© 2024 KOSU
Play Live Radio
Next Up:
0:00
0:00
0:00 0:00
Available On Air Stations

'Recoding America' author Jennifer Pahlka on how we can reshape our government to work for us

The U.S. Capitol stands. (Drew Angerer/Getty Images)
The U.S. Capitol stands. (Drew Angerer/Getty Images)

Too often state and federal government fails to provide the services its laws intend.

“We’ve been trying to fix this problem with more money for technology in government, more oversight and more rules,” author Jennifer Pahlka says. “And the evidence shows that’s not working. We got to take a different approach.”

Outdated technology is part of the problem. But in Pahlka’s new book “Recoding America,” she argues the biggest issue is how little policymakers care about implementation.

“They see implementation as a sort of detail that less important people should deal with,” she says. “And until we change that, we’re going to continue to have problems getting the outcomes we want.”

Today, On Point: How government culture gets in the way of better outcomes.

Guests

Jennifer Pahlka, author of the new book “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better.” Former deputy chief technology officer in the Obama White House. Founder and a former executive director of Code for America, a nonprofit that aims to help government agencies with their tech issues.

Also Featured

Erica Chan, helped many Californians navigate the unemployment process during the pandemic.

Natalie Kates, former product manager for the United Services Digital Service and former CTO for the COVID response at the Biden White House.

Michael Nugent, worked in government for about 25 years, most recently as the director of the Defense Language and National Security Education Office for the Department of Defense.

Book Excerpt

Excerpt from “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better” by Jennifer Pahlka. Not to be used without permission of the publisher. All rights reserved.

Transcript

Part I

MEGHNA CHAKRABARTI: Erica Chan remembers March 12, 2020, like it was yesterday. The pandemic. She was sitting at home when she got an email out of the blue saying work was canceled, indefinitely.

ERICA CHAN: It wasn’t just for me. All of my friends was getting work canceled. And so it wasn’t uncommon, and it kind of seemed like it was happening to everyone.

CHAKRABARTI: Erica works in the film industry in Los Angeles, and the pandemic hit the media and entertainment world hard. She knew work in her industry wasn’t coming back anytime soon. So she decided to file for unemployment, something she’d done before when work dried up.

CHAN: So I posted on my own personal Facebook telling people that I know it’s confusing, and if they need help applying that they should ask me whatever questions that they had.

CHAKRABARTI: Erica’s Facebook got flooded, but the problem was Erica didn’t always have the answers.

CHAN: I decided to create a Facebook group, because maybe other people can help other people. And so I started that Facebook group, I think maybe around March 17. And I think by April, May, it ended up growing to maybe 20,000 members. Maybe by June, it was like 40,000. And as of today, it’s like 70,000 members who ended up joining. So it became a really big thing, unexpectedly. And I realized how much people actually really needed that help.

CHAKRABARTI: This is On Point. I’m Meghna Chakrabarti. By July of 2020, about 4.4 million Californians were receiving unemployment benefits. But hundreds of thousands of others who needed help were not. Erica was trying to help them. She says the applications length wasn’t the problem. It took only about 30 minutes to 60 minutes to fill out. The real issue, it was hard to fill out correctly.

CHAN: It was very confusing. Because the way they would word the questions didn’t really make the most sense. And I think partly it was because maybe the government has to write their questions in a way where it was legalese, where it was proper.

But to a common person, sometimes that stuff doesn’t really make sense. You have to read it a couple of times to make sure you’re answering things correctly. And especially with the pandemic, a lot of things changed. They had a lot of different questions.

CHAKRABARTI: For example, there was one question that asked, “Are you too sick or injured to work?” The question didn’t take into account the pandemic and COVID spread. The applicant must answer no, even if that wasn’t true for the applicant at the time. If someone put any other answer then no, applications would be flagged or denied. Once the application was filled out, it still took 3 to 4 weeks to process. If you didn’t get an automatic approval after that time, you’d have to call California’s Employment Development Department. And Erica says that was a nightmare.

CHAN: When I was helping people call, I would block out 4 hours to get through. Because first you have to call in. And you speak to a tier one, the tier one, which is usually somebody who couldn’t help you, they could just take information, and make sure you weren’t asking something basic.

And when they realized that you have to actually speak to a tier two, who can actually adjust your claim, they transfer you. This whole process takes about 3 to 4 hours. Once you reach this tier two and they, let’s say, are somebody helpful, and they want to make changes to your claim, and they say, “Check back in two weeks.”

