Why does government so often fall short of its goals — or even fail catastrophically? Jennifer Pahlka, in her important new book Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better, offers what is perhaps the most incisive explanation yet for government failure, particularly in the realm of technology. This is a book that every policymaker should read and take to heart.
In Pahlka’s view, declining state capacity has resulted from a political culture that prioritizes politics and policymaking over implementation. And government especially falls short of its potential for good when well-intentioned policymakers fail to understand technology, pay attention to citizens who suffer the consequences of poor delivery of government services, or emphasize outcomes over processes. She writes, “When systems or organizations don’t work as 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.”
Jennifer Pahlka comes to her granular understanding of government failures through long experience with the digital delivery of government service at the federal, state, and local levels. In 2009, she founded Code for America to attract technology experts to work on public problems. In 2013, she became the U.S. Deputy Chief Technology Officer in the Obama administration. She played a significant role in rescuing the healthcare.gov website after its botched rollout and helped to create the U.S. Digital Service. In 2020, California Governor Gavin Newsom appointed her to a task force to salvage the state’s unemployment insurance program when it collapsed under the weight of a tenfold increase in claims during the Covid pandemic.
In this podcast interview, Pahlka discusses the complexity of government computer systems that become unworkable through decades of layering-on of technologies and policies, policymakers’ failure to understand why they pass laws that can’t be implemented, and the dilemma of civil servants caught between contradictory pressures to deliver outcomes while also adhering to the rigid processes on which their jobs depend. She describes how the government is caught in a hierarchical “waterfall model” of program management while the software industry has moved on to a decentralized model of agile development, and how technological developments are doomed by unworkable technical requirements that aren’t actually mandated by government policy — even though bureaucrats and contractors have come to believe that they are. And although listeners will share Pahlka’s evident frustration at the many examples of government failure that she cites, she also shares numerous examples of courageous leaders who have overcome structural obstacles and outdated thinking to deliver results and show what government can be at its best.
Jennifer Pahlka: The main concept of agile is that you have all of the stakeholders in a cycle, so they’re able to talk to each other and learn from each other and change what they’re doing based on what they’re learning, which you just don’t see when the information and power flows only one way.
Geoff Kabaservice: Hello! I’m Geoff Kabaservice for the Niskanen Center. Welcome to the Vital Center podcast, where we try to sort through the problems of the muddled, moderate majority of Americans, drawing upon history, biography, and current events. And I’m thrilled to be joined today by Jennifer Pahlka. She served as the Deputy Chief Technology Officer of the United States in the second term of the Obama administration, during which time she came up with the idea for the United States Digital Service, which is a sort of an inside-the-government consultant for federal agencies on information technology.
She’s also the founder and former executive director of Code for America, a nonprofit organization launched in 2009 to provide technologically adept young people with fellowships that let them work with local governments to improve their delivery of services. The Washington Post called it “the technology world’s equivalent of the Peace Corps or Teach for America.” She served on the Defense Innovation Board for the Department of Defense under presidents Obama and Trump and chaired California Governor Gavin Newsom’s strike team on unemployment insurance during the pandemic. Welcome, Jen!
Jennifer Pahlka: Thanks for having me, Geoff.
Geoff Kabaservice: It’s great to have you here. And to the point of our discussion today, you are the author of a new book that will be released, I think June 3rd…
Jennifer Pahlka: 13th.
Geoff Kabaservice: …June 13th, entitled Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better. It’s a fascinating work that I hope and believe will have a major impact on how citizens and political leaders alike think about and understand why government too often fails to achieve its aims. This is a book about the dysfunctions in government and bureaucracy that somehow manages to be not just interesting but illuminating and revelatory and profound. And much of it is written with a kind of controlled outrage that I find compelling — I’m sure we can get into why. And I’m inclined to agree with some of the early reviewers who feel that politicians should be required to read it and no public servant should be hired who has not read it. So congratulations, Jen, on this real breakthrough of a book.
Jennifer Pahlka: Well, I have to sit here and absorb that praise for a second. Thank you. It’s very, very kind of you.
Geoff Kabaservice: So before we get into the book, can you tell me something about where you grew up, where you were educated, and how you came to the career path that you did?
Jennifer Pahlka: I grew up in Texas, New York, Connecticut, and then moved out to California after college — so I just hit all the big states, basically. My education undergrad was at Yale and I did American Studies. And the big joke is what do you do with an American Studies degree? You found an organization called Code for America that has nothing to do with American Studies.
Geoff Kabaservice: Interesting. I want to say that you went to the Bronx High School of Science too?
Jennifer Pahlka: I did. Thank you for mentioning that. I’m a proud graduate of the New York City public school system.
Geoff Kabaservice: So was my dad. He also was a Bronx Science grad. And speaking of that trajectory, I did a study of admissions at Yale in the early 1950s, and I happened to use Bronx Science and the Phillips Academy at Andover as case examples. And I found that over a five-year span in the first half of that decade, Andover sent 275 undergraduates to Yale — and Bronx Science, which of course was far more academically selective, sent seven. And there were various explanations for that forty-fold disparity, but one of them was that particularly in the 1950s, science and technology were looked down upon in many of the elite colleges and universities of that era. And Yale was very much of an Anglophile institution, and it had inherited that mentality that you described in your book where the British civil service traditionally divided the work of government into the intellectual and the mechanical, as they said. And it’s interesting to see that that mentality persists in the governing class to this day.
Jennifer Pahlka: Yes, I think it does. I also think it’s changing, which is great. But that is something that I felt when I was there to some degree — even though I worked in the Office of Science and Technology Policy. We were definitely the science nerds. But I did feel a sort of resistance… I think actually it was less science, especially in the Obama administration, and more amongst the sort of political class and the policy class in D.C. against really anything that had to do with implementation. I think science policy was okay. But if you were, like me, not actually working on policy to govern science and especially the growth of the tech platforms for instance, but how government delivers services — that fell in this category of the details that the White House shouldn’t really be involved with. And to be fair, there are really good reasons why the White House doesn’t want to touch implementation on a practical level. But I think the sort of pushing away of having tech expertise in these centers of government does come from this deep cultural tendency that probably goes back to the Brits.
Geoff Kabaservice: Fascinating. Can you tell me how you came by your scientific-technological-implementation expertise?
