How to Actually Implement a Policy

"To me, this is all screaming: you need leadership"

At Statecraft, we’re interested in what happens to policies after they get passed as a bill, executive order, or regulation. Take the failure of healthcare.gov, or America’s broken environmental review process, or decades-long waiting lists at Veterans Affairs. As today’s guest understands, sometimes policymaking isn’t done when the policy is made; often, it’s just getting started.

Jen Pahlka has been obsessed with policy implementation for a long time. She founded Code for America (modeled on the Peace Corp and Teach for America), served as the Deputy Chief Technology Officer for the Obama administration, helped found the United States Digital Service, an elite federal government tech unit, and co-founded US Digital Response in 2020. These days, she’s a Senior Fellow at the Niskanen Center and the Federation of American Scientists.

I was joined by my IFP colleague Alec Stapp to interview Pahlka (Alec wrote a great review of Pahlka’s new book, Recoding America: Why Government Is Failing in the Digital Age and How We Can Do Better).

What You’ll Learn

• What the IRS is doing right

• Whether Schedule F can fix civil service

• Why federal contractors fail

• Why it takes nine months to hire in the federal government

For a printable transcript of this piece, click here:

---

Obamacare was perhaps the most infamous government technology rollout of the century. The website managed to serve, on its first day, a total of eight people. What the hell happened? 

Eight lucky people, and many hundreds of thousands of unlucky ones. You had a congressionally mandated deadline and a very big task. And you had an agency that pretty much had to use one of the vendors already on contract, because the procurement process can take a really long time and they only had two or three years. 

So you already had a set of actors, including CMS itself [Centers for Medicare & Medicaid Services] and the vendor and others, who were operating in a government technology mode, not a consumer technology mode. There's not that much that CMS builds that's actually citizen-facing, and certainly not that quickly. This was something that had to serve millions of people. 

Lots of people have had their say about what went wrong, and they’re largely right, but the thing they don’t say is that healthcare.gov didn't have a product manager, in the tech-world sense of that term. The distinction between a project manager and a product manager is that a project manager gets stuff done, but a product manager decides what to do in the first place. 

We have these assumptions in government that there's no choices to be made. This is the policy, these are the requirements derived from that policy and other policies from domains like cybersecurity, for example. And you just implement them. It's a process of checking off a lot of boxes. 

When we make good consumer software, we set priorities. We try to really understand what people want, and we don't try to do everything at once. The people fundamentally in charge believe that's the only way to do it, despite the fact that you had great people in the system saying, “What if we had a staged rollout? What if we only served this kind of customer from the beginning and added the others?” If the site had launched to a limited population, it's not that others wouldn't have been able to sign up for health care. They would have gone through the call centers and the in-person service centers that were designed to handle people with complex edge cases. 

But the CMS team was told that wasn't possible. And so they tried to do every single possible edge case from day one. When you try to do that, to serve everyone from the beginning, you serve no one. Contrast that with the IRS’s Direct File. That's serving a very small portion of the population [141,000 successfully submitted tax returns], learning what works, working out the bugs, seeing what people like and what they don't, where they get confused. That means they can make it better by the time that they bring the next tranche of users on. And it’s working!

This is actually the way that good software gets built iteratively, by adding people and use cases over time. We've come so far from healthcare.gov, but I've taken several calls from journalists who want to know why the IRS is doing Direct File this way. They're like, “It seems terrible that they would serve so few people.” And I just point out, did you try to sign up for Bluesky? You went on a waiting list, and they added you when they were ready. Did Yelp start out serving every single city? No, they started in San Francisco, and then they added New York and then LA, as they got it better over time. 

And what they're able to do is the opposite of that quote from Clay Shirky: “The waterfall method amounts to a pledge by all parties not to learn anything while doing the actual work.”

Not learning anything while doing the work is what the healthcare.gov team were forced to do, and you saw the results of it. Now you've got a Direct File team able to learn and improve as it goes. So by the time the rest of us get invited to use Direct File, it will be far better. 

Could you explain what agile vs. waterfall means in this context? 

It's two styles of building software. Waterfall essentially says you have these distinct phases: gathering requirements, designing it, building it, testing it. And it's a sin in waterfall to go back. Once you've built your requirements, never change them. Once you've done your design, don't change it. 

