掌管200万联邦雇员人事的人

Scott, Greg, it is so wonderful to have you both here. Today, we are talking about a subject that I think is top of mind for everyone: making government cool again. We're pleased today to have Scott Kupor, who's the director of the Office of Personnel Management, and Greg Barbaccia, who's the Federal Chief Information Officer. Both of you have very different backgrounds coming to Washington than what we've previously seen.

But we are seeing this renaissance of people in the private sector, particularly people in tech with tech backgrounds, wanting to serve their country and come into government. So, I think I'd start with you, Scott. You were employee number one and managing partner at Andreessen Horowitz. You had long been in the private sector, helping to build companies and investing in companies. What made you say, “I'm done with managing a 600-person organization, and now I'm going to manage 2.3 million people in the federal government”?

Well, first of all, thanks for having us. It's really great to see you again, obviously, as a former partner. I definitely was not running away from a16z, just to be clear, in case Mark and Ben are listening. I love a16z.

It was very simple. I thought it was a confluence of events that made sense. We know we're on this precipice from a fiscal perspective, right? We're clearly spending way more money than we should, and we've got this problem, which I know Greg is well versed in: Technology continues to advance rapidly, and, just to be blunt, the government is nowhere near prepared for it.

Some of that is obviously a procurement issue, which I know Greg and his team are working on. A lot of it is a talent issue. We just don't have the right talent here. So, it was a combination of those things, and an administration that I felt actually made this a real priority for them. We've had incredible support from the president around everything from what David Sacks is doing on the AI side to a lot of the merit hiring proposals that we're doing.

I don't know. It was almost serendipitous: the right time for me personally, and, luckily, the right time for Mark and Ben to be able to say, “If this is what you want to do, go do it.” It was an opportunity to really do some things that I thought were important for the country.

Totally. I think a lot of our tech listeners probably have never heard the term OPM before, so maybe just give us an overview. What does OPM do? And when you were thinking about, as you said, talent is so important, it's something that I think a lot of normal Americans don't think about: the talent in the federal government. What does that even mean? Maybe talk about your thesis and philosophy for how you can help reshape that.

Absolutely. It's not surprising if people don't know what OPM is. As I've said publicly before, I didn't know what OPM was either until the transition team started talking to me about these opportunities. It stands for the Office of Personnel Management. I think the very simple way to think about it is that we are the talent organization for the federal government.

Our job is everything from hopefully bringing the right talent in to making sure we have policies around the hiring standards for government and what we're going to do on performance management. The goal in my mind is very simple: How do we attract and retain the very best people in the country and make sure that we can deploy their efforts on behalf of the American people? That's, at a very high level, how to think about it.

Just to level-set again for the organization, we started the year with roughly 2.4 million civilian employees. We don't have responsibility for the uniformed military, but there are about 2.4 million civilian employees. We've done a lot of work over the course of this year, a lot of which was catalyzed by Elon and the DOGE team.

As a result, we expect that we'll end up with somewhere around 2.1 million employees at the end of the year, so roughly a 300,000 reduction. A lot of those reductions came from the voluntary programs that DOGE put together, these Deferred Resignation Programs and other things. Despite the fact that, obviously, anytime you're talking about people losing jobs, it's a serious issue, the good news is that I think we were able to design programs that gave people the option, if they just didn't want to be part of the go-forward organization, to self-select and volunteer to go do other things.

Absolutely. We'll certainly get more into organizational design and how you remake something that big. Greg, I want to come to you because you have a very different background, having worked at Palantir, served as a veteran, and done a host of different things, but really always focused on technology. What was it that made you say, “Okay, now I'm going to bring this private-sector technology expertise into government”?

Like Scott mentioned, it was the right administration and the right timing. It's certainly in the zeitgeist lately that this is a time for change. This is an administration that wants to lean forward and isn't afraid of upsetting some of the ecosystem that's out there.

So, I saw this as a real opportunity, especially where I'm at in my career, to make some different change. I jumped at the opportunity.