CHAKRABARTI: Estimates from the state’s unemployment office show that with the sharp rise in unemployment applications due to the pandemic, only about one in every 1,000 calls were ever answered. So a little industry sprung up around that. Phone services, where people could pay someone else to wait in the queue for you.

CHAN: Honestly, [I] think a part of the issue is the disconnect, because you’ve got a lot of people who are working the phone lines, but not a lot of them understand the struggle and the pain of actually applying and going through it. It just seemed that way. Because when you speak to somebody, have they understood what was actually going on the ground, maybe they could have been more efficient with fixing what the actual issues were.

CHAKRABARTI: In 2020, Erica ended up spending most of her time helping other people file their unemployment claims. So much so, that after about eight months of doing this, she had to start charging a fee.

CHAN: Well, there’s obviously something broken with the system if people are trying to find answers elsewhere. Had the system been working, had people been able to speak to the agency, to be able to resolve their issues immediately, there wouldn’t be a need for outside sources to help. So there’s something wrong. There’s something that’s not working.

CHAKRABARTI: Erica Chan lives in Los Angeles. Erica also told us that some of the people she started helping in 2020 just got their unemployment claims approved this year. Now California is not alone. When the COVID pandemic hit, agencies, social service agencies across the nation that administer unemployment claims, food stamps, free school lunches, rental assistance, you name it, they were all inundated. Congress acted quickly to make sure funding was made available.

But getting that money and services into the hands of Americans was a different story. Jennifer Pahlka talks about that in her new book. It’s called “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better.”

Pahlka also served as deputy chief technology officer in the Obama White House. And she’s founder and former executive director of Code for America. It’s a nonprofit that tries to help government agencies modernize their use of technology. Jennifer Pahlka, welcome to On Point.

JENNIFER PAHLKA: Thank you so much for having me.

CHAKRABARTI: So you were actually tapped by the state of California to co-chair a task force looking to improve or fix the very problems that Erica talked about in California’s unemployment system. So, first of all, what were the goals of that task force?

PAHLKA: Well, I believe that Governor Newsom asked us to look at the user experience of applying, which is what Erica was just speaking to. But I think we decided as a task force that the experience that users wanted was to get their checks, to get the actual benefit. So we decided that the most important thing would be to clear what was clearly a mounting backlog of claims. I think the day before I started, there was a hearing in the state legislature in which the director of EDD testified that she thought there were 239,000 cases that they hadn’t been able to look at yet.

But the members of the state legislature were so overwhelmed with calls from their constituents. I mean, their entire offices were just swamped with these calls from people who hadn’t gotten their checks, that they were pretty sure it was more than that. And we set out to figure out, first of all, how many there were. Thought it would take us a couple of days. It actually took seven whole weeks for us to figure that out. And by then, the number was about 1.2 or 1.3 million claims that hadn’t been looked at yet.

CHAKRABARTI: In the backlog. Okay. Wow. So what explains even just that, that delta in that one data point?

PAHLKA: Well, they weren’t counting. It was very, very hard for them to count these. The reason it took us seven weeks is that it’s just very hard. And the reason it’s hard is that while people tend to think that there’s a computer system on the back end, the system that makes the claim, it’s not really a system. There are layers of systems that date back to the 1980s, and they’ve sort of been, you know, accrued one over the other, over the decades. Some people have called them layers of paint, because it’s true that if you paint on something too many times, those layers of paint will start to crack. And that’s what’s happening.

But I started to think of them as sort of archeological layers of technology. But I realized in working with these amazing public servants, trying to get this to work, that those archeological layers are really an expression of archeological layers of policy. And the complexity of that policy is almost hard to explain, except through a little anecdote that I heard through one of my colleagues who was working next to these claims processors every day.

There was one who kept answering her questions by saying, “Well, I’m not quite sure about that. I’m the new guy. I’m still learning how this all works.” And after about the 10th time he said that, she said, “Well, how long have you been here?” And he said, “I’ve been here only 17 years.” The folks who really know how this works have been here 25 years or more.

So if it’s that complicated to know how to process a claim, of course, these systems are not going to scale up in times of need, because you can’t have more claims processors all of a sudden. And I think people don’t recognize, not only how fragile the layers of technology are behind that application, but also how sort of mind bogglingly complex the policy, and regulations and processes that have been, you know, added since the 1935 Social Security Act have become.