Jennifer Pahlka: Oh, I’m faking it. I don’t have any of that. I was, as you can tell… I actually did love science in high school, but I was one of those women who dropped out of the science and tech pipeline. I was very into biology, but obviously did liberal arts in college. I’m also not a programmer and I am not a designer. I am none of the things that I admire, and my friends who do technology and design will tell you I am absolutely terrible at them. I have no design sense whatsoever.
But I think that my skill is in translating between very deep technical fields and the broader culture. And I think I really developed that, to be honest, when I started working in the video game industry. I came back from traveling around the world in 1995, sort of the beginning of the dot-com boom in San Francisco. I was completely out of money and needed a job, so I took a job working for a company that did conferences for software developers. And I very soon moved into the conference for video game developers, which is a very technical field. They really push the boundaries of technology — were doing so in the mid-’90s and are still doing so. And I think I just got really good at understanding… not necessarily everything that advanced programmers were saying, but understanding where the heat was and where the conflict was and where the meaning was. And I got kind of good at faking tech.
Geoff Kabaservice: I have no technological expertise myself. In fact, the only college course that I took on computer science was a programming course in Pascal, which is a now-obsolete computing language that I believe has been superseded by C++.
Jennifer Pahlka: Yes.
Geoff Kabaservice: But can I ask about how you came to start Code for America?
Jennifer Pahlka: I was working on these Web 2.0 conferences, when Web 2.0 was a thing. It really isn’t a term we use much anymore, but it was a big deal and quite transformative back in 2007 and 2008 when we realized that probably the best application of the principles and values of the participatory web were in government. If there is an institution that is built for everyone, that is government. And that was foundational to the idea of Web 2.0.
So we started Gov 2.0 and kind of had to fight, I think, to define it as something more profound and meaningful than just, “Let’s get federal agencies on Twitter,” though I think that’s a good thing to do. But we were in my Web 2.0 world, I was putting the very early Facebook and very early Google and very early Flickr on stage. And they had these lightweight, very user-friendly ways of working with technology that of course changed the world.
And the contrast to how government was building technology — lightweight is not the word I would use. Iterative, user-centered — they’re just the absolute opposite of that. It really shocked me. So when we went to D.C. and we’re asking people, “How do you build and buy government technology that serves people?” I could not have been more horrified, in a sense, by what I was hearing. And I believed that people would want to help if they could see the difference that they could make. I happened to know enough people that were able to help me start this organization. And I couldn’t believe it, but the very first year… I thought, “No one’s going to show up. This pays this tiny stipend for a year of work.” And we had something like 525 applications for 20 slots the first year.
Geoff Kabaservice: So obviously a lot of appetite for this kind of public service, in its way.
Jennifer Pahlka: And that was 2010. There’s a narrative that nobody good in technology wants to go into government. And day after day after day, I see that that is not true.
Geoff Kabaservice: There’s another somewhat related question here, which you’ve touched upon. Women have historically been underrepresented in the STEM fields. And although there has been I think considerable improvement overall in recent years, I believe that women still make up only about a fifth of undergraduate computer science graduates each year. Do you have thoughts on this underrepresentation and did that come into play when you were running Code for America?
Jennifer Pahlka: We had fewer female programmers our first couple of years. But of course we’re not just programmers. If you’re trying to make services that work well for people, very critical skillsets are product management, user research, design. Those don’t suffer from quite as much gender imbalance as programming, and don’t get as much credit for the results of great technology as they should.
I’ll also say… You may have noticed in the book, and I didn’t try to do this, but so many of the great government technologists that I profile are women. I don’t know the statistics, but there is certainly not as much of a gender imbalance in government technology as there is in private-sector technology. You look at the three top jobs in tech in federal government right now: Clare Martorana is the federal CIO, Mina Hsiang is running the USDS, and Robin Carnahan is running the GSA, which of course has the Technology Transformation Services in it under her. She has Raylene Yung, a former colleague of mine, running the Technology Modernization Fund. So actually, I’ll count four women there. There must be something (and I won’t presume to say what it is) about government technology that attracts great female technologists.
Geoff Kabaservice: Yeah, I did notice that a disproportionate number of the heroes of your book were in fact women in those positions — as they actually were in Michael Lewis’ The Fifth Risk.
Jennifer Pahlka: Yes, Charity Dean. She’s fantastic.
Geoff Kabaservice: Absolutely. So as I said, I don’t know much about computing science and such, but I’m extremely interested in human networks and connections. So I hope you’ll indulge me while I mention a few that seem relevant to your work and your book. About 20 years ago, I worked for the Advisory Board Company in Washington, D.C., which is a kind of consulting company for hospitals and health systems. And my boss two levels up was Jeff Zients, who is now President Biden’s Chief of Staff — and earlier than that, he was President Obama’s first U.S. Chief Performance Officer and took charge of the effort to salvage the botched healthcare.gov rollout, which you were so involved with yourself. And another one of my colleagues at that same company was Aneesh Chopra, whom President Obama named as the first U.S. Chief Technology Officer, who was succeeded by Todd Park, with whom you served as Deputy Chief Technology Officer. So I’ve heard something about the world that you’re describing through some people who have known those principals.
Jennifer Pahlka: And of course, Todd and Jeff worked very closely when healthcare.gov had its problems, we’ll say. And they worked very closely with some wonderful public servants at CMS who don’t get enough credit for their role in bringing that site back from the brink.
Geoff Kabaservice: Right. There has also been an upswell of people from both the left and the right who are concerned with the issue of the United States’ declining state capacity, which is basically the ability of government at all levels to get things done, to actually execute and implement the programs that get passed. And you and I took part in a conference back in April that was co-sponsored by Columbia University’s Center on Capitalism and Society and Philip K. Howard’s Common Good Organization. And I thought it was fascinating to see at that conference how many people are thinking along similar lines about this problem of state capacity, people on both the left and the right. And some of the participants were also people who show up in your book, in one way or another, and also the Volcker Alliance, which I believe supported you as a fellow during the first year that you were writing this book.
Jennifer Pahlka: That’s right, yeah. And it’s funny… I think nowhere in the book do I use the words “state capacity,” but it is absolutely a book that’s advocating for state capacity. And I think what I’m trying to do is not just say, “We need more people,” or “We need more money,” but ask us to take a long, hard look at what kinds of skills and capabilities we need, and the kinds of jobs too. I mean, we have this problem where we don’t look at government programs and service delivery from a design perspective. We tend to just say, “Oh, we need more people.” And that is really the wrong answer.