Agile, that is all the opposite. You're in a build, measure, learn cycle over and over again. You build something, if it works, you learn from it, you build more. In a waterfall paradigm, the software never touches users until the very end, in what’s called user acceptance testing. That's very different from agile, in which you are working with users from the very beginning.

User acceptance testing sounds a bit dystopian. 

It is. And it's very different from the mantra, “understand user needs and meet user needs.” You have to be working with users all the way up through your development to achieve that. 

In the book, I use this dichotomy as a metaphor. Our entire system of government is built as a waterfall, especially when you have to think about implementation of the policy.

At the very top, you have whatever legislative body writes the law. It gets handed off to agencies who write regulations based on the law and policy, and it descends all the way down to the people who create the interfaces that administer that law. In the case of the ACA, that was healthcare.gov. In the case of the IRS, it's now Direct File or your regular tax forms. Those people are at the very bottom of the waterfall. If learning only goes one way, the people who are actually interacting with the users of the system have no way to tell the people who wrote the law what's working and what isn't. And we continue not to learn. 

You mentioned that some folks advocated for a staged rollout of healthcare.gov, and they were overruled. IRS is also getting criticized for its staged rollout of Direct File. 

Are these issues bigger problems with progressives who value equality and inclusion, and think that on day one, a new government service should be available to 100% of people? Or is this an issue inherent to government, even when Republicans are in charge? 

I think that there has been a misinterpretation of the progressive value of equity to mean that staged rollouts would harm folks, because there some people wouldn't be able to use it. And, to be honest, in the case of healthcare.gov, it's true. For instance, there was a small segment of immigrants who were eligible for healthcare under the ACA. And it's very hard to code software for those far-out edge cases and get it right from the beginning. 

So it does feel sometimes like you are disadvantaging the least advantaged. You're failing to serve the most vulnerable. But the reality is, if you don't do a stage rollout, you're very often going to serve no one at all.

Certainly the ACA was extremely politicized, but the people who built it were neither Republicans or Democrats. They're bureaucrats. It’s less the left or right creating this environment, and more what Ezra Klein has called “cramped, small, professionalized thinking.” It’s very lawyerly, but it’s not always lawyers who act this way, and not all lawyers act this way. “I'm interpreting these pieces of law and policy and regulation in the most rigid, persnickety way and trying to get exactly the thing that we think the law says, rather than what the law intended.”

There's a real divide between those people who are constantly looking at the letter of the law and those who are returning to the intent of the law, trying to have flexibility and judgment, and create space for design in the implementation of law.

Klein has argued that this lawyerly approach is more associated with the left, and that may be true, but in a bureaucratic context, you don't even know if the person's on the left or the right. You just know that they're saying, “A staged rollout is illegal because I'm interpreting this equity statute that way.” 

Larry Kramer has this notion that the law is supposed to be interpreted by normal people, in a book called The People Themselves: Popular Constitutionalism and Judicial Review. He argues there's a real history in the U.S. of saying the law must make sense to normal people, not just lawyers and judges. 

I talk in my book about this woman, Natalie Kates, on a project at CMS, trying to fight the policy team. The policy team is saying, “There are nine different definitions of a medical group, and we’re going to make doctors read through these nine different definitions as they start the process of complying with the new quality payment program.”

And Natalie says, “We can't do that. There's enough in common in these definitions that we can resolve them down. We're not burdening doctors with that.” She doesn't get all the way to one definition, but she does get to two, which makes it a lot easier. What she's saying is, I get that it's complicated, but it has to make sense to a person.

People with specialized knowledge, whether it be law or policy or whatever — and you meet a lot of people like that in government — are good people, and they're trying to do the right thing, but what they don't realize is they're so attached to that specialized knowledge, which they feel gives them value and status, that they're missing the forest for the trees.

And Natalie's right. It has to make sense to a person. There’s roots of that in American history and culture that we need to tap back into. I think that's the way out of this small, cramped, professionalized thinking, where you're just so proud of all nine definitions, instead of saying, “No. We can't have nine definitions.” 