CHAKRABARTI: You know, what you’re describing in those archeological layers of policy and technology, in fact, I have to say it sounds familiar to me. Because we did a show a while ago about customer service in general in the private sector. And why it just, every time you get on the phone with anybody, pick on the airlines again for a second, it’s an awful experience.

And we had private sector consultants say, “Well, the problem is, their systems are built upon older systems, which are built upon older systems, which are built upon older systems.” And so, there’s never been sort of a top-down review of how to make customer service a better experience. Or an adequate top-down review. And a more efficient technological process. So, I mean, it’s not unfamiliar in the private sector either. What makes it even more painful in the government and policy world?

PAHLKA: Yeah, no, it’s absolutely true. And I think people will call out COBOL as the problem.

CHAKRABARTI: (LAUGHS) Sorry.

PAHLKA: COBOL is this programing language. That’s okay. You know, it was started, I think, in 1959. So it sounds really, really old. And the truth is there’s COBOL in California’s unemployment system, there’s COBOL in every unemployment system, but there’s also COBOL behind, you know, making a reservation with an airline.

Those layers are not necessarily the problem if they connect well, and if they are, the policies that govern them are so complex that it takes 25 years to learn. I think my big lesson from all of this was, “Wow, we can’t just work on the technology. We’ve also got to simplify the processes, and the rules and the regulations, and that’s not in the hands of the customer service agents.”

Part II

CHAKRABARTI: Jennifer Pahlka joins us today. She’s author of the new book “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better.” Jennifer, I’m sorry we had to take a quick break there, but I wanted to let you finish your thought about how fixing the technology, when it comes to implementing policies like unemployment or food assistance, isn’t enough. That you have to go further upstream, and go ahead and finish that thought.

PAHLKA: I just don’t think that we can actually fix the technology with the current complexity of policy and process that it’s supposed to fulfill. And that just means that our elected leaders, and the people who create all that policy that comes down from the top, are going to have to look a little in the mirror and say, wait, instead of yelling at the bureaucracy and saying, “Why aren’t you able to deliver these services?” I think they need to ask themselves, “What are we doing to make it harder for the bureaucracy to deliver? And what could we do to make it easier?”

And I have a very long list if anybody wants to ask me what those things might be. But they start with, don’t just keep layering on mandates. You have to remove some of the mandates so that these people can get their jobs done.

CHAKRABARTI: Okay. So I want to hear actually more about how, let’s say, that layering on of complexity in original legislation, it sounds like, how that’s manifested in the end users’ experience, or the individual Americans’ experience. I mean, you know, you’ve talked about before SNAP assistance in California, so food assistance. And again, we don’t mean to pick on California, but it’s really a terrific example, because it’s a huge state. How does that complexity end up landing before the person who’s applying for assistance?

PAHLKA: So in California, back in 2013, when we started working on SNAP access and enrollment, there was one of the forms the Californians use to apply online had about 212 questions in it. It didn’t work on a mobile phone. And so a lot of people trying to apply for SNAP just never got through that original application. You know, if you went to a library computer, it times out after half an hour, and it’s pretty hard to get through it in half an hour. So you’d have to go start all over again.

And, you know, I think a lot of people look at something like that and they say, “Oh, there must have been bureaucrats who really didn’t want people to get through. This must be an intentional design choice.” But it isn’t, always. I’m not saying there isn’t that sometimes.

But a colleague of mine who was working on this, a guy named Jake Solomon, one day he was asked to come present to the consortium that had created this application with the 212 questions. He had been quite critical of it, and it was sort of surprising that he had been asked to come. Because he had created this alternative application that was on a mobile phone, and you could do it in about 7 minutes. And it was much clearer, and easier in much the way Erica was talking about. Like the questions made sense.

And when he was at the meeting of these 23 counties, that all contributed to using that form, and managing it and paying for it. What he found is the way they made decisions about what went into that application form was that the 23 counties got to vote, which makes sense, obviously. I mean, they’re the stakeholders in it. But what that meant was that there were 23 different entities that all represented sort of what the government needed. But not 23 people who were representing what was good for users.