And when you hear the words chief information officer, I think a lot of people probably think very different things about what that could mean. What are your tasks and roles? What does a CIO do inside the federal government?

It is not directly analogous to what you would see in the private sector. I am responsible for technology policy and budgeting decisions across the entire executive branch of the federal government.

One of the big priorities of the administration is the OneGov effort. How do we move forward as one government? A lot of the issues we've had in the past have been where individual agencies do things in silos. How do I galvanize the CIO community to move the administration's priorities forward together as one?

Yeah. One question I always ask people when they come from government into Andreessen Horowitz is what has surprised you most? I always see this culture shock on people's faces. Having left the government after 20 years and coming to work for us, it's a very different culture.

I want to ask you all the same question. Maybe let's start with you, Greg. You've been in seat for how many days now?

About 230-something, so I'm just counting.

It is by days, right? Every day counts. What has surprised you most in those first, say, 100 days? What was the thing you wrote down and said, “I just can't believe that this is how it works,” or, “This was not what I was expecting”?

The compliance and regulatory regime is incredibly complex. Coming from Silicon Valley, where we move fast and break things, you do not have that opportunity in the United States government at this level.

To satisfy some clichés, often things do take an act of Congress, and oftentimes we do need a Fed to make a federal case out of something. That was very eye-opening for me.

Yeah. And, Scott, you've been in seat about 110 days or so right now. For the first 100 days, what's been the thing that shocked you the most?

I want to give a little nuance to what I've observed, at least relative to what Greg was talking about. Maybe this is the same thing: There is no question that there is a different compliance and regulatory environment, for sure. A lot of those rules are there for good reasons.

What was most interesting to me from a culture and people perspective is just this over-obsession with risk. I sat in a couple of meetings in my first couple of days and thought, “Why do we do it this way?” The answer was, “Well, if we don't do it this way, we might get sued, and then something might happen.”

I asked people, “How many times have we actually been sued? How many times has that actually happened?” The honest answer was blank stares. Nobody actually knows, and nobody actually tracks it.

But we've embedded in the culture this idea that risk is a pass-fail thing. You either have risk or you don't have risk, and if you have risk, God forbid you should ever go there. That, to me, was the most eye-opening thing.

I guess I shouldn't have been surprised by it, but it was almost like a cult of obsession. Anything that involved some element of risk—that was the primary thing we were trying to do: mitigate risk.

There was no concept of either how big or how small that risk is, and then certainly no concept of the world that you and I—you live in and I used to live in—which is kind of how unbounded the upside is relative to that risk. That conversation just never happens, and that was super eye-opening for me.

Have you reflected on why that is? I mean, there’s always been the joke that Washington is a town of lawyers, right? So everything is a legal issue. It’s not a business issue; you can’t look at it from any other lens. You’re a lawyer. What is your takeaway as to how we’ve gotten to the point where we never look at upside and only look at, “Let’s just make sure that we don’t fail”?

I think there are a couple of issues. One is, I think that’s right. Look, there are a ton of lawyers that run around here and make a living basically scaring people into worrying about risk. And as a former lawyer myself, it’s the full-employment act for lawyers here in D.C., there’s no question. Some of my best friends are lawyers, so I don’t mean that in any derogatory way.

I think the other thing is that part of it is the political system. If we just forget about the executive branch—people like us—and go to the other side, down this way toward the Hill, the problem is that anytime somebody makes a mistake, the opposite party basically wants to use that mistake as an example for why their entire position is wrong.

The one I love to talk about is that, in our conference room in Menlo Park, we have an old Solyndra PV—I don’t even know what it is, a panel or something—that Marc somehow picked up at one point in time. If you remember the whole debacle over Solyndra, this was in the Obama administration. They did this big loan, Solyndra went bankrupt, and everyone was like, “Look, Obama’s an idiot. This is terrible. The government should never be doing this stuff.”

Now, the government probably shouldn’t have been doing that, either, but the point is, nobody ever said, “Okay, what are the other 50 projects that we’ve done if we think clean energy is important? What are the other 50 things we’ve done?” Yes, some are going to fail and some are going to succeed. Hopefully, some succeed wildly. But there’s no concept of a portfolio.