I’d love to give an example in the context of one of the quick stories in the book. We started working at Code for America with the San Francisco DA’s office about the problem of people who had former marijuana convictions needing to clear them off their records once marijuana was no longer a crime. And they just weren’t going anywhere. You had the law saying one thing, but the databases said something entirely different. So people were not getting the relief that the law intended. And it was because the process was this Kafkaesque thing of going and finding forms from the government, filling out other forms that don’t make any sense, filing those forms back with different parts of the government, waiting, going to court, getting other forms back, and then filing them back again with the exact same people you got the documents from in the first place — which made it so that really almost no one could persist through this process. But if you pull back and you say, “None of this is necessary, and these are just fields in a database. We can find these records and change them and we don’t need any paperwork,” you can solve the problem without saying, “Oh, wow, we need like 5,000 more pro bono lawyers helping these people.” That really wouldn’t help at all.
But in fact, it’s sort of no one’s job to do that. That’s one of the things I think people don’t realize. It is not that someone fell down on the job, it’s that no one is assigned to design processes that work for people. And so when we talk about state capacity, I don’t want it to get mixed up with sort of, “Let’s just grow the government at federal, state, and local levels.” We need to grow it in the right ways and move people into jobs that are going to be valuable instead of throwing bodies at processes that were just never designed at all. I don’t mean just not designed to work, they literally were not designed.
Geoff Kabaservice: They were “vomited up,” as someone else says elsewhere.
Jennifer Pahlka: Yes. And I think, to continue on that, that the left has this narrative that we are starving government by design. And my caveat on that, though it’s sometimes true, is that we are starving government of design because we don’t think it’s part of state capacity — and it needs to be.
Geoff Kabaservice: So just to go back to the marijuana issue, it was California’s Prop 64, also known as its Adult Use of Marijuana Act, that made marijuana legal and also then made hundreds of thousands of people statewide eligible to have their felony convictions expunged. And many of them might well have wanted to go into the legal marijuana industry. But actually there were only 23 people who got as far as filing the initial paperwork to get their records expunged, because the process was so cumbersome.
Jennifer Pahlka: That was a year after passing the law. They’d only gotten 23 applications, and they expected that none of them had or would really make it through to an actual expungement.
Geoff Kabaservice: So that just seems to be a perfect illustration of the difference between passing a law and expecting change to happen, and actually executing the change necessary for that to happen.
Jennifer Pahlka: Exactly.
Geoff Kabaservice: So on this issue of state capacity, there’s a lot of people coming out of California, where you live, who are involved in this, I think because they have seen that just having Democrats in power at every level is no guarantee that you’re going to get results. And I’m thinking of people like Ezra Klein, although I think he’s moving back East, but also the people at the Modern Power Substack. And in fact, Misha Chellam, who’s a writer there, predicted that your book will be to the state capacity movement what Rachel Carson’s Silent Spring was to the environmental movement. That’s high praise!
Jennifer Pahlka: Misha’s also been very generous. And I’m not saying this because he’s been so generous to me. But I think Misha is doing some of the most important work in the country by trying to create an actual constituency in political power behind the idea of state capacity. And what we need is more Mishas for other states. And we need a Misha for the country, someone who’s looking at it on a national level. Niskanen is absolutely part of that as a think tank, but I think there’s something different that Misha’s doing in terms of trying to get real energy and power behind it.
Geoff Kabaservice: Yes, I absolutely agree. There was a Medium post that you wrote a few years ago about delivery-driven government, which is not really a phrase you use in the book, but I think it’s pretty much along the lines of state capacity. And there was a quote from that post that I wanted to bring up. You wrote: “Policymaking is in a quiet crisis. While there are meaningful differences of opinion in the United States and elsewhere about which policies will result in the kind of society we want (often driven by different visions for our society), it’s also true that many of the policies that result from our democratic processes simply do not live up to their intent. It’s like a car where the steering wheel is only loosely connected to the wheels: We might fight over who is in the driver’s seat, but regardless of who is driving, you’re not going to get where you meant to go — and you’re going to hurt people along the way.” I thought that was a perfect analogy for the situation we find ourselves to be in.
Jennifer Pahlka: And I wrote that, of course, before the pandemic. I think during the pandemic you saw lawmakers at all levels say, “Wait a minute. We’ve hit the gas on recovery funds, we’ve hit the gas on economic stimulus, and the car’s not exactly jumping into gear.” I mean, we did get a lot of stimulus out, we did get a lot of unemployment insurance out, but there were significant problems with all of those things. There were people who were missed, and of course there was an unemployment insurance, many people who waited far too long to get their payments, and then the fraud problem. And I think that’s a great time for us for leaders to reflect on this issue of state capacity, because you don’t think about it until you’re trying to turn the car and it’s not turning.
Geoff Kabaservice: Yes. You had another quote from the book that I thought was just really on point. You wrote that when systems or organizations don’t work, it’s not because the people there are stupid or evil, but rather because “they’re operating according to structures and incentives that aren’t obvious from the outside.” And behind that is often what you called “the waterfall model” of policymaking, but also maybe more generally the mentality that’s present at all levels of government as well. Can you tell me something about the waterfall model and why that matters so much?
Jennifer Pahlka: It’s a concept borrowed from software development, where there’s been a real revolution in the past 25 years away from the waterfall method, which is essentially a hierarchy. It’s a cascade from top to bottom in which you do things sequentially, and the lower tiers of the waterfall don’t get to give any information or insights back up; you just move forward. It’s a one-way track. And I hope people realize that I’m not talking just about software development, I am talking about a much larger way of thinking and acting that’s really deeply ingrained in our culture. So when you have a hierarchy from policymaking down to regulation down to implementation, I think part of the big problem that we have is that we’re not learning anything from the people who are doing the implementation.
The contrast in software development is to something called “agile development.” I’m not really sure “agile” is the right contrast to “waterfall” in the big cultural sense, but the main concept of agile is that it’s a cycle. You build, measure, learn, and you have all of the stakeholders in a cycle so they’re able to talk to each other and learn from each other and change what they’re doing based on what they’re learning — which you just don’t see when the information and power flows only one way.