Or, in the case of a program at Medicaid, Congress has exempted doctors who take very few Medicare patients from a particular rollout. The people trying to deliver in a way that makes sense say, “Great, let's exempt them based on last year's data.” But people with small, professionalized, cramped thinking say, “The most accurate way to do what Congress said is to make everybody go through the program the first year, then see who's under the threshold, and then exempt them.” But making everyone go through a program is extremely burdensome and it actually drives people away from the program. That's the big divide, between lawyerly thinking and user-centered thinking. 

I agree that we have a lot of lawyer brain in government, maybe disproportionately on the center-left and among progressive groups.

Is this because if you were to survey the personalities and political ideologies of students in law school, they tend to lean a lot more progressive, and then they want to do big things with government, and we now have a lot more lawyers in government relative to other types of professionals?

Or does causality run in the other direction? U.S. institutions are highly litigious, and the government needs a lot of lawyers to defend against litigation. So it draws more lawyers into government at the expense of other types. 

I really don't know. I do observe that the successes of the progressive left, say the civil rights movement, have all come through laws that were then strictly enforced.

There's just a belief that law is the way progress happens. I believe that, but I also believe that the delivery end is where this happens, actually giving people what they need and expect from government. That involves a different mindset. Progressives have too much faith, sometimes, in the power of law. I would encourage them to think about the power of actually delivering what the people need, which sometimes involves being a little bit flexible with the law.

One more question on the healthcare.gov case study before we move on. How about the power of government contractors? Before reading your book, I'd have had a different belief about government delivery of services. You totally persuaded me that these contractors are often terrible and can very often be worse than doing stuff in-house.

But before reading it, I might have thought that large contracting firms can hire a lot of product managers, they can pay them market salaries to produce good software and meet deliverables. 

To me it's not an in-house or out-of-house thing. The terms are set by what the government decides to contract for. Teams learn how to contract for a product instead of a project. But once the government puts out a Request For Proposal [RFP] with 6,700 requirements — and that's not a made-up number, that's a big state software project — you're already in a death march of fulfilling requirements without any prioritization or understanding of user needs.

A great contractor with fantastic product managers is not going to succeed, even with the right staff, because the project’s success has been defined as a process of checking off requirements. And there's never a requirement that the software actually work for the people who use it. That's just too vague. 

But if you put out an RFP that is structured differently, with a different definition of success, then yes, a contractor can come in and say, “We've got exactly the right product manager for this, here's where we're going to deliver it.” 

Even then, though, they have to rely on there being product ownership on the inside. A government software project can’t succeed if you don't have a very involved partner on the inside resolving ambiguities in the policy with you. In a traditional, legacy development framework, the government essentially abdicates responsibility for the outcome to the vendor. When it doesn't work, it's not our fault: it's the vendor’s fault. And even those vendors who are used to working in this requirements-heavy, death march style will say, “It's actually impossible to do this because there's no one helping us make product decisions.”

Contractors should not be interpreting regulations without somebody from the government sitting by their side, because they're essentially making policy decisions, which should be done by public servants.

Even the contractors that I consider less helpful in the ecosystem would love to see changes like that. But the contractors offering fantastic services don't just want those changes, they require them. 

So there are incumbent contractors that have mastered the current system. They're generally opposed to changes, because they make good profits under the current system, regardless of the effect on the end user of government services.

In a world where the system changes, do some of them see a clear path towards a win where they deliver better services and it's also better business? Are they reading your book and getting excited? 

Some of them. 

Name names. I'm ready. 

Look, there's always a difference between people in companies and companies themselves. It's hard to find somebody who works for one of the big Beltway bandits who wouldn't complain about the same problems.

Here’s a good name: Nava PBC. It's a contractor that works with CMS and a lot of states, very good at product management, and they deliver excellent user-centered products. They’re the kind of contractor that requires that product ownership by their government partners. A Beltway bandit employee will complain about the same thing because, on a personal level, they want to deliver good product. But as a company, they may not be as eager to see change in the ecosystem, because they definitely benefit from the system.

I believe that there is plenty of work. And if we went to a product model more frequently in government, you would see smaller contracts go out, but more of them, and you would have a greater variety of contractors, which is really good for the system. 

But those big Beltway bandits would still have a lot of projects. They just might have more smaller ones. There'll be plenty to go around, but I'm not sure they necessarily can get their heads around that and really advocate for the changes that need to happen to get there because they do benefit from the status quo.

---

Jen, it’s time to rank federal agencies. Which agencies have improved the most over the last ten years? 