Everything is a one-off project. When a one-off project fails, whichever party didn’t sponsor it says, “Those guys are idiots. This is why you should never vote for them. This is why you should never trust them.” So I think that’s a bigger part of it.

The other thing is the number of oversight agencies and the amount of activity this government spawns. Greg, I’m sure you deal with it too, but every day I have a report from the GAO, or I have a report from the OIG, or I have a report from God knows who, or some Democrat on my committee who doesn’t like what we’re doing. They’re all well-intentioned, right? They’re all intended to be oversight. But then you’ve created this culture of oversight, which therefore means we should never take any risk because nobody wants to see their department show up in a GAO report, basically.

We’ve just overdone this concept of risk as a way to direct behavior.

Yes.

Totally. I’d love to hear your thoughts on it, too. Where do you see the misunderstanding or miscalibration of risk going wrong in terms of working with technology or bringing technology into the federal government?

Like Scott mentioned, for me specifically, I have more auditors auditing my team than I have team members. That’s not hyperbole.

In the private sector, lawyers are focused on how we enable the business in a way that’s responsible and how we give our leadership a sober analysis of risk. In the government, it is very hard to lose your job and very hard to get fired. One of the only ways that happens is when you get sued and, also, like Scott mentioned, when you get dragged up to the Hill and all the victory laps the party that’s not in power at the moment takes. That encircles everything we do.

And I would just add—and again, Katherine, you know this from the world that you live in today—I’m not sitting here saying anything different than Greg is saying. None of us is saying we should go do the Silicon Valley risk here, right? We need to make sure people get their Social Security checks and that we actually have a national defense system that works.

But we’re so far on the other side that just some modicum of what I’ve been calling “measured risk” is what I’ve been telling people on my team. That’s an objective for the organization. What I’m just trying to get them slowly to say is, “Look, there’s no such thing as 100% on any test. Everything has some risk. Let’s just understand it.”

If it’s catastrophic, and if we get it wrong, and that means the national security of the government is in play, then we probably shouldn’t do that. But the vast majority of things we do don’t involve sovereignty or national security, or whether somebody actually gets a Social Security check.

Yes. No, that is a good point: not everything is a crisis, and this is a town that specializes in crisis.

We love a crisis. Yes.

But you actually said something very interesting. I believe it was on either ABC or NBC, but you said that only 4% of federal employees are ranked below expectations, 60% are ranked at meeting expectations. Going back to this culture of “we can’t fire anyone” or “we can’t actually talk about anyone’s performance,” maybe talk a little bit about how, coming into an organization where everyone is a superstar and no one is bad at their job, you shuffle through, “Okay, these are the high performers and these are the low performers,” given that everyone is ranked on a grading curve, as you’ve said.

Let me give you some new numbers, because it’s actually even worse than what you said. The good news is, it’s worse.

The narrative still plays, right? Just to take one step back, everybody in the government gets ranked—well, not everybody, but most people get ranked—1 through 5 in their annual review. Five is the best score, 1 is the worst, and 3 means you’re meeting expectations. You’re doing what you’re supposed to do.

Fours and fives, so the very, very top, are roughly 65% to 70% of people. When you add threes, fours, and fives, that’s 99.7%. So 99.7% are a 3 or higher. Therefore, 0.3% are a 1 or a 2.

In a 2.3- or 2.4-million-person workforce, you might expect that not everyone is above the median. You would expect that there’s enough of a population to have some normalized type of distribution, right? That’s just not the case.

One of the things we’re doing, at least with respect to what are called the SESs, which are the most senior-level people in government, is effectively a modicum of a forced distribution. We put out guidance that says only 30% of the people can be ranked a 4 or 5.

We didn’t put in any guidance as to what the others are, but we’re really trying to slowly change the culture to say, “You know what? Everybody actually can be measured against one another.” It’s not a competition issue. I gave an example in a blog post I wrote: if you’re a head of sales in a private company and you do 105% of your quota, you probably are going to do better than the CIO in that company who achieved 90% of her objectives.