Obviously this is built into every structure of our government. Federalism is a waterfall, the policy cascade is a waterfall. And it’s a very hierarchical culture, and policymakers are at the top and implementers are at the bottom. There’s no silver bullet for being able to learn and be more agile and build state capacity. And this certainly isn’t it, but fighting that waterfall and rejecting it — which is things like having different people at the table when you’re discussing policy — does start to get you away from some of the real detriments of that process and that way of thinking.
Geoff Kabaservice: You quoted from the “Manifesto for Agile Software Development,” which came out in 2001, which was rejecting the precepts of that waterfall project-management model. And it included aspects like you welcome changing requirements even late in development, and everybody takes part in this and gets to look at the software and try it out while it’s still being coded. And you put the usefulness of product over fidelity to process. And you also quoted General Stanley McChrystal, with what I thought was an extremely powerful quote: “Don’t do what I tell you to do. Do what I would do if I were in your situation and had the information that you have.”
Jennifer Pahlka: I thought it was a great way of putting it.
Geoff Kabaservice: Yeah, it was a brilliant quote. And I actually thought about that when I was reading an analysis of why Russia’s army is not having the expected results in Ukraine. It’s because they don’t have an empowered junior officer corps. Instead it’s a deeply hierarchical organization, and you cannot take independent action further down the chain unless you have explicit authorization from further up the chain. And it seems that this mentality does pervade so many aspects of the way life is lived in so many societies across the world.
Jennifer Pahlka: Yes. And what I had hoped to do in the book is also show examples of the waterfall being disrupted by public servants who say, “Okay, I hear what you’re saying you want me to do. But I know that that’s not the right way to do it. And if I do exactly what you said, regulator/ policymaker/ lawmaker/ whoever it is, I’m not going to get the outcomes you intended. And it’s more important for me to honor the intent of the law than the actual words of the law.” I tell you, this is quite controversial. I think there are many people who will absolutely disagree with this, that we should not have public servants interpreting law, policy, and regulation using their own discretion.
To that I would just say, look at the results of both sides of it. Certainly we need to have… I am not suggesting that people go do something just because it more matches their values or they don’t like the law or policy that was handed down. But I do think we need public servants to use their discretion to get the outcomes that both the public and our lawmakers intended — especially in these areas where we’re really not fighting about politics, we’re really just trying to get the job done.
Geoff Kabaservice: I was impressed that throughout your book, you do not take a political position in places where it would’ve been easy to put some kind of ideological interpretation on things. And this is also not a book with villains. The villain is not government. The villain is not even those civil servants who actually seem to be performing an obstructionist role. There was one character you brought up named Kevin (or who you gave the name of Kevin), who worked at the Veteran’s Administration and particularly was dealing with the slow performance of the Veterans Benefit Management System and the administration of getting relief to vets in need of it.
And you had a quote from him, which I’m going to paraphrase here, but he said something to you, something like, “I’ve spent my entire career training my team not to have an opinion on business requirements. If they, the higher-ups, ask us to build a concrete boat, a boat that’ll sink right off the pier as soon as it launched, we’ll build a concrete boat. That way when it goes wrong, it’s not my fault.” So that mentality seems terrible, but he’s got a point that he’s going to be judged not on outcomes — because that’s not the way that civil servants are evaluated — but on whether or not they had fidelity to process. And that accounts for a lot of the really bad results that we do get from government.
Jennifer Pahlka: Yeah, I’m glad that you don’t see any villains in the book. I mean, I felt sort of gut-punched when he said that, just because of the real-world impact of the VBMS not working and us not serving veterans the way that we should. But I do say that he had a point, and that it is important to be angry at the system, not the people. The system is what’s broken. And furthermore, vilifying public servants… I would say he’s not a great leader, but I don’t think it makes sense to vilify people like him, or the woman earlier in the book at the EDD who is also sort of disempowered herself within this waterfall — or I would say the waterfall has disempowered her. Because it’s not going to change anything. What will change is if we change the system and the ways of thinking and the environment around these folks.
That said, I think the folks in the book who do make better choices and demonstrate leadership deserve to be uplifted. And we should make all public servants want to be like Yadira Sanchez, who I profile later, as someone who says, “I’m going to interpret this to get the outcomes that I need” when she’s trying to improve Medicare at the Centers for Medicare and Medicaid Services.
Geoff Kabaservice: The leader at the EDD, whom you pseudonymed again, was Paula. And Paula was somebody who just seems really wedded to the status quo and the established ways of doing things and can’t really deviate from that even when there’s an obvious pressing need because you’re getting perverse results. So tell us something about what happened during the pandemic when all of these unemployment applications started to go up. I should add that it’s the California Employment Development Department that we’re talking about. And when the pandemic hit, it experienced a tenfold increase in claims and the system just buckled under the weight. But as you dug into it through your work for the task force that Gavin Newsom appointed you to, you found out that it was a mixture of technological problems plus administrative and mentality problems. Can you just tell us something about that episode and how it was educative for you?
Jennifer Pahlka: I think that pretty much everybody accepted the notion that the fact that there was COBOL code in unemployment insurance systems — not just in California but everywhere — must be the problem. And I don’t think that’s really borne out by what we saw. Because in fact, the part of the system that was working really well was what I’ll just call the assembly line: the automatic part where if you got verified early on in all these ways, you got a check pretty quickly. And the department deserves a lot of credit for serving a lot of people relatively quickly. And that was all the COBOL code.
But the complexity of the system is almost hard to describe. It takes chapters in the book. But the best way of describing it, in some sense, was the experience my colleague Marina had when she was talking with one of the claims processors. She was just asking these guys how this works so she could figure out where the bottlenecks are, and one of them kept saying, “I’ll have to check on that, I’m not sure. I’m the new guy.” And finally, after about 10 times, she says, “Okay, how long have you been here?” And he said, “Oh, I’ve only been here for 17 years. The folks who really know how unemployment insurance policy and process work have been here 25 years or more.” And that’s not just like they happen to have been around that long, but in fact there are a lot of rules governing who can touch what claim at what level with what number of years of experience.