I have to start with Centers for Medicare and Medicaid Services because there were some wonderful people there who were kind enough to let me feature them. 

In particular, Yadira Sanchez. You can see the shock from healthcare.gov that caused the change, and you can see the ways that folks like Yadira shifted practices. When they delivered their next product after healthcare.gov, the quality payment program, it was a huge success: on time, under budget. People were calling the call center, not to complain, but to say, “This is so easy, I must be on the wrong website?”

I should also shout out the VA [Veterans Affairs], which is going to surprise some people, because there are things still very wrong with the VA — and frankly, there's still things going wrong at CMS. But in 2013, Marina Nitze was recruited to be the first CTO there, and the transformation that she kicked off has been profound. [We interviewed Nitze for Statecraft: How to Salvage the VA.]

And then Charles Worthington ended up succeeding her at VA, and has been there for I think 10 years now, which shows the power of people sticking around to see change through. Yadira Sanchez has been at CMS for 24 years now. The VA used to be one backlog after another. Marina and the team start by fixing this veterans healthcare application and showing that there's a different way of doing this. They brought a video of a veteran trying to use the legacy healthcare application to the deputy secretary and they made him watch how hard it was for a user.

And that's the beginning of a culture shift, which says, It doesn't matter if all the boxes were checked, if they can't use it, it's not finished, it's not done. Fast forward to COVID, the VA was texting veterans saying, “Click this link to schedule your vaccine.” It was three clicks and they had a vaccine appointment. It isn't perfect, but they now are delivering services to veterans that are world class. That's an enormous transformation, but it has taken quite a while. It’s taken people like Charles showing up, sticking around, and making sure the agency doesn’t backslide when it makes progress.

Those are great examples. You mentioned that Yadira Sanchez has been at CMS for 24 years. I'm curious about civil service reform, and getting the best people into positions of leadership. Has Sanchez been fighting the good fight for 24 years, and now we're seeing the fruits of that as she transforms CMS?

Or is she someone who changed over time, and learned a new set of ideas? I guess my question is whether we mostly need to empower existing people, or help well-intentioned folks do a complete 180? 

I think it's a combination, but the biggest thing we could do right now is find and empower Yadiras. They're there. She didn't have the language for agile development until she worked with the healthcare.gov team, but it's not like she was a tabula rasa. She's incredibly creative and mission-driven, and was doing fantastic work before healthcare.gov

I think there's been this thesis that the way to make government better is to bring people in from the outside. It's important to bring people in. I value that. But it’s a big mistake to hold them up as saviors, when we should really be investing in and empowering the strong leaders we have today, giving them exposure to ways of working that legacy frameworks just don't allow for. We have great people in government; we need to unlock their potential.

Empowerment isn't just the knowledge; it's also the will. When our public servants are standing up for the user, they need to be able to win. And so much of our government is constructed to fight against them. 

There's a big conversation going on right now about changes to the civil service, including a second-term Trump administration invoking Schedule F, removing protections from many civil servants, and bringing in more political appointees. 

What do you think of that? For people who are skeptical of the Schedule F approach, what’s the alternative agenda? 

I think Schedule F is dangerous, because civil service protections are important, and there can be a disregard for what public servants do. But I also think the status quo is intolerable. We need to make it easier to hire better based on real assessments of people, better support, better training, and also easier to fire.

The left doesn't like to talk about that, but I absolutely believe that is important. And it can be done without significant congressional action. It can be done by setting a different tone from the top about what's important. It's less important to check every box and make sure that the process that you run to hire somebody is unassailable if it’s a nine-month process. It's more important that you just hire the person in the timeframe that's needed. 

And we need leadership. Whoever it is needs to set that tone. I do not want Trump in charge doing Schedule F, which is like setting off a nuclear bomb, but nor do I want us to continue on our current path. 

From my understanding, time to hire has gone up in the Biden administration, not just relative to the Trump administration. It’s also up relative to the Obama administration. I think that's the wrong direction, in a way that’s far more critical to the future of our country than people realize.

We're in this crisis of state capacity, and there are really only three ways to build state capacity. You can have more of the right people doing the right things, you can focus them on the right things, and you can burden them less.

I'm very interested in burdening them less, but we also are going to have more of the right people doing the right things. If we take nine months to hire people and don’t assess their skills when we do, it’s incredibly destructive. 