That’s reasonable and fair, and you all did a good job. Everybody gets a nice congratulations and a nice certification, but that’s not what we’re talking about here. We’ve just got this massive grade inflation.

What’s really important are the 2 things that follow from that. One is the compensation and reward system, which is messed up, because if everybody’s great, that means bonuses just get peanut-buttered out. There’s no differentiation. Promotions just get peanut-buttered out.

You’ve got this opposite problem, which is that the people who probably should be ones or twos don’t actually ever get ranked that way. Some of that is what Greg said: it’s really hard to fire people. You can understand the managerial behavior. If it’s going to take me 12, 18, or 24 months of appeals to do something, I don’t want to do it.

As a result, sometimes people say it’s better to keep a warm body in a job than to go through the process of removing them, which just makes no sense.

One of the things we’re trying to do is create a high-performance culture, because the only way that we attract and retain great people is that great people want to be surrounded by great people. They want to work in an environment where their activities are rewarded and where they can be promoted because of their merit, as opposed to how many years they’ve sat in a seat.

It’s a fundamental cultural change. It’s going to take a lot of heavy lifting to get done, but we’re starting at the margins to figure out a way to incent the right behavior.

What’s been the reaction? In some ways, you would expect that there would be some people who would be shocked by this, but others who would say, “Finally, finally,” as a high performer.

I think that’s exactly right. I hear both things. Look, there’s no question that if you survey the government and ask people—there’s this thing called FEVS that the government conducts every year. We didn’t do it this year just because we missed the timing for it. It’s basically an employee survey, right?

One of the questions they ask in FEVS is, “Does my manager actually manage performance effectively?” In particular, does my manager hold people accountable for behavior? You get extremely low scores on that. So there’s no secret—it’s not a secret. I think most people, if you ask them, will say, “Yes, there are people in my organization who everybody knows”—and it’s true in any organization, right? It’s unfortunately true even in a smaller organization like a16z. Everybody actually knows who the people are who are really carrying their weight, and people also know who the small group of people are who are kind of just dialing it in, basically, right?

So it’s no secret. You hear that from people, and that’s good. There are some people who say this is a good thing.

The other concern I’ve got on the other end is, well, these are all really high-performing people, and so it’s unfair to not give everybody a high grade also. Gee, are we going to create competition among these executives? My answer to all of those has been—on the first question, I was like, look, it’s not unfair. This is the world, and the reality is we’re not all equal in terms of our contributions in an organization. We just need to get over that fallacy and recognize it.

Of course.

And I’m not worried about competition either, because if we want teamwork, which I think we do, then you incentivize people for teamwork, right? As part of your annual review—I know this right at a16z—one of our goals is that we believe teams beat individuals. One of the things that we review Katherine on every year is: Are you working in isolation? Are you leveraging the full strengths of the team to win deals or to do other stuff?

Therefore, your behavior obviously reflects that if you want to be in the top performance. Of course, it’s the same thing in government, right? What’s the behavior we want? Let’s create an incentive structure around it. Let’s actually measure people for it. Let’s put it in the reviews. If we do that, I think we get the behavior that we want.

Absolutely. Greg, I want to come to you on this question, because talent is changing so quickly in the private sector, particularly in tech, around AI and all of these new paradigm shifts that are happening at once. When you think about recruiting those people or bringing those people into your world, and bringing more technologically advanced people into the government, what have been some of the learnings? How do you think about how talent impacts your work?

I think about what incentivizes people to go to high-performing organizations in general. We give them the tools that they need to be successful, we trust them to have the freedom to innovate, and we need to give them access to problems that are sufficiently difficult and interesting to them.

While it’s very hard for the government to compete with Silicon Valley on compensation and things like equity—this is something Scott and I talk about at length—we do win on mission. You get access to some of the world’s hardest problems in the government. You get agency to affect potentially over 300 million people, and you get access to things that outweigh your tenure at the organization—something that, if you went to a large company, would take you perhaps a decade to get to: such valuable work on behalf of the company.

