Earlier this year, Elon Musk arrived in Washington, touting the promise of his so-called Department of Government Efficiency (DOGE). It was supposed to be a moonshot, a bold experiment to replace the bloat of bureaucracy with the soft, frictionless hum of artificial intelligence. Everything would be cheaper, faster, and more productive. The government would actually work for the people, and save taxpayers at least $2 trillion. 

It didn’t turn out as planned. In just a few short months, Musk and his minions arguably violated the Constitution, touted billions in savings that didn’t exist, created a potential cybersecurity nightmare, and wasted, according to a July report by Democrats on the Senate’s Permanent Subcommittee on Investigations, an estimated $21 billion. And that doesn’t include the human toll: the thousands of public servants who were fired. Not to mention the ripple effects of the programs they guillotined, including lifesaving food aid for millions around the world. 

By the time the Tesla CEO left Washington in May, DOGE had proven to be an effective smashing machine, but little else.

For Jennifer Pahlka, the results were disappointing. More than a decade ago, she helped found the U.S. Digital Service, a small team within the federal government designed to improve its digital tools and services — from healthcare to veterans’ benefits. It was the very group that President Donald Trump’s administration reconfigured as DOGE. Pahlka also created Code for America — a nonprofit devoted to bringing new tech to government. She has spent her career trying to make Washington work better for Americans — not just in terms of passing laws, but actually implementing them in ways that improve people’s lives. 

Pahlka’s seen firsthand how overly complex systems, outdated requirements, rigid interpretations of the law, and a work culture that prioritizes doing what you’re told and protecting your job over doing what’s right can cripple even the most well-intentioned programs. Yet she’s also seen what happens when technologists and civil servants work together, when the people who build the tools understand who those tools are meant to serve — and have the freedom to make them better. As she put it in her acclaimed book Recoding America (2023): “When systems or organizations don’t work the way you think they should, it is generally not because the people in them are stupid or evil. It is because they are operating according to structures and incentives that aren’t obvious from the outside.”

Big Think spoke to Pahlka about the future of DOGE and how leaders — from President Trump to New York Mayor-elect Zohran Mamdani — can truly modernize government. Spoiler: She’s optimistic.

The following interview has been lightly edited for length and clarity.

Big Think: In Recoding America, you write that we need to update the operating model of government. What do you mean by that? 

Jennifer Pahlka: People care about the government not for its own sake, but because it’s supposed to provide key things to our society, whether that’s a social safety net or our national defense or education. 

If you want a government that can actually achieve its policy goals, it has to have the right people, which means we probably need civil service reform. Those people need to be focused on the right work, which means we need significant procedural reform. We have created this world in which public servants are spending their time on anything and everything but what the public expects them to be spending their time on. That means taming those procedures that have accumulated and accreted over decades. 

We need the government to have purpose-fit systems, not just the mess of technology that we’ve accidentally acquired today. That means reforms in the way that we’re going to build and buy that technology and maintain and keep it up-to-date over time.

Lastly, government needs to be able to operate in test-and-learn frameworks. Public servants who mean well are often stuck executing a plan that was conceived of years or even sometimes decades ago, even as they can see that it is not working. That’s what we call the waterfall method. You sort of establish everything up front and then you just go do it rather than learning along the way and adjusting. 

These people, civil servants, are caught in a system that we, the people, created. It’s totally unfair to judge them for their behavior when they are stuck in something that everybody in this country is responsible for helping shape. 

Big Think: What kind of feedback did you get from people inside government about your book? 

Pahlka: People in government told me these exact words over and over again: “I feel seen.” They felt like we’re finally naming the problem accurately. And that is the key to starting to be able to work on the solution. A lot of people working in government tell me they had family members or people close to them read it. And that they finally felt like they could explain their jobs to people close to them. I’ve had a lot of leaders in government assign it to their whole teams. I’ve had people who are running for office say that it’ll be their Bible.

Big Think: Can you name anybody? What about Elon Musk? 

Pahlka: I know many people who told me they were going to get it in front of him. I do not think he read it. I do know that many people who came into DOGE did read it because some of them have told me that. And I know that a lot of them took what’s in it seriously. 

Big Think: What do you make of how DOGE turned out? 

Pahlka: What I’ve heard is that Elon came in with a real sense of urgency. He wanted to do stuff very, very quickly, and he had no confidence in Congress to take any action. They had talked about a more fundamental transformation of government that involved looking at the rules and regulations that have accumulated over time and made it hard to do the work. But they came in and just cut the workforce without cutting the work that is assigned to them. They ate dessert first. It’s harder to do the decluttering of the policy and regulatory environment than it is to fire people. 

The Elon Musk moment is over. But we’re really just less than a year into the Trump administration. You’ve still got people from this team in the White House, some of whom are sticking around and going to try to play the long game. And I think that’s where the interesting stuff will happen. We need to go to Congress and have our civil service system actually renovated for the modern era.

Big Think: If you were to write Recoding America today, what would you add or change? 

Pahlka: Since the book came out, I’ve become a huge proponent of the use of AI in government for a number of reasons. One is simply that AI is going to change society around us, and it’s the government’s job to keep up. Another is AI is a tool to tackle that procedural and regulatory bloat that we didn’t have before — and it has come at exactly the right time. 

During the pandemic, every state accrued a pretty tragic backlog of unemployment insurance claims. It was a real crisis. But there were more than 7,000 pages of active federal rules and regulations related to these claims. This program dates back to the 1935 Social Security Act, and we’ve added and added and added, but we’ve very rarely subtracted. If you wanted to reduce those pages into something that was the right size to make a scalable, robust program, you kind of can’t do that without the use of large language models.