And so this became very important information because the state had allowed EDD (or really ordered EDD) to hire about 5,000 new workers in the hopes that they could help process these claims. But the ones that were stuck required the high-touch, high-knowledge expertise of these people with 17 to 30 years of experience. And the folks who just got hired really couldn’t do anything. But what they did do was they took up the time and attention of that very limited number of claims processors. So every person the state hired actually slowed claims processing.
This was not something anybody wanted to hear, but it was the truth. And I do think it speaks to what we need to do if we’re going to have services that scale during times of crisis. It is not necessarily just a matter of getting rid of that COBOL code. Moving a system that complex to the cloud is still going to be a system that is going to require human intervention that simply cannot scale up when you need more people. And we really need to look at that clutter and accumulation — not just of the technology but the policy that drives it — if we want to make systems that are going to work for people.
Geoff Kabaservice: I agree with everything you’re saying. However, I did think that your description of that outdated technology was also quite fascinating. Because it’s not just that these are systems that date back to the 1950s in part and have COBOL at their center. It’s that there have been all kinds of what you call sedimentary layers that have been added on to this technology over the years, including macros and workarounds that no one has recorded. So it’s like oral lore passed down from programmer to programmer. There’s an 800-page instruction book, but even reading that thoroughly won’t get you but part of the way toward understanding how the system works. And these programmers seem almost like, I don’t know, priests serving a temperamental and possibly insane god.
Jennifer Pahlka: Well said!
Geoff Kabaservice: So there is a technological element to it.
Jennifer Pahlka: Oh, absolutely. I think that’s the problem. It’s that everyone who sent us in there conceives of the technology that runs unemployment insurance as a system that was built to do a particular thing. And if you think of it that way, you’re imagining a system that’s been built. And it’s just not that. Tony Scott had used the metaphor of layers of paint, and I think that’s helpful because you think about how at some number of layers that just starts cracking, but you never go back and scrape them off and start over again. But I really felt like, when we were looking at them, that we could date these layers back to the 1970s and ‘80s, that it was archeological layers in the sense that they’re made up of stuff that was once useful and is now garbage and has been covered by just whatever has fallen from the sky. And it’s very, very hard to deliver under those circumstances when that’s what your technology looks like.
And I think there’s a good question to ask: Why do we add a new layer instead of just go back? Well, I think that goes back to the ways in which the folks who make decisions don’t really understand and don’t care to understand the operations. And to them, why would you spend money investing in something that we already have functionality to do? So every appropriation for new technology has to add something new but not go back. Because there’s not an understanding that refactoring code and moving things to new platforms and getting rid of that fragility and all of those dependencies back so many layers is worthwhile. The concept in Silicon Valley or in programming is “technical debt.” Periodically, you have to retire your technical debt. We have just no appetite for that in government, nor do we have an appetite for retiring our policy debt.
Geoff Kabaservice: Yeah, and that’s important too because there are a lot of organizations and agencies out there that have similar problems of legacy systems. And one of the ones you mentioned is the Internal Revenue Service, which since 2000 has been planning to replace one of its core systems, the Individual Master File. And now maybe they might have that happen by 2030, which means they probably won’t. But you also point out that the IRS has real trouble modernizing its system because there’s more than 73,000 pages in the statutes and regulations that the agency has to implement. And between 2001 and 2012, you wrote, “The tax code changed 4,680 times, or an average of once per day.” And you quoted then IRS Commissioner John Koskinen testifying to Congress, who told them, “I didn’t write the tax laws, you did.” So you have to have a certain degree of sympathy for these public servants who are given a thankless and possibly impossible task of trying to implement the law as written with systems that can’t handle it because the policy has never been thought out to that level of implementation.
Jennifer Pahlka: I think every technology person or implementation person who’s ever been called up in front of a congressional committee or a state assembly committee and grilled on their failures has wanted to say that, and good for John Koskinen for saying it out loud. He’s 100% right. And I don’t think our elected leaders take responsibility for that complexity in the way that they should. Instead, when things go wrong, they throw on more mandates that make more complexity, that distract the agency from the core work of retiring policy, retiring technical debt. And they mean well, but because they don’t see the system the way that, say, my colleague Marina could see the system, their actions are more harmful than helpful. And I do think that’s probably why people are saying that public officials should read the book, because I do think they’ll act differently when they see the real guts of the machinery of government.
Geoff Kabaservice: I hope so. There’s another aspect of policy implementation that you described, analogizing it to the game of telephone where a person will whisper a message to one person and then they will whisper that message to the next person and to the next. And inevitably that message degrades with every transition, and by the end it gets to be something completely unrelated to what the original message was. And some of that seemed to be at work in the interesting episode that you describe about the U.S. Air Force awarding a contract to try to develop its Next Generation GPS Operational Control System and all the disaster that has ensued from that. And the hero there (although he doesn’t win in the end) is Matthew Weaver (better known as Weaver) of Digital Services.
Jennifer Pahlka: His real name in this case.
Geoff Kabaservice: Yes. Can you tell us something about that episode?
Jennifer Pahlka: This was actually something that Weaver had written up and I adapted because it was so fascinating the way he dove into it. But essentially he gets assigned to go help this team at Raytheon deliver on the next-generation software for satellites. For those in the know, it’s the OCX project, which has something of a reputation around defense circles and procurement circles.
So they’re stuck. This team is stuck on something, which is getting the data from the satellites back to the ground stations. It’s taking too long, and it’s really obvious why it’s taking too long. There’s a protocol that is standard used everywhere in the internet, a basic thing that does this. It’s the obvious way to do it. And it would be sort of like a screwdriver and a screw: it will work, you can go to any hardware store and get it. But instead there’s this ridiculous Rube Goldberg machine of software that’s inserted in the middle. So imagine you’ve got a screw here and a screw here, but in the middle is some crazy thing that’s turning and dropping apples. And of course it won’t get there in time! And he is like, “Why haven’t we taken this” — it’s called an Enterprise Service Bus — “out and just use UDP as the obvious solution?” And the answer is that it’s in the Air Force’s requirements, so you can’t take it out.
And so he traces that requirement from the contract with Raytheon to the Air Force RFP, to the Air Force Enterprise Architecture, which derives from the DOD enterprise architecture, which derives from the federal enterprise architecture. And where did this federal enterprise architecture come from? It came from the CIO Council required by Congress, and it was required by Congress in order to make technology better.