You could come into government and very quickly do something that every citizen and taxpayer would feel. That’s something we’re optimizing on right now.

Yeah.

Yeah. Greg and I have been spending a lot of time on this, and I would just augment a couple of things. One is, look, there’s compensation. We’re never going to solve the compensation problem with the private sector, but as Greg said, I think that’s okay. I think there is a class of people who care about public service and who recognize that this has value.

I think part of the problem we have in attracting talent is, number one, that we don’t sell the right narrative, quite frankly. I’ll give you an example. One of my first meetings when I started here was with somebody on my team who’s a career government person. She said, “Hey, you guys”—meaning, I think, collectively, DOGE—“you guys are destroying the message we’ve always said about government, which is the beauty of government is we attract people because you have, effectively, lifetime employment. You have tenure.”

I said two things. I said, one is, “Look, I hate to break the story to you, but that’s just a myth. There’s no such thing as lifetime employment. Nobody has that. Maybe that was true in the 1950s and 1960s; it’s not true.”

But I said, “Look, that’s totally the wrong message.” The message we want is exactly what Greg said: If you want to come here and solve really hard problems and do something really cutting-edge and, oh, by the way, you want to serve your country, that’s the mission we want to tell young people on the technology side. We’ve never told that story, basically, until we’ve had people like Greg sitting in the seat. Now we have a good cohort of technology people who can do that.

One of the other problems we have with getting engineers here—and Greg and I have talked about this too—is that many jobs in engineering organizations are managing contractors, right? If you’re some super-smart person coming out of MIT Computer Science or something, you actually want to do something cool. You don’t want to manage external contractors.

So we’ve got this endemic problem: We’re missing talent, therefore we go to contractors, therefore we can’t actually recruit talent because the job is to manage contractors. We’ve got to break this mold and actually bring people in. Greg and I are working on an initiative, which I’m very excited about and that I’m very confident we’re going to pull off, to really target how we get particularly early-stage technology people into the government and make them effective at actually solving the most complex problems that we have.

Totally. That was going to be sort of my next question. We’re sort of obsessed with pools of talent in Silicon Valley. It’s like, where’s the next pool of talent, or where’s the next great place to recruit from that other people haven’t discovered?

Are there untapped pockets or pools of talent where you’re saying, across America, these people would be fantastic for exactly what you just described? Maybe they don’t know we exist. Maybe they’re early in their career and they haven’t even thought about it. Or maybe they’re more senior in their career and they just haven’t thought about making the leap into government. How are you getting the message out, and what pools are you attacking first?

Yeah. At least from my perspective—and Greg may have a different view—the pool I’m trying to attack first is the early talent pool.

And there are a couple of reasons for that. One is, as you well know, AI and all this stuff is moving really fast. We need people who are really, really conversant with this stuff. It’s not that people who aren’t early-career can’t do this, but we do know that adoption of technology sometimes takes new people who have very open minds on things.

We want to do that. The other thing is that we have a particular early-talent problem broadly in the government. Just to give you a sense, 7% of government employees are under the age of 30. That compares with close to a quarter of the normal workforce outside of government. On the other side, we have almost 45% of people over the age of 50, which compares to about 33% in the non-federal workforce.

Again, this is not an ageism comment, but if we’re going to lead on AI and we’re going to lead in new technologies, we need people who are probably at an earlier stage in their career, where they’re willing to take some of those risks.

So that’s the first thing I think we need to do: really go after that. The program that Greg and I have been working on, which we’re hoping to roll out soon, is to target how we get early-career people in, number one.

Number two is that we have a technical-management problem also in the government. A significant portion of people in technical organizations are not actually technical, and they’re managing people who are technical. Therefore, they’re just not, quite frankly, able to provide career development and other things that happen.