So I would be looking for far bolder leadership from the left, or frankly anybody, on this issue, so there’s a sane reform agenda on the table instead of the insanity that is Schedule F. Part of that bold leadership is not using the excuse that since Congress hasn't acted, we can't do anything. There are programs like the Subject Matter Expert Qualifying Assessments [SMEQA] that show very clearly that you can do speedy, accurate, fair hiring of skilled professionals with the existing law and policy that we have. We do not need Congress to change anything to do SMEQA. But you have this incredibly successful pilot that scales across government very slowly, because our leadership isn't calling it out as a priority, and isn't saying that the results we get through SMEQA are more important than all of these little rules that have accumulated over the past 100 years in the agency hiring manuals. This is a solvable problem by others besides Trump. 

Let me cut in here: some readers are not going to understand how it could take nine months to hire somebody and not run the kind of assessment you're talking about. Will you back up and explain what the heck's going on there? 

I don't think I can accurately, in less than an hour, describe all of the touchpoints in having someone get hired, but I can describe some of the perverse nature of it. 

This is a good example of culture eating policy. Let's take, for instance, laws to promote meritorious hiring and our Merit System Principles. We moved in the late 1800s from a spoils system, where people got jobs in government because the person who got elected owed them a favor, to a system where we objectively decide on the right person to hire.

So we put all these rules in place. And then civil servants, especially in HR, have taken this issue of fairness really literally. Specifically, they've taken it to mean that every single applicant needs to be treated exactly the same and no bias can come into the process. That means that if we use a subject matter expert to decide if I'm a better Java programmer than Bob over here, that’s dangerous, because it might insert bias into the process. Whereas if I'm objectively reviewing their resumes and cover letter, then the decision will be more objective. 

But here’s the way it works. I have a job open. Let's say I get a thousand resumes for that job. As the HR person, I have a really difficult job to do: come up with a slate of candidates that the hiring manager can choose from, and the process that gets me to that slate has to be unimpeachable. So we first do a resume review. What we look for is exact matches in the wording between the resume and the job description. 

Many hiring managers have told me— I’m not making this up — that people cut and paste from the job description into the resume and don't even reformat it. They don't change a single word, and they go to the top of the hiring list, even if it's completely obvious that it's a cut and paste. 

Jack Cable won the Hack the Pentagon contest several years ago, genius programmer. By definition, he’s one of the most qualified people possible to work on the Pentagon’s cybersecurity. He then submitted a resume for a job at the Defense Digital Service, but instead of cutting and pasting from the generic job description, he included a list of the programming languages he knows.

And he was rejected something like five times. They told him, “If you want to get a job here, you could go work at Best Buy selling computers for a year and then reapply, and then you'll qualify.” So there's this insane down-select: whose resume most closely matches the job description? 

The second down-select is a self assessment where they send those candidates a form to fill out that says, “Here are the characteristics we're looking for. How would you rate yourself?” The way to get through that down-select is to rate yourself as “master” on every single one.

So you’ve down-selected twice. Let's say we now have 100 resumes. Then you can apply “veterans preference” to that candidate pool. And that's your slate. Technically you have done everything right, but you have not given the hiring manager anybody competent in anything but cutting and pasting – and lying. 

And so there's two ways that you see culture eating policy here. One is that our laws are designed to give us a meritorious system, but, through its implementation, the system is now deeply unmeritorious, because the only way to get selected for one of those jobs is to know how to play the game. It's an inside job.

The other is that the law was designed to preference veterans, but the evidence shows that applying veterans preference in this way creates significant bias against veterans. Because then an HR person takes that 10-person slate and gives it to the hiring manager. But the hiring manager understands that because the process did not actually assess the candidates for their ability to do the job, that slate is full of unqualified people who happen to be veterans. Hiring managers learn that if it's an all-veteran slate, those people are unqualified, and they reject the slate.

Fun fact: half of all hiring actions in the federal government are just thrown out the door. But there were often very qualified people in the candidate pool, and some of them were veterans! It’s just that the process didn’t select for them, and hiring managers know that. So it doesn’t help the veterans who make the cert, because the hiring manager throws the cert out anyway. And it doesn’t help the veterans who were actually qualified, because the hiring manager never sees their resume. And that’s how a law intended to help veterans actually creates bias against them.