And I think that … we tend to think that when something is really hard to use, that there’s, you know, a bureaucrat behind the scenes who had all this power and, you know, decided to make it hard. But what we saw was that really nobody has very much power. Everyone has a little bit to say, “We need this in, and we need that in.” And, you know, that’s how you get to 212 questions. But nobody has the power to say, “Wait a minute, this is going to be really hard to use if everybody gets to pile on their requirements, but nobody gets to pull back and say, “Now it’s sort of, you know, what I call a concrete boat. It’s so big and heavy it won’t float.”

So we kind of need to recognize that, in fact, it’s not … concentrated power. It’s really diffuse power in the bureaucracy that can end up with these things that feel incredibly burdensome and overbearing and often sort of insulting.

CHAKRABARTI: Okay. So the diffuse power then also leads to lots of different stakeholders throwing in, you know, their desires, regardless of whether or not it’s actually good. As you say, you’re saying, for the end user. And it also does this thing where no one in particular is responsible. Right? So, accountability is more challenging.

But if I could just take a moment to see it from the other perspective. The reason why this problem, I think, accrues more rapidly in government is that, take those counties that you’re talking about. You know, they’re not just like individual executives, as you said. They actually also represent the other residents, and citizens and probably most importantly, the taxpayers in those counties who may or may not be one of the users receiving the assistance.

And I point that out because, you know, we were looking at some of those 212 questions for the SNAP benefits. Right? And they included things like, “Do you own a burial plot?” I guess I’m not quite sure what that’s supposed to say about current ability to pay for food. But then they have questions like, “Have you or any member of your household ever been found guilty of trading SNAP benefits for drugs or guns, etc.?”

Now, you could make an argument that from the perspective of whatever county put that question in, that it’s actually quite important for them to know on behalf of the taxpayers in that county, if the SNAP benefits are being used for their intended purpose. Right? So there’s more than just the end user in mind that government has to keep in their minds.

PAHLKA: I mean, you have to remember that question derives from some regulation. Someone put in somewhere. That’s a fair regulation. I’m not debating it, but is asking the question on the form the right way to ensure that? Are they going to answer that honestly? And which of these things are really important to ask in, you know, right up front? So that burial plot question is a good example. I tell a story in my book of actually, you know, I spent years sort of using that as a negative example. Why do you need to know that?

And then I was working with somebody in federal government who I was trying to convince that we would do something much simpler, and clearer and with less clutter and fewer questions. And I used this example to explain to him, like, this is why it’s bad when we ask too many questions. And it turns out he was the one who had written the regulation in the first place. So it was a very awkward conversation.

But because I was sitting right in front of him and I had direct access to the author of that regulation, I could ask him why. And it was really simple. He was not a bad person. He was just trying to be very thorough. When I said, “Why did you include that?” Which has now resulted in this being asked on these applications, which are now, you know, 212 questions long. He said, “Well, Congress asked us to make sure we assess their assets, a burial plot is an asset.”

And it was very clear to me that he didn’t mean harm to the users. He had just been given an assignment by Congress and he wanted to be as thorough as he could possibly be. And I just want to always recognize that good intention, but also help steer it in a different direction. And, you know, I think he’s somebody who since told me he recognizes now that maybe that wasn’t the right way to write that, because it results in these sort of burdensome systems and these low enrollment rates.

CHAKRABARTI: Yeah. You know, it makes applying for SNAP benefits sound even worse than applying for a new mortgage, which I’ve always considered to be like the financial version of a colonoscopy. (LAUGHS) Right?

PAHLKA: I did it recently. It wasn’t so bad.

CHAKRABARTI: Oh, really? (LAUGHS) Maybe it was just my financial assets that needed lots and lots of extra scrutiny. But I guess my point is, is that you’re saying there’s no need for lots of key government services that people rely on to actually be that complicated. We could successfully and in fact, improve implementation. But, you know, I’m trying to think about the policy-making process, which is really what you’re critiquing here. In the case of these SNAP benefits, what would you have changed, and where, in the process?

PAHLKA: Well, I’m critiquing both the policy-making process. And an example I would give you outside of SNAP, back to our unemployment insurance backlog, was that right in the middle of this, you know, really incredible crisis, you know, it was very clear to me that the Employment Development Department was trying very hard to clear the backlog and they weren’t able to. They didn’t have the right competencies and capacities to do it. But right in the middle of that, we had members of the legislature putting forth new bills for new mandates, and they were things like language access, which I think is very, very important.