One of the things I think we can lean on the private sector more for is to help us with that problem. Let’s go to all the great technology companies and say, “You know what? Show us your directors, senior directors, and VPs. Let’s put them on a 2-year secondment, for example, in the government, and have them come be managers of early-career technology teams.” Let’s start this greater kind of connectivity between the public and private sectors.

I think this is the other challenge we have with young talent, at least having talked to a bunch of people: we've created this false dichotomy, which is, you're either a career public service person or you're a career private sector person. And I just think that's a silly dichotomy, right? The reality is, look, most people change jobs all the time, and we should make it fluid, right?

I think there's huge value to the government and huge value to the private sector by having a lot more commingling of that. I would love to have young engineers come here for 2 or 3 or 4 years and, quite frankly, then go to the private sector and realize that the private sector values the experience they have. They value the knowledge. If they want to maximize their earnings, great—go maximize your earnings in the private sector. That's totally fine.

But this idea that at 22 years old, you have to decide, “Am I going to go be a 40-year career civil servant, or am I going to go be a 40-year private sector person?” I just think that's an outdated mode of thinking about employment.

Absolutely. Hey, Greg, I'd love to hear your thoughts on that, because this concept of a tour of duty—I mean, sometimes you talk to people in Silicon Valley and they say, “Okay, if I leave my fast-moving job right now, where I'm seeing cutting-edge work, and then I go into government, there's no guarantee I'll be able to go back. There's no guarantee those years will be the best years of my life, where I've learned something.” What is your pitch, or how do you implement a tour of duty? Have you thought about how you can bring people in for a couple of years and give status to that, where then they'll be very attractive to the private sector coming out?

Absolutely. I think the new-grad and younger-talent idea is phenomenal. I've been thinking a lot lately about people who just had perhaps their first exit or had some success for a few years in the private sector. Come to the government and see how you do, because it is a totally different operating environment.

Everything you do after working in tech in the government will be easier, I assure you, and you will learn how to operate in environments of opacity that you probably don't see in the private sector. I wanted to touch on something Scott mentioned: this technology-management problem. This is a very real problem.

First, we have it from a talent-screening perspective. The people who are screening resumes for technologists in the government perhaps don't have the background to know exactly what we need, what exactly we're looking for. Oftentimes, they're the same people screening resumes for a bunch of different roles under their purview.

This also translates to contracts. People ask, “How do these big contract sprawls happen? How do we get all these tech projects that don't make any sense?” If you walk through the White House complex, it looks like a suburban dealership. There are Suburbans everywhere. But every American has an intrinsic understanding of how much a Chevy Suburban should cost.

If I were to hand you a contract for a fleet of Suburbans and they were $900,000 each, you would say, “This is objectively insane. That is not how much a vehicle costs.”

But we are missing that crucial talent in the government. This is why the people who are charged with approving contracts oftentimes do not understand the technology problem deeply. They don't understand what an appropriate solution is, how long it should take to solve it, how many people we need to solve it, or what type of person we need to deploy to the problem.

If we're speaking about contractors, this is how these large Beltway bandits, as they call them, or systems integrators, get these indefinite-delivery, indefinite-quantity contracts that don't make any sense when you're trying to solve the problem. These middle managers coming from tech are really interesting. We need the people who understand the roles and what they're doing to help adjudicate the talent that's coming in and also the contracts. Otherwise, the sprawl is just going to be an indefinite problem.

Greg, if I could add one thing—Greg knows this, but, Katherine, just for you and the listeners, as Greg mentioned, this all starts even at the very front end of evaluating talent. So we literally have, again, people who are nonfunctional experts evaluating people's resumes, and their evaluation to date, until very recently, has been—I would call it—self-assessment.

The candidate says, “I have the following skills,” and the HR person says, “Great, those are the following skills.” It turns out these people are smart enough. It's not by accident that their skills exactly match what's on the job description. But we do not actually verify those.

We don't do a software-coding test to hire a software engineer. You would probably fire one of your CEOs in any of your companies if you found that they were hiring engineers without actually assessing their technical talent for real.