They wanted the CIOs to come up with a plan that would make data and systems interoperable across federal government. And so they had them write this document in which the use of an ESB — which was maybe an okay thing to do in the ’90s, though it was already fading from use; it was sort of outdated already by the time that it got in there. But it’s less a problem of the fact that it was already somewhat outdated when they wrote it in the document, and more a problem that nobody intended it to be required. But every level down, people would see it in the document from which they were deriving their document and go, “Oh, they must want us to do that.” So by the time it gets to the Air Force RFP, it’s seen as absolutely necessary and you can’t take it out because “Someone above told us we had to do it.”
And it’s sort of depressing that once there’s a lot of oversight on a project, everybody feels like it’s too risky to do something like take out a requirement. Whereas in fact, when you look around, I interview these other people and they’re like, “We just pretend there’s an ESB in our software if somebody asks about it.” If no one’s looking, you can get rid of these things. If everyone’s looking, it gets stuck and your software never ships.”
Geoff Kabaservice: Yeah, it’s fascinating. You wrote that “ESBs became mandatory and practice in DOD through overzealous interpretations of law, policy, and guidance combined with lack of technical understanding.” And I think that last part is just so key. People don’t understand the policies they are requiring. It’s like a cargo cult of sorts.
Jennifer Pahlka: Another really good example of that is FISMA, the Federal Information Management Securities Act, which is famous amongst government technologists as just being really overreaching. There’s these 300-and-something security controls that people say you are required to use. And in practice, everybody is required to do all 300 of these security controls, even though FISMA only suggests these as options and sort of examples, the same way that the federal enterprise architecture suggests the ESB as an example. But when you have a culture that’s always “better safe than sorry,” and the people doing the implementation have no idea what these security controls mean…
Let’s assume you’ve got a chief of security on a federal software project who’s great at their job. They know exactly what to do to secure this application, and they say, “Here’s the 25 controls we’re going to do. Oh, and by the way, we need lots of testing. We’re going to make sure that this doesn’t disrupt the usability of it, we’ve got features that need to go in. So we’re just going to do these 25.” They’ve got to get permission up the chain to do it. And the people up the chain have no way of knowing if those are the right 25. They only know that if number 26 needed to be on there and wasn’t, it’s their butt on the line. So they say, “You know what? It’s better if you just do all 300.” Well then suddenly so much of your budget of time and attention is gone to something that is not going to help secure your application. And mostly it will come out of the one thing that would help secure your application the most, which is testing.
And we actually make our systems less secure because of FISMA. It’s not the fault of the people who wrote it, it’s the fault of the people who interpret it. And I think there’s two solutions to that. Get more technical people in that chain to say, “Yeah, that’s a good 25 controls for this particular application.” Or have those people in the chain trust the folks at the bottom who know their stuff.
Geoff Kabaservice: Yeah, absolutely. And that is in fact the Federal Information Security Management Act. And you point out that this requirement to march through all 300 checklists adds months or even years to these projects, which is part of why the government delivers so slowly.
Jennifer Pahlka: And they’re part of getting an ATO, an authority to operate, which is even a larger set of things, which just can take forever. There’s been a lot of efforts to streamline it in various ways, but there’s so much focus on compliance and so little focus on things like, “Have we tested this thing? Does it work for people?” And that’s just got to shift.
Geoff Kabaservice: One of the valuable aspects of this book — one of the many — is that you also go through the history of how the U.S. government in effect hamstrung itself when it comes to technological expertise. And it was not ever thus. You point out that in the early days of computers, government dominated the industry and most of the expertise was there. But you point to a rather obscure set of measures in the mid-‘60s… In 1965, Congress passed the Brooks Act, also known as the Automatic Data Processing Act, which specified that technological acquisition in the federal government should go through the GSA, the General Service Administration. And then in 1966, OMB released Circular A-76, which says that the federal government won’t do commercial activity. And it’s significant, that memo, because it formalized the distinction between functions that are “commercial” and those that are “inherently governmental,” which in the latter case means management of government programs requiring value judgments. And so this over time, you write, has led to a government that just doesn’t know enough about the technological policy that it is implementing to actually have effective policy.
Jennifer Pahlka: Yes. I think this definition of inherently governmental deserves a relook in the context of digital. I think when computing was, “We need to buy a bunch of hardware and processing time,” it was pretty much a commodity, though it was still changing the world obviously. But once we hit the early ’90s and the internet and this digital age really started coming at us in a way that was going to fundamentally change the world, we needed to look back and say, “Is this whole field really about getting computers at a cheaper cost? Or is there something more profound that we need to grapple with?” And I think that our leaders kind of mistook that change as a change in the tools of implementation, not a change in how our society was going to run.
And if the definition of inherently governmental is that public servants are going to have to make value judgements in doing the work, then I think the creation of lots of technology, what features go into a service, who it’s going to serve — all of these things are absolutely judgment calls. And the idea that that is simply something you shove off to a vendor because all they have to do is find a bunch of requirements and fulfill a bunch of requirements and not make any judgment calls… It was a huge mistake that has set us back many, many years.
Geoff Kabaservice: And to actually make software that serves the people it’s intended to serve, you really have to engage in user research. And yet that is actually all but illegal in the federal government thanks to the comically misnamed Paperwork Reduction Act of 1980. Tell me about that.
Jennifer Pahlka: Well, it’s actually not illegal; it’s interpreted to be illegal. The Paperwork Reduction Act, for the record, does not include user research. You are not required to clear almost all kinds of user research through OIRA, the Office of Information and Regulatory Affairs, under the Paperwork Reduction Act. But so many people do exactly what I just described with FISMA, which is “Let’s better be safe than sorry. Let’s send it through for a review process, because I’m not quite sure what this is. The words ‘user research’ sound funny to me. And so let’s have somebody above me sign off on that.” And that just kills user research, because you don’t have nine months to wait for that clearance.
Geoff Kabaservice: Yeah. The norms, you conclude in that case, are more powerful than the intent and even the words of the law.