We need to make these things available in the languages that the people who are going to use them speak, but that mandate was already on the books. They were already out of compliance. So it was just piling on more without recognizing that you couldn’t take the current application form for unemployment insurance and just translate it into other languages. It wasn’t going to be any good, as you heard Erica say. Like most people who read and write English very well couldn’t figure it out.

So why throw this mandate on, layered on all of the other mandates, instead of saying, “Hey, EDD, understand, your top priority now is clearing the backlog, we’re going to back off for a little while.” But I also want to clarify that I’m also critiquing the interpretation of these laws and policies. And I have a lot of stories in the book of public servants who pull back and say, “I see what this says, this regulation that’s been handed down to us, and we can interpret it in a very, very literal and rigid way, in which case it may create a lot of burden for the users.”

Or we can say, “What’s the best way to interpret this that actually gets the best outcome for the user?” And I think that that decision that people in the bureaucracy make every day, they just need support to choose the latter, and not the former.

CHAKRABARTI: Okay. So what you just said, Jennifer, gets us really to the core question of sort of work culture within these state and federal agencies. It sounds like you’re saying that whether the overt or sort of just cultural encouragement is to make those literal interpretations. Is that what you’re saying?

PAHLKA: I think, yes. The reason that public servants often make that overly literal and overly rigid interpretation of a law policy which ends up in what I call, and I hope this is okay to say on air, ‘policy vomit,’ where you just take the actual regulation and throw it right into the form. The way that the burial plot just got thrown right into the form, didn’t need to be, but it did. Because it was just sort of like, “Okay, that’s what you do. You just take it from the regulation and put it into the form that people are going to use.”

The reason people sometimes don’t exert their judgment and just say, “That’s not needed, we’ll figure that out some other way.” Is this hierarchy and sort of power structure within government that says the policy people are important and they’re at the top. And the people who do the implementation, that’s just a detail and they’re separated in a lot of ways, organizationally, temporally, spatially from the people who are making the policy.

And so they don’t have the ability often or the affordance to go back up to policymakers and say, “Can we interpret this a little loosely so that we can get the outcome that I think you intended?” And, you know, if you think about most organizations are hierarchies. Companies are hierarchies. But look at, say, sort of your classic tech company, and I’m not lionizing tech companies here. I have many, many criticisms of them.

But implementation, the actual product is at the top of the power structure in those organizations. They’re started by people often who are, you know, programmers, and designers and what the actual app looks like, and how it works for people is the key consideration. And the people who do sort of compliance work tend to be at lower levels with less power.

They still, of course, have their place, but they’re supporting the user experience. That’s the top, top consideration in these organizations. And it’s flipped in government. Where the actual experience that the user has is so far down the chain, so far downstream of the important work of policymakers that it’s very, very hard to get it right. And I think it’s transformational when you kind of break that, and allow the people at the bottom to be talking to the people at the top.

CHAKRABARTI: Yeah, you know, it suddenly occurred to me that maybe a somewhat accurate metaphor might be in architecture, right? Like an architect can design as crazy looking a building as she wants. But at some point in time, she’s going to have to talk to engineers and say, “Will this thing stand?” Right?

So, you know, it comes earlier in the system other than here’s the completed, you know, design and drawings and blueprints. And now you guys figure out if this building, you know, won’t collapse on us. But I think, you know, the inversion that we see in government, the problem there is politics.

And that continues to, I think, be the singular difference between non-governmental examples we can think of. And those are coming from state and federal agencies. Because the people who are creating the policy, as you said, their accountability feedback mechanism is votes, right? Whereas you could make the argument that an executive in a different kind of company, their accountability feedback mechanism is indeed user experience. So there’s a major disconnect in government. How could we get some kind of accountability feedback mechanism that I don’t know, leapfrogs politics? So that it matters more to the people making the policy, whether or not that policy effectively reaches the Americans it’s supposed to serve?

PAHLKA: I mean, I have a couple of answers to that. And the obvious one is just leadership. So leadership is realizing and connecting the dots and saying, “My constituents actually do care about getting their unemployment insurance benefits. They do care about the outcomes that we’ve sort of said we’re going for here. And even though they’re not asking me about implementation, and they don’t seem to be voting for me based on my ability to get the bureaucracy to effectively implement, I’m going to have to do that anyway.”