---

Are there agencies that have concretely improved in recent years specifically in getting talent into the right positions?

Some agencies are trying to use the SMEQA process, which they report as being very effective. Part of SMEQA is that you can do pooled hiring. So let's say there's an identified need, for instance, for data scientists. A bunch of agencies need data scientists. Rather than every person applying for every individual position in every agency, they say, “We're going to hire 50-100 data scientists.” Then the resume and candidate assessment happens at the interagency level. The agencies can interview from that pool, and it's way easier. That's been effective.

But I don't think there's an agency that feels like it's gotten hiring to where it needs to be, by any means. And I hear from some folks that the Office of Personnel Management [OPM] attempts to scale SMEQA are watering it down in ways that they're not happy with. [On April 29th, after our conversation, the Biden administration announced it would rework hiring requirements for certain IT positions to be more skills-based and allow candidates to “self-assess” their technical qualifications.]

And you hear that from being plugged into this discourse generally? 

I ask a lot of people a lot of questions. 

How very gumshoe of you. 

In the past 10 years, what agencies have really backslid in terms of state capacity?

I don't think I'd call out anybody in particular. But there are a lot of people who think it's more risk-averse now than it ever has been. One commenter on an op-ed I wrote complained about how bad it’s gotten and said “We're murdering the mission.”

What are the background factors that contribute to that increasing risk aversion? Speaking in broad strokes. 

That's such a good question. I don't think anybody really knows, but you can certainly point to the polarized environment. Process is easily weaponized. “I don't like the outcome of that decision, or I want to give you a hard time. So I'm going to write letters to agencies, demand answers, and responding to them is going to end up taking a lot of their time.” And the more you're the subject of oversight in all its forms, the more you tend to want to defend your actions, and the more you rely on, “I followed this process, so it must be right.”

So part of it is partisan politics. But there is also just an unhealthy mindset of respect for the status quo. I'm a patriot and I love government, I’m pro-government. I think government is the most important thing in our country. But I am not pro-status quo. And I feel like sometimes people of all different stripes, but maybe it is more a little bit on the progressive or liberal side, have this respect for how we do things today in government that gets equated with love of government.

I think that if you really love government, you want it to change. I'm someone who feels very strongly that equity is important and that we need to do a better job of it. And that the processes that we're using to ensure equity are not ensuring equity.

In Alec's American Affairs review of your book, he used the term “adversarial legalism” to describe the use of litigation to fight political battles. 

In our last call, you talked about the need for leaders who tell their agencies, and I’m paraphrasing, “Don't worry so much about the letter of the law, here's what we're trying to get done.” But adversarial legalism exists: we live in a world with partisans on the other side looking for the opportunity to bring you down. Tactically, how do you deal with that?

Let me edit something I said earlier. I don’t want leaders to tell their people to be flexible with the letter of the law. Generally, the law is written in a way that's highly flexible and gives bureaucrats enormous amounts of ability to exercise judgment. It's following very maximalist interpretations of the procedures that derive from that law that are the problem.

So again, take veterans preference. It's totally fine. According to the law, you're supposed to apply it after a qualifying assessment. If we were actually doing what the law said, we would be honoring the intent. It would be a totally equitable and fair thing to do. That’s just not what’s happening, because people lose sight of the intent and interpret merit principles so rigidly that we don’t actually assess our candidates. Out of fear of introducing bias into a process, we miss the whole point of merit – and do a great disservice to our veterans in the process.

Or NEPA, right? The authors of NEPA [the National Environmental Policy Act] did not say there's a four-year review period before we can approve a permit. They did not say that you're going to have 1,000-page documents circling around 12 different agencies before these things can be approved. Marc Dunkelman said the authors of those laws thought that two bureaucrats were going to talk for 15 minutes, make a call, and it would be done.

And now this 15-minute chat has turned into a four-year process involving literally hundreds of people, full-time. From law, we have iteratively derived and ensconced policy, regulation and procedure that is now perceived as law. What I want leaders to do is to go back and say, “What was the intent of this law? Are we fulfilling the intent of the law?”

Now, absolutely, a lot of this four-year process that NEPA requires now isn't just because bureaucrats maximized everything. It's because they're going to get sued. But those two things come together. 