Jennifer Pahlka: And we have had, I want to say, at least three successive presidential administrations issue guidance, trying to tell people not to send them. But there’s something going on at OIRA too, because it’s not just the desk officers and the agencies who are sending this stuff up the chain for approval. OIRA is still taking things that should be just a rubber stamp, send it right back, let them do it — and saying, “Oh, nope, we’re going to post this to the Federal Register.” So my friend who was trying to get approval for user research she was frankly already doing, that had been cleared years before, was told by someone in her agency, “Okay, we’re going to have to clear that through OIRA.” She writes it up, sends it over, and OIRA’s response is, “Let’s post it to the Federal Register so that we can get public input on how and whether you can get public input. And in the meantime, let’s have no public input.” And it’s just deadly to her project, where what she is trying to do is to understand the needs of users at the CFPV and help make student loan applications (or I can’t remember exactly what it is) a lot more easy to access.
Geoff Kabaservice: Yeah, it’s Kafkaesque in the most literal sense of the word. It seems to me that your approach is not moderate or centrist per se, but you are pointing out the limitations of both the left and right approach to policymaking and implementation. And you point out that the right wants to shrink government by starving it of knowhow, digital or otherwise. But it hasn’t shrunk government through this approach; it has actually ballooned it. You write that “We spend billions of dollars on satellite software that never goes into space, pay vendors for web forms that don’t work, and make applying for government services feel like the Inquisition.” This is the opposite of the kind of non-intrusive, small government that conservatives intended.
And on the liberal side — I feel like this has been much more covered in recent years — but liberals approach all the problems of policy with a really overly technocratic expertise and a legalistic one, which prevents them from achieving the ends that they say they want, such as building affordable housing and addressing climate change. So what advice would you give to both left and right about how to move beyond this unproductive paradigm?
Jennifer Pahlka: Well, I think they should move beyond the unproductive paradigm. I think it’s also important just in the general public to not just think about the divide of the left and right, but to think about the divide between people. There’s sort of two different divides. There is obviously a divide between those who want to improve state capacity — and by that, I do not mean make government bigger, I mean the things I talked about earlier: have the right skillsets so that we get a lot more outcomes for our money, which includes thoughtfully designing government services instead of just letting them accrue over time — and those who want to dismantle the state.
There’s a lot going on that front. I have the “Mandate for Leadership” from the Heritage Foundation here on my desk, all 900 pages of it. And someone pointed me at Chapter Three, where the folks who are trying to advise what they would assume be would be an incoming Republican president, they want to advise that person on what they would do for the civil service. And they really just want to be able to fire people, basically. They want to reinstitute Schedule F.
I want to be able to hire people. I would like there to be a similar plan written not by the Heritage Foundation and not necessarily by a left-wing think tank — maybe you guys can write it at Niskanen — about how we make it a lot easier to hire people. Because we have fantastic people who want to join government, who have exactly those skills that will make government cost less and be better. It’s “tastes great, less filling.” It takes nine months to hire them or to make them an offer — in which time they’ve gone on to another job.
So in a certain way, what I want to recognize about the folks who penned this “Mandate for Leadership,” which is very much about dismantling the state, is that we are both looking at the same problem. I come to a very different conclusion, but I’m as frustrated with the status quo and what I think they call in this report “obstructionists” — I’m frustrated with them too. We’re both looking at demosclerosis and saying this is unacceptable. I think, in a sense, the real divide is between people who are fed up with the demosclerosis and willing to do something about it, and those who aren’t. And not many of these divides fall clearly on left and right. There are many, many folks on the left, as you mentioned, who are so skeptical of government they don’t want it to be stronger. And there are many folks on the right who are very engaged in the building of state capacity that I want to be a part of.
But there are folks who are just ignoring it, and in some ways that is as unacceptable to me as those who want to dismantle the state. Because I feel like we can have a healthy debate about whether the correct answer is to fire people or hire people — I think I will win! — but we do need to do something about it. And if the Heritage Foundation wants to be very bold on their agenda, at least let’s get the conversation started. And let’s bring in those people who are not spending the energy they should be on things like civil service reform.
Geoff Kabaservice: I agree 100%. And in fact, there are in your book examples of innovation on the conservative side, even though I don’t remember if they’re identified as such. But you point, for example, to the United Kingdom’s 2011 creation of the Government Digital Service. Well, that was under David Cameron’s government, the Tory Party, and it has been extraordinarily effective. And in fact, I want to say that you modeled the USDS after the example of the Government Digital Service.
Jennifer Pahlka: I really was deeply inspired by their work. Like everything, it has changed over time. It has had more power and less power, more success and less. But I was actually just there, what was it, last month actually, checking in on how they’re doing. And they’re still doing great work. For example, they are working on digital identity. They have the same challenge — not quite — a lot of the same challenges that we do on identity, in the sense that their populace is not going to be tolerant of something like what India did with everyone having a national ID. So they’ve got to find creative ways around that. They’re making a lot more progress than we are, and I think we should continue to look to them.
Geoff Kabaservice: Absolutely. And there also is the example of the Information Technology Management Reform Act of 1996, which is better known as Clinger-Cohen. Well, both Representative William Clinger of Pennsylvania and Senator William Cohen of Maine were Republicans.
Jennifer Pahlka: This is true.
Geoff Kabaservice: And the report in particular that you cite — you probably found the only copy that exists —came out of Senator Cohen’s office in 1994. It was really very prescient in foreseeing the kinds of chaos that would be caused by a lack of coherent policy for actually implementing technology in government.
Jennifer Pahlka: I think that what happened around Clinger-Cohen in ’96 deserves a lot more attention: the fact that they were identifying these issues very thoroughly, documenting them, and I think their analysis of what went was going wrong then was really on target. The problem is that they failed, ultimately, to do the one thing Clinger-Cohen should have done — I don’t mean Clinger and Cohen the people, I mean what the Clinger-Cohen Act should have done — which was get the White House to take on digital strategy. And though there is still a Clinger-Cohen Act — there are many provisions of it, it has had a lot of impact — they just failed to get them to take that seriously until Obama came into office and created the CIO and CTO positions.
Geoff Kabaservice: Absolutely. And you also have this great example of people like Yadira Sanchez, who you mentioned earlier, who do succeed in getting results. And you point out that there actually are great people working in the civil service. The idea that they’re all sort of the sisters from “The Simpsons” putting people through hell is wrong. And you’re right on what I think is something that people from both sides can agree, which is that government really needs to treat its users with dignity: “When services respect our time, minimize intrusion, use language we understand, and don’t make us feel stupid, that deeply broken relationship with government can start to mend.” And I really think that’s an ideal worth striving for.