Big Think: Should we be feeling hopeful or cynical at this moment in terms of our ability to modernize government?

Pahlka: I think it’s always a good idea to choose hope. We just launched the Recoding America Fund, an organization to fund the ecosystem that helps governments update their operating models. 

Remember, we don’t just have a federal government — we have states and counties and cities. What I think DOGE did show is that people want transformational change. I don’t think they wanted the dessert first, but they did want something that was bolder than what they’ve been given in the past. We have plenty of opportunities to give that to them. 

There were states already doing this kind of work more quietly, and I think probably more successfully than DOGE in the long run. Colorado is a great example. The Colorado Digital Service is slowly and quietly transforming how the state does delivery. Let’s see over the next couple of years if Coloradans really feel that difference.

At the federal government level, though, change hurts, and it’s hard. I don’t want to condone any actions that DOGE may have taken that were truly illegal, but I do think that sometimes you need to clear things out in order to build back what is fit for purpose today. I don’t think that work is done. I think so many people are concerned that the work now is building back when, in fact, I think there is still more clearing that needs to happen, and I’d love to see it happen more thoughtfully.

Big Think: What are some low-hanging fruit to, as you put it on your Substack, Marie Kondo the government? 

Pahlka: In the city of San Francisco, where the municipal code runs 16 million words, we’ve used LLMs to figure out a couple of key questions: How many boards and commissions did they have? And how many required reports were written into the code? This is important. You’re talking about a city that has estimated it would cost $1.7 million to build a public toilet. There are huge processes involved in getting anything approved, so being able to understand even just how many boards and commissions we need is really critical. 

Over in Denver, Mayor Mike Johnston was facing a budget shortfall. Instead of just saying, “Okay, we’re going to have to do layoffs. It’ll happen by seniority,” he said, “What’s most important is that we protect the level of services that Denverites get. Therefore, I’m going to really look at where we’ve got too many people, where those jobs could be automated. Where we don’t have enough people, we can move people to the jobs where the positions are open. But also, if we do have to lose people, we will lose the underperformers, not the people we hired last week or last month.” That’s a hard political thing to do. But that’s the kind of cleanup that’s really necessary. 

Big Think: To what extent do you think the failures of implementation shape our politics today? And by that, I just mean the sense that people don’t seem to feel like anybody can get anything done?

Pahlka: Leading up to the recent presidential election, you saw polls in which people were asked how much does the system need to change, and the number of people who answered at least “major change” was, I believe, nearly 70%. That’s a lot of people who are really, really frustrated. I don’t think it’s all tied just to implementation, but I think a lot of that stems from the sense that government just doesn’t work.

I’ll give you an example from my time when I was living in Oakland. Two years ago, I finished up my Zoom calls, went into my bedroom in the middle of the day, and there was a man going through my drawers. I screamed, ran out of the house, and I dialed 911. I waited for seven minutes, and the call dropped. I waited for three minutes, and it dropped again. I called back, waited for five minutes, and only then did the 911 center answer. They said, “Oakland Police Department will be there as soon as they possibly can.” They showed up two whole days later.

We neglect government successes at our peril. I can get on a plane with a reasonable sense that it is not going to fall out of the sky because government does a great job actually making sure that those planes are really safe. We don’t ever think about how valuable that is, and we should. But we also think, “If I call 911, someone will pick up the phone.”

There’s been a lot of debate about public safety policy, but if you look at what was going on in Oakland the day that I called and didn’t get an answer, it really didn’t have that much to do with that policy. It had to do with the operating model of government. They didn’t have the right people, and they didn’t have enough of the people. It’s very hard to hire. They’re way behind in their hiring. It takes a long time to get somebody on board and trained. 

Those people are also focused on the wrong work — 85% of their job is paperwork. Part of the reason for that is that the Oakland Police Department is subject to so many different oversight bodies. Part of the reason they are spending so much time on paperwork is because those different oversight bodies ask for data and information in different ways. So you have police officers who spend their time entering the same information into multiple different systems instead of responding to calls like mine. 

Not that we shouldn’t hold our police officers and police departments truly accountable and constrain what they can do. But not to the point that they are not actually doing the job that taxpayers expect them to be doing. That’s not an issue of public safety policy. It’s an issue of the operating system of government and our neglect of it.

Big Think: In New York, Mayor-elect Mamdani has offered a vision of government that works better for people. If you were in his shoes and you wanted to make all these big changes — whether it be free buses or city-owned grocery stores — where would you start?

Pahlka: I would first assemble a team that was 100% committed to the hard work of transforming government. That is the healthy meal you have to eat in order to earn your dessert. All of these promises rely on the ability to do things at a certain speed and a certain accuracy — and to be able to learn while we’re doing them. 

One of the things Mamdani said in one of his interviews is that he wants to try things as pilots and not necessarily just plow forward. I think that’s a good principle. But what matters most for any new mayor or governor or county executive is this: Do you have the team that’s committed not to administering the policies as they exist and/or changing the policies, but to creating the machinery that can actually deliver on any new policy that you come up with? That has to come first.

I’m not saying you don’t go do a couple of important things because you need to keep people excited. But most of the time, elected leaders realize that the operating model of their jurisdiction is sorely out of date about halfway or a little more than halfway through their term. At that point it’s too late. They say, “Let’s fix this transit problem,” and it turns out that procurement to fix this problem is going to take maybe a total of six years, and they go, “Wait, that won’t be done on my watch.”

Big Think: Do you think Mamdani has read your book?

Pahlka: I would be very happy to send him a copy.