But I do think it does come back to, you know, how do we, the people, hold our elected leaders accountable? I mean, have you ever been asked by a candidate for office, for your vote or for your money and said, “Yes, but I want to know what you’re doing to have a healthy civil service in this jurisdiction. I want to know what you’re doing about hiring. I want to know how you’re following up on the implementation of the policies and laws that you’ve just promised you’re going to pass.” We have to change how we relate to politicians.

CHAKRABARTI: Well, Jennifer Pahlka, stand by here for just a second, because when we come back, we’re going to talk about some examples on how to make this system work better, both culturally in government and technologically.

Part III

CHAKRABARTI: Today, we’re talking with Jennifer Pahlka. She’s author of the new book “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better.” Now, Jennifer, I want to sort of dig deeply into another example that emerges from your book, and that has to do with Medicare. Because in 2015, Congress passed MACRA, or the Medicare Access and CHIP Reauthorization Act. It was the largest modern change in how Medicare pays doctors.

So the idea was to move from fee-for-service payments where doctors and hospitals get paid for each test or procedure they do, to what’s considered value-based payments. And that’s a system where doctors and hospitals get paid based on outcomes. Essentially, they get paid more if they’re keeping patients healthy.

So this new system required a new way to get data to the Centers for Medicare and Medicaid Services, or CMS. So that CMS could determine the quality of care that doctors and hospitals were providing. Well, the United States Digital Service, which is a federal agency that provides I.T. consultation services to other agencies, was tapped to help CMS with building a new website that would receive the data. Natalie Kates was part of that team.

NATALIE KATES: When we first got to CMS, we saw a pretty typical government software project. And what I mean by typical there is there were a lot of people, a lot of contractors, a large team writing the policy and the technical requirements in tandem. And a lot of project management. Not so much thought into how doctors were actually going to interact with this program, how the program would be launched, how it all worked together, and sort of how it would benefit the people it was trying to serve.

CHAKRABARTI: So in order to answer that question, Kates and the team wanted to figure out how to explain to a doctor, in a simple and straightforward way, how the new program was going to work. But their primary resource was a 2,000-page long Word document. That document was also constantly changing. So the goal was to boil down the program to a one-page website for doctors that would sum up, in plain language, what they should expect.

KATES: It should have been a very simple ask. As many technologists know, putting a static page on the Internet is not that hard. There’s one page of content. It should have taken a couple of people a few hours to write, got cleared through the clearance process at CMS and we should have been able to put it on the Internet.

Instead, it took me ten months to get that single page website out. Because it turns out when you explain a 2,000-page complicated policy in a single plain language website, it highlights how much disconnect and questions there are about that policy that have not been answered with that 2,000-page document.

CHAKRABARTI: Okay. So under the policy, a doctor could participate in the new Medicare program as an individual, or as part of a group. But the issue was when it came to being part of a group, there were nine different definitions that doctors could fall into. And Kates thought that was way too many for doctors to have to choose from. It would make the software more difficult to build and create a lot more hassle for administrators to process.

KATES: I probably had 100 hour-long calls with different parts of the team, the team as a whole, to try to get those groups whittled down to two. It does not seem like something as obvious as making a definition of a group make sense to a doctor who is going to have to choose and get paid based off of that distinction. Should be hard. But it’s hard because the team is incentivized to be technically right, not reasonably right.

CHAKRABARTI: Well, Kates and the team were eventually able to get the number of definitions down from nine to two, but it was a struggle.

KATES: In order for government to get better at delivering policy, the people who understand the policy goals, and the people who understand the technology have to be in the same room from the beginning to work through what is feasible, what technology is helpful with, and what technology isn’t helpful with.

KATES: Well, that was Natalie Kates, formerly of the United States Digital Service. On January 1, 2018, the new Medicare site launched on schedule without any issues, and users reported it was easy to use and understand. So, Jennifer, I love this example. Because ultimately the right thing did happen. But what could have been changed in the process so that it wasn’t, you know, as difficult and painful or as much of a struggle as Natalie describes? And I ask that because you talk about in the book the need for product managers, not just project managers, when it comes to implementing government programs. Tell us a little bit about that.

PAHLKA: Yeah, I think it’s a great example, too. And it was lovely to hear Natalie talk about it. It was very inspiring for me when I talked to her for writing the book. Product management is a thing that is very well-known in the consumer tech world, and very easily confused with project management in government. So when I talked to government people about product management, they usually say, “Oh yes, we have good project managers.”