Jennifer Pahlka: Yes. I think that most public servants want to do that. And they’ll fight to do it, and they’ll fight harder to do it if they see that there are tools around them and that they have support structures around them to do it. They don’t want to torture or insult their fellow Americans, and when they do it, they’re generally doing it by accident because the system is requiring it. I really think that public servants are the hero of this book, and I think the hero in my mind of this whole agenda. You might have noticed I dedicated the book to public servants everywhere and I said, “Don’t give up.” That’s all I have to say other than “I’m rooting for you.”
Geoff Kabaservice: And I thoroughly applaud that sentiment. But I guess here’s the problem. Elected leaders, when they look at the problems that are caused by all of these things you’re describing, think they can throw more money at the situation or put more regulations on it or add more oversight. And yet these actually worsen all of the problems. The solution, you say, is to put skilled technologists within government who are empowered to make the necessary decisions. Is that happening and what is the path that should be followed to get that to happen more often and on a greater scale?
Jennifer Pahlka: I don’t think it’s happening yet. I have seen some glimmers of it maybe starting to happen.
Geoff Kabaservice: You cite that letter from poor old Kenneth Glueck, who was then the SVP at Oracle, complaining that in-house government know-how was actually getting developed in IT, whereas really this is just what the capitalists in the private sector outside should be doing. The fact that he felt moved to make that complaint suggested that there was some change being made to the mentality.
Jennifer Pahlka: Yes. I think once you start having enemies, it means you’re making some progress. And I do think that there are folks in the executive branch and other leaders like that who are seeing the problem and doing something about it. And I call out, and will call out here as well, Cecilia Muñoz, who was in the Domestic Policy Council under Obama, who just fundamentally understood this in a snap. It just made sense to her and nobody needed to really explain it to her. And she’s just an amazing leader.
I think for elected leaders who obviously… There’s two problems with elected leaders. One, they’re not incented to deal with state capacity. They’re incented to sort of extract what outcomes they can on a particular topic out of the increasingly dwindling state capacity without ever renewing it. It’s sort of like trying to get more crops out of a garden without ever fertilizing it. Is that a reasonable metaphor? And they’re not inclined, I think, to look at the machinery of government.
But also they really believe that their levers are money and oversight. I think oversight is probably more powerful; if they could change how they do oversight, it could be super powerful. And they can change how they do money, how they fund projects. But they kind of only know a couple tricks: give them more money, give them less money. It’s not like: How are we giving them money? How are we deciding what gets more money? How are we deciding what gets less money? How are we lifting up public servants who are doing great work instead of just yelling at the ones who are failing or having trouble?
But I want to believe that, those misaligned incentives aside, a big part of the problem is that they just don’t know what to do. Because I hear from staff members and electeds from time to time saying, “Can you just explain to me what to do to fix this problem?” And the problem is, it’s just been hard to explain, which is why I wrote the book. And I think we hopefully are at the beginning of a conversation that will give elected leaders different tools and different ways of intervening and using their power that make things better instead of making things worse. Because I think they are making things worse unintentionally.
Geoff Kabaservice: I agree, and I think you put that very well. But you also mentioned in your book that, as you put it, “If our elected leaders have the wrong incentives, perhaps there’s another layer up the chain to implicate. And that is we the public.” What can we the public do to have a better understanding not just of what government should or shouldn’t be doing, but what it can and can’t do? How can we approach this whole question of better governance with better results?
Jennifer Pahlka: I think in general many, many people could do tours of duty in government, in technology or not, that would help them understand this machinery, and why it’s difficult, and have more empathy for public servants and therefore change the ways in which they personally hold government accountable. I’m not saying we shouldn’t hold government accountable; we absolutely should. We just need to understand what makes it better and what makes it worse and ask for the right things. Public service, to me, is a really big way.
I will also say, for technologists and designers in particular, I mentioned a lot of places in the book where they could go work. But you don’t even have to jump right in with a job. I’m on the board of an organization called United States Digital Response, which has thousands of amazing technologists doing pro bono work on a short-term basis and a part-time basis for cities, states, and even a little bit for the federal government. It started in COVID when there was huge, huge need for people to jump in and put up forms or fix data pipelines. And it’s just turned out that the government still wants their help, and technologists still really want to help. And it’s a great way to do this without quitting your job or without taking a job in government. So that’s another way.
But I do think that whether you’re just somebody who’s going to show up at a town hall or maybe a city hall meeting, or you’re in the donor class, for sure we need to ask our elected officials about this stuff. Nobody does that. We’ve got to start.
I’ll give you a really concrete example. If you care about the climate and climate change, we have just passed the Inflation Reduction Act. We need to implement the climate provisions of the Inflation reduction Act. One of the things that means is that your local building permitting office is going to have to start permitting residential solar installations and upgrades of electrical and all the installations of heat pumps at a far higher rate than they have been. Well, is that office prepared for that kind of scale? Is anybody asking them about it? That’s the kind of intervention and pressure that the public could put on their government that we just haven’t really thought of.
Geoff Kabaservice: And this is an existentially important question. Because as you write near the end of the book, “Our global standing in the world has declined since the mid ’90s. Our response to COVID was embarrassing. Our tax system is the joke of the developed world. Our military might is waning and our health outcomes ranked dead last among eleven peer countries.”
Jennifer Pahlka: Did I say all that? Who let me write that?
Geoff Kabaservice: Harsh but true. And you conclude, “We can no longer afford not to learn new tricks.” And I think you’ve really given us a roadmap for progress in understanding the problems of government and how it could work better, and I think that’s been a tremendous public service. So thank you very much again, Jennifer Pahlka, for joining me today, and congratulations again on Recoding America.
Jennifer Pahlka: Thank you so much. This has been a really fun conversation, Geoff. I hope to have it again soon.
Geoff Kabaservice: Thank you, Jennifer. And thank you all for listening to the Vital Center Podcast. Please subscribe and rate us on your preferred podcasting platform. And if you have any questions, comments, or other responses, please include them along with your rating or send us an email at firstname.lastname@example.org. Thanks as always, to our technical director Kristie Eshelman, our sound engineer Ray Ingegneri, and the Niskanen Center in Washington, D.C.