When you ask what I want a leader to do, let's take NEPA review as an example. You have a project that is put forth, and it's going to need to get through this process. Let's say 12 different agencies will likely have to touch this. What I want is for leadership to say, “Here's the lead on this who's going to shepherd this through all 12 agencies. And that person's job is to really assess where the risk is.” 

There's several places where you are open to lawsuits on a project. But then there are thousands of pages of things that absolutely no one cares about, and they're not going to sue about, and they aren't required by law. They're just the standard procedure that people go through unless they're told otherwise. 

Again, there's this concept in tech of product management, which is, “What is the important stuff for us to do here? Let's do that really well. Those are our priorities. There's a million things we could do, but trying to do all of those million things instead of the important stuff does not get us to the outcome that we need.” 

I'm looking for product management in the sense of a leader saying, “We need to get an A+, 100% on these areas, because that's very likely where we're going to get sued. These tiny little details over here that were required for some project ages ago, because it was this weird edge case, don't really apply to us. Don't do them.” 

And I want them to navigate this project through 12 different agencies simultaneously, instead of sequentially, and to constantly get all of the people who touch it together to say, okay, where are we now? That's leadership, and it's all leadership in the implementation bucket, not the policy bucket. We just get shafted on that. 

The layman in me is saying, “What do you mean there's not a lead on these permitting review projects?” That’s insane. 

There are affordances in this process for what to do when there's more than one agency involved. “One or more federal agencies may act as the lead agency or joint lead agencies and coordinate the need to effort.” It's not used that much, but when it is used, it doesn’t work the way it should. The lead agency, and someone within that agency, needs to have enough authority and power to move things along faster. 

The Niskanen Center has a great report out now on the challenges in transmission permitting, where they cite the example of the TransWest Express project, where “although routine coordination calls were held weekly and monthly for more than five years, “major issues remained unresolved as decisions were only finalized if there was a ‘consensus.’” Was everyone on those calls just too polite and continually deferred to their colleagues in other agencies? Or was there no one there with the authority to resolve conflicts, make decisions, and move the project forward?

I guess what I’m asking is, isn’t that obvious? Doesn't this exist in the business world? If you want something done, you tell someone this is their project. Help me understand why that's such a foreign concept. 

[Pahlka laughs] Yeah, it's totally obvious. And I don't think it's happening. It's not to say that individual people aren't trying to take on that role. It's not to say that anybody here is acting with malfeasance or even incompetence. It's that our system doesn't value that kind of leadership in implementation. It’s valued in politics, and in policy. But when it comes to implementing that policy, the people with the power to make things happen are busy with other priorities.

There's this fantastic book from 1973 called Implementation, and its subtitle is a full paragraph long [How Great Expectations in Washington Are Dashed in Oakland; Or, Why It's Amazing that Federal Programs Work at All, This Being a Saga of the Economic Development Administration as Told by Two Sympathetic Observers Who Seek to Build Morals on a Foundation]. The authors spend the first three chapters explaining what happened in this EDA project [Economic Development Administration] that had high expectations but went nowhere. 

And then they take all of the reasons that people give for why it failed and examine them, and none of them are really the right answer. But the one that's the most interesting to me is that, of course, they say it needed better coordination. Pressman and Wildavsky have this brilliant stuff to say about coordination: 

[Pahlka reads aloud.]

“No phrase expresses as frequent a complaint about the federal government as does ‘lack of coordination.’ No suggestion for reform is more common than, ‘What we need is more coordination.’

Yet when an evil is recognized and a remedy proposed for as long and as insistently as this one has been, the analyst may wonder whether there is not more to it. Participants in a common enterprise may act in a contradictory fashion because of ignorance. When informed of their place in the scheme of things, they may be expected to behave obediently. 

If we relax the assumption that a common purpose is involved, however, and admit the possibility, indeed, the likelihood of conflict over goals, then coordination becomes another term for coercion. Since actors A and B disagree with goal C, they can only be coordinated by being told what to do and doing it. Coordination thus becomes a form of power… 

Coordination means getting what you do not have. It means creating unity in a city that is not unified… It means compelling the federal agencies and their counterparts to act in a desired manner at the right time, when achieving this purposely is precisely what you cannot do. 

