfare system is rife with fraud and most people collecting resources don’t need them. That assumption turned into some very brittle rules that were part of the technological system, like a rule that said, basically, if you made any mistakes at all in your application (which can run anywhere from 35 to 120 pages), then you’ve failed to establish your eligibility and can be denied.

“I think the real solution for us as a nation is to get our souls right about poverty. Until we do that, we will continue to produce systems that profile and punish poor and working families.”

Sean Illing

In other words, politicians are making a value judgment that the welfare system is itself a scam, and most of the people relying on it are fraudsters.

Virginia Eubanks

Precisely. And they’re designing the technologies to produce the results they want: less and less people getting the resources they need.

Sean Illing

You cover a lot of individual stories in the book. Can you tell just one of them here to give people a sense of what you mean?

Virginia Eubanks

One story is about a guy named Gary who has been living on the streets on Skid Row, Los Angeles, on and off for 10 years. He’s trying to get access to housing through a system there called the Coordinated Entry System.

The Coordinated Entry System is supposed to match the most vulnerable homeless folks with the most appropriate available housing. There’s this incredibly long and very intense survey called the VI-SPDAT that seeks to give people a vulnerability score, looking at how close to death or institutionalization or hospitalization a person is. The stated goal is to prioritize folks who are most vulnerable for available housing, which makes a lot of sense.

Here’s the issue: the questions on the survey are questions that can require unhoused people to admit to illegal activities. In fact, the more of these you admit to, the higher you score on this survey, and the more vulnerable the system sees you as. The problem is, this information is then shared with more than 168 different organizations, including the Los Angeles Police Department, which can access this data without a warrant or oversight.

So Gary is extremely vulnerable — he’s over 60, has health problems, has been in and out of the criminal justice system. But instead of getting access to housing, which is the only reason he applied to this system, he ends up getting arrested for something he claims not to have done, and it’s partly because his name was flagged in this database as a suspect.

When he comes out of jail, he’s lost all of his resources. He’s lost his network of people he knew on the street, he’s lost his tent, he’s lost all of his paperwork. But the system actually now will see him as less vulnerable, because it counts jail as housing — it would see him as being housed for the last six months. He’s become lower priority.

Sean Illing

So these systems are getting more efficient at denying people the help they need and at the same time bringing them under more and more scrutiny?

Virginia Eubanks

That’s exactly right. There’s this double-edged sword of system integration. On one level, it’s incredibly difficult to navigate the public services system. You have to fill out a million applications. If you want home heating assistance, food stamps, cash assistance, you have to fill out a separate application for each of these.

Integrating these systems could be a very positive thing for people if it were set up to provide people with the resources they need to meet basic human needs, but that’s not what we have. The system we have has been set up to pass moral judgment on people because it works on the assumption that poor people are poor because of some personal failure on their part.

It’s also about creating a network of information between agencies that makes it harder for people to get the help they need, but does a very good job of using their personal data to create new barriers for them.

Sean Illing

And marginalized groups are more exposed to this system of surveillance precisely because they need to access the benefits and programs connected to it?

Virginia Eubanks

And because they live in what I think of as low-rights environments. The kinds of scrutiny that happens in poor working class neighborhoods, or in communities of color, or in migrant communities — those processes of surveillance are much more intense in those neighborhoods. They’re overrepresented in all of these databases because they access public resources, but also because their communities are oversurveilled.

Then, because we have this data, we create new tools that only focus on them and we reproduce this feedback cycle, what the Stop LAPD Spying Coalition calls the feedback loop of injustice.

“Ultimately, these systems make our values visible to us in a way that calls us to a moral reckoning.”

Sean Illing

Are poor people basically being used as guinea pigs in preparation for a much broader use of these technologies?

Virginia Eubanks

I think that’s absolutely true. In one of the early parts in the book, I write about a previous project I did that focused primarily on families on public assistance in my hometown, in Troy, New York. A real wake-up moment for me was when I was sitting with a young mother and we were talking about technology and her use of the public assistance EBT debit cards. I told her that I’ve heard people say they like EBT cards because they’re convenient and more efficient.

She said, “Sure, it’s nice to not have to carry around food stamps, but it’s also invasive because my case worker uses it to track where I buy food and spend money.”

She then told me, “You middle-class folks should watch out, because you’re going to be under the microscope next.” This was in 2000. I think there’s good reason to expect that these tools we’re talking about will be used far more broadly in the future.

Sean Illing

Are we entering a kind of soft, digital tyranny that is too nebulous or distant for most of us to notice?

Virginia Eubanks

That’s a difficult question. I think that these tools allow us a moment to think deeply about what we allow government agencies to do on our behalf. These tools are incredible indicators of what we already think, both about poor working-class communities and families, and about government.

Ultimately, these systems make our values visible to us in a way that calls us to a moral reckoning. A lot of readers want me to give a 10-point plan for creating better technologies for public services, but I’ve resisted doing that because I think the real solution for us as a nation is to get our souls right about poverty. Until we do that, we will continue to produce systems that profile and punish poor and working families.

More in Technology

The real argument artists should be making against AIThe real argument artists should be making against AI
Artificial Intelligence

AI companies are inflicting moral injury on artists.

By Sigal Samuel
The Trump administration says cheap goods aren’t part of the American dream. They’re wrong.The Trump administration says cheap goods aren’t part of the American dream. They’re wrong.
Future Perfect

Everyone likes cheap consumer goods that will get more expensive under Trump’s tariffs.

By Kelsey Piper
A surprisingly useful tool for this year’s tax seasonA surprisingly useful tool for this year’s tax season
Even Better

Doing your taxes is technical and boring. It’s perfect for AI chatbots.

By Adam Clark Estes
The self-inflicted death of American science has already begunThe self-inflicted death of American science has already begun
Future Perfect

Trump’s crackdown on foreign students and scientists will do irreparable harm to the country.

By Bryan Walsh
Will AI become God? That’s the wrong question.Will AI become God? That’s the wrong question.
Podcast
The Gray Area

Tech pioneer Jaron Lanier argues for a different way to think about AI.

By Sean Illing
Why are car headlights so blindingly bright now?Why are car headlights so blindingly bright now?
Podcast
Explain It to Me

Bright headlights can make it easier for drivers to see the road. But there’s a dark side.

By Hady Mawajdeh