And I have to make a point of making the distinction. And the distinction, and my words are the words that sort of borrowed from colleagues, is project management, which is incredibly important, and we need great project managers, is the art of getting things done. But product management is the art of deciding what to do in the first place. And when you have a project like the one that Natalie and others were working on, it seems it’s really hard to do that when there’s sort of this assumption that you have to, for instance, code for all nine definitions of the group.

And when you say, actually, we’re going to need to boil those down to one. And I’m glad they got to two, but one would have been better. It’s really hard because the people say, “Well, that’s not your lane, that’s not your call.” And people like Natalie and her colleagues at CMS, it wasn’t just the USDS folks. It was also the CMS folks. Had to stand up and say, “We need to continue to have a conversation about that. We’re going to need to really talk it through with you. And yes, we’re tech people having a discussion about policy, but that’s how this is going to get to an outcome that we all want.”

And it does mean that exactly as Natalie said, you have to have the tech people in the room for the discussion, long before they are normally invited. Because by the time you already have those nine definitions and you’re just coding for them, you’ve built a concrete boat.

CHAKRABARTI: Uh huh. I mean, you’re saying that those tech people, or those product managers, they need to be in the room from the start when the policies are, you know, at the natal stage and being created?

PAHLKA: Often, yes. And let me give you an example outside of Medicare, though, there’s many more, you know, great examples of it, and that team that did MACRA. But I talk in the book about a process in California where many states have decriminalized marijuana. And in doing that, they have said if you have a criminal record from marijuana, we should be able to take it off. It’s very, very hard to live under the burden of a felony record. The thing you’ve done a crime for is no longer a crime.

So let’s remove that from your record, so that you can get a job, etc. And, you know, it turns out that that process is incredibly complicated. It’s much worse, actually, than applying for SNAP or some of the other things we’ve talked about. There’s many, many steps. You have to go to court, and file a whole bunch of paperwork and fill out a lot of confusing forms, and so no one was really doing it. So we had this huge gap between the idea of the law, and the actual real-world outcomes that the law had intended, and that is a gap of implementation.

And so a team at Code for America had started working with the San Francisco DA’s office to sort of say, “Look, you know, none of this paperwork process is necessary. That felony record is just a field in a database. We can find all those records that are eligible and change the field and the database, essentially.” Now, much like Natalie getting from nine to two, but not all the way to one, they didn’t quite get all the way there. But they did realize that they didn’t need people to petition for this change on their record at all. They could just work directly with the bureaucrats on the back end to change it.

But one of the things they found later was that in previous expungement laws, they had been written in such a way that you couldn’t actually go in and just query the database and say, “Who are all the people who are eligible for expungement under this particular proposition in California?” If you can’t actually search on those folks, then you can’t do it automatically. And if you can’t do it automatically, really no one’s going to end up with the expungement.

And I’ll give you a little example. In this case, the law was written to say that if you had been convicted of a burglary that was under $950, and I think it had to be commercial or residential, I can’t remember which. That was what made you eligible. But that exists, if at all, that information exists in like paper forms that, you know, policemen, police officers have scribbled in hard-to-read handwriting.

And even if you could read that handwriting, you’re not going to necessarily know, you know, was this, you know, Cannon Sure Shot 10 worth more than $950 in 2011? And so, the solution ends up being, when you’re writing that law, have somebody who understands the data structures at the table, and then you can make a decision about whether you’re going to be able to make it automatable.

CHAKRABARTI: I see. Because, you know, it was occurring to me that of the folks writing the law they have the, ‘you don’t know what you don’t know problem.’ Right? They wouldn’t even know.

PAHLKA: Exactly.

CHAKRABARTI: Yeah. yeah. And so you get around that by having people who understand the data structure right there with you as you’re trying to create the policy.

PAHLKA: And they are, I would say, almost never invited to the table at that stage. But it is often incredibly helpful to have them there, not just in the cases of data structure, but, you know, “How should we actually implement this law? What technology issues or data issues are going to come up?” And it’s happening more and more now.

CHAKRABARTI: Well, you know, you have this remarkable fact that the federal workforce between 2015 and 2020 was about 2 million employees strong. But out of those 2 million, only seven people, seven out of 2 million had product manager —

PAHLKA: (CHUCKLES)

CHAKRABARTI: Or product management in their job title. So it sounds like —

PAHLKA: It was the federal workforce.