You can't learn how to do it from a slogan that tells you the way to get what you want is to already have it. The invocation of coordination does not necessarily provide either a statement of or a solution to the problem, but it may be a way of avoiding both when an accurate prescription would be too painful.”

To me, this is all screaming: you need leadership. You cannot just invoke the notion of coordination. It is a meaningless thing to say. 

The book’s ultimate conclusion is not that this failed because of a lack of coordination, or any of the other things that they talked about. The problem is the separation between policy and implementation. That is what caused that project to fail, and what causes most projects to fail. 

You talked about how veterans preference is in some cases an anti-signal, and a full slate of veterans will be thrown out.

As I understand it, that’s a similar dynamic to that of “Ban the Box” initiatives, where taking away information results in greater bias against the people you're trying to take care of.

I think it's part of a whole set of activities and interventions that we try to do to increase equity that have the opposite effect. Because we're talking about these things in an ivory tower, instead of seeing how they play out within a bureaucracy. To me, that is another example of culture eating the policy intent. 

Is there a generalized through line in the Ban the Box case and in the veterans preference case that we should be thinking about? I don't know how to articulate that theme, but it seems so obviously like the same phenomenon.

I don't think it's just taking information out. I think it's a general category of restricting how people are going to make decisions. The bigger principle at play here is that decisions are a hot potato in government. Nobody wants to be the person who made the decision, because making the decision is risky. And, not only can you shove decisions to another person, you can shove it all the way to the process, and say, “Look, this is what the process gave us.” 

When people talk about using AI in hiring, they say, “This is great, because the AI can now select the candidates.” That's just doubling down on the dysfunction of not wanting to be responsible for having made a decision. 

Sure, in the Ban the Box case, it didn't work because you're still letting people make those decisions and they're still biased. But there's this whole trend of saying, “Because people are biased, let's constrain them in this way, and then we won't experience the bias as much.” And it pops back up everywhere. 

You're better off dealing with the problem of human beings than trying to say, “Let's not let humans make this decision. Let's make sure the process chooses it.” And I should be really clear: those slates of veterans usually have some very qualified veterans. But you don't know, because the process didn't actually assess their skills. 

I really love what you're saying here, that there's no getting away from human judgment in a democracy. Some people have to make decisions and they have to be accountable to those decisions. 

Yes. 

That's a consistent theme in these interviews. Someone has to choose. You'll obscure responsibility or you'll hide it, but someone will make a decision, it will be a good or bad decision.

There's a new book called The Unaccountability Machine by Dan Davies. He says, take any scandal over the past 20 years. When the news first breaks of this scandal, everyone goes, we're going to have to hold these people accountable. 

And then you follow the commission that investigates it, or the journalism, or the official investigation. They almost always come out with, “There's not really anybody to blame. It's the system that we created.” This idea of systems as accountability sinks. 

There's this Catholic philosopher, Romano Guardini, who a bunch of my friends are interested in, but he's got a book, The End of the Modern World, written right after WWII where he flags this problem, that the whole of Western society has decided we can take decision making away from people, because good judgment is rare. In practice, it seems like you're going to have to rely on the good judgment that exists in the world. There's no other method. 

And that's the biggest criticism I get of my work. When I talk about empowering digital service teams, people say, “That's all well and good when you've got really good digital service teams. But, Jen, we have to constrain them, because a lot of the digital service teams aren't any good, or they don't exist.” 

And I'm like, sure, yes, I get that. That is a capacity problem. But that path of thinking — they're not good, so we should constrain them — is the exact path I'm trying to get people off of. If your instinct is to perceive a lack of capacity, a lack of expertise, a lack of judgment and you constrain them — there's no logical conclusion to that that's any good. You're just going to continue to constrain them. There's a different reaction, which is, “We have a lack of capacity. Let's build the capacity.” 

When I talk to folks on the Hill about this, they say, “Tell me what constraints and mandates we should impose on agencies.” I ask them how the constraints and mandates they’ve already imposed are working. They’re not working, and they know it.

I ask them to consider an enablement and capacity building framework, which instead says, “Obviously you're over-constrained. What could we do to reduce the burdens that you're carrying and increase your productivity? And how can we help you build your capacity?” That's just not in our language yet, but that's the magic wand I would wave: getting people in power to move to an enablement framework.