CHAKRABARTI: Federal workforce. Right. Exactly. So, I mean, it sounds like, again, we come back to this issue of culture that it’s not even something that has been recognized as a need in within agencies.

PAHLKA: I think that’s right. And I think it also derives from this sense that the further down the hierarchy you go, the less decision-making authority you should have, the less judgment you should use. A product manager is someone who’s empowered to make decisions to interpret law, policy and regulation in the service of users. And that can make people pretty uncomfortable.

But I think they don’t realize that they are also quite uncomfortable when they pass a law or a policy, and then the outcome is exactly the opposite of what they intended, which, for example, was about to be what would happen with, you know, before Natalie and the team at CMS really, you know, took charge of the implementation of MACRA.

You had the situation where doctors were threatening to leave the program because they were so frustrated, which would have degraded the quality of the care, not improved it. So I think people with power in government have to choose, are you okay getting the wrong outcome? Are you going to be okay empowering people to use their judgment a little bit lower down in the hierarchy?

CHAKRABARTI: Well, there’s another aspect of how work culture in government could positively change, and it has to do with team autonomy. Right? And so Michael Nugent, he’s someone who worked in government for about 25 years, first in the Department of Education, then in the Department of Defense as the director of Defense Language and National Security Education Office. They focused on foreign language, culture and regional expertise for DOD personnel. So, an important job. And he says his team had a tiny budget, but it had a lot of autonomy.

MICHAEL NUGENT: As more and more layers of people came in to scrutinize what you’re doing, it slows things down. So, for example, a pilot who’s flying a large plane has the authority, go or no go, based on what his or her decisions are on what’s going on with the aircraft. He or she does not have to go out and say, “Oh, gee, I went through this checklist and now it needs to be looked at [by] 15 other people, and then it needs to be sent somewhere else before I could pull the plane out of the gate.”

That’s kind of a similar analogy to having autonomy in an office. It needs to innovate. Trust the leader of that office to make decisions and hold them accountable for that. But once you start putting on layers and layers and layers of people, you have every single time that goes through something, it goes to a different interpretation, the more likelihood there is for delays in innovation.

CHAKRABARTI: Jennifer, could you take a quick minute to talk about how we’ve seen some successful innovation in government. I mean you talk about how eventually we did eventually have some fast shipment of COVID rapid tests for COVID, for example. Just take a quick minute to describe that.

PAHLKA: Yeah, I think a lot of people really liked using COVIDtest.gov. That came out, I think, right in the beginning of the Biden administration. I used it, I timed myself. It took me 11 seconds to order my tests. They came just a couple of days later. That’s really delightful service delivery. And of course, they did that in just a couple of weeks. And in fact, Natalie Kates, who we heard from earlier, was on that team. And, you know, it’s easy to say, ‘Well, that was a simple thing that, you know, wasn’t as complicated as, you know, getting doctors in a value-based care program or even clearing criminal records.”

But the truth is, they could have made it much more complicated. Think of all the things they could have asked. Your vaccination status, your health insurance information, the size of the household you live in, the age of your kids, and they decided not to. And I think that was, what I want to highlight about that is, they made an active decision to say, “What’s important here?”

Scalability, ease of use, accessibility. It launched in more than one language, which is really important to the country. And they said, if we’re going to make it those things, you know, it’s going to have to handle a very high volume of use and it’s going have to be accessible to more people. And we’re going to have to not do those other things. And they made the right choice. And that is great product management.

CHAKRABARTI: You know, even though we’ve been using a lot of technical language this hour, it seems to me that, you know, fundamentally a democratic system of governance is supposed to work for people, right? And when too many people feel like the government isn’t working for them, they seek other forms of leadership. Right?

I think it’s one of the things that they do when democratic governments, small d, fail. I mean, it’s one of the things that helps usher in authoritarianism. So this does matter to all of us. Jennifer Pahlka, the book is “Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better.” Terrific book, and thank you so much for joining us.

PAHLKA: Thank you, Meghna. This was lovely.

This article was originally published on WBUR.org.

Copyright 2023 NPR. To see more, visit https://www.npr.org.

KOSU is nonprofit and independent. We rely on readers like you to support the local, national, and international coverage on this website. Your support makes this news available to everyone.

Give today. A monthly donation of $5 makes a real difference.