There's a historical reason for this. Forty years ago, we used to have a civil service exam. It was basically taken away at the very end of the Carter administration and the beginning of the Reagan administration because it was deemed to have disparate impact. There were lower scores for minority candidates on it, and the government entered into a consent decree and said, “We will henceforth no longer do these things.”

That literally sat on the books for 43 years until, luckily, one of the first things we did in this administration was exit that consent decree. We've now given guidance as part of our merit-hiring plan that everybody who's going to apply for a job in the government has to have some functional assessment done.

If you're a software engineer, we're going to give you a coding test. If you're applying for an HR job, we're going to figure out the right way to actually test your skills. So you've got this compounding problem: you have people hiring people who don't have domain expertise to evaluate them, there's no check on that, and that just compounds the problem Greg's talking about.

Those people get promoted, and now you have nontechnical people sitting in managerial roles. We have to dismantle each brick one by one in order to actually solve the hiring problem.

Totally. It's interesting, because what you're talking about is, in many ways, an opportunity with AI. I'm going back a couple of years, and this was at Andreessen Horowitz, but also, I think, across Silicon Valley: there was the earth-shattering moment of ChatGPT's release in late 2022. I think everyone woke up and said, “We have to completely reorient our organizations, completely change our thinking, because AI is the next 10 to 15 years, at least, of how we're going to be thinking about building companies.”

And it's 3 years later. It's not too late for the federal government to have that moment, too. But how do you even evaluate these companies, as you said, if you're someone who's never worked with AI or never even really understood the term? How are you going to know which companies to work with?

Maybe talk a little bit about the tactical way that you reorient a culture to say, “Okay, AI is going to be very important for the federal government. We're going to have to learn new things. We're going to have to hire different types of people. We're going to have to work with different companies where maybe we can't pronounce the names as well.” Just talk about how you tactically implement that culture for what you're doing.

Certainly. The government loves process and ritual. We're slowly moving toward, “Hey, we don't have to do these same mundane, repetitive tasks because we could automate these things.” This goes back to the early days of Silicon Valley: process versus product.

In the government, oftentimes, you celebrate the fact that you checked all the boxes and completed a process, disconnected from the reality of whether this moved the needle for whatever outcome we're trying to drive for our mission. Reorienting the culture of employees in the government means saying your success at your job is now based on the outcomes we're driving for the mission and being focused on that.

We're going to use this technology to change what you were doing all day. Instead of just going through these tasks, we're now going to automate everything that we can do in whatever way is responsible to the mission being driven. Now you're going to do higher-value work for the taxpayer.

What's important is that we did this in the past. We moved from budgeting from ledgers to Excel, and eventually we moved away from Post-it notes to email. Instead of doing all these things physically, we did them digitally. Now it's time for another modernization.

We have to get people comfortable with their roles intrinsically changing because of this technology, in a way we haven't seen in at least a few generations.

Yeah, I think we have to attack it from a couple of angles. One is—and huge kudos to Greg, David Sacks, Sriram, all the people, and the President for his leadership here—we now have at least a top-down directive. The President, through that AI Action Plan, basically said, “Look, this is a critical agenda for the US government.”

For people like Greg, who get to talk across all the CIOs, he at least now can say, “Hey, look, the President's been very clear: this is an important agenda item.” We can have his leadership do that, so I think that's awesome.

At the micro level, I tell you, in our organization—we're a tiny organization, we're like a couple thousand people.

You have to just slowly change the behavior. One of the things we did, literally—believe it or not—until this week was that we did not have ChatGPT on anybody’s actual desktop. I’ve been pounding my CIO and pounding my CEO, and we finally got it, thanks to the new deal that GSA did—the $1-a-year deal. I’m sure Greg was part of that, too. We now have ChatGPT. Great.

What I’ve told everybody on my team is: It is your obligation to learn this. You can pretend that it doesn’t exist and put your head in the sand, and that’s fine, but I’m not going to be responsible for that, because the technology is going to advance whether you use it or not. The question for you as an employee in the government is: Do you want to be modern? Do you want to find ways to implement this?

I’ve said, look, just find little things. What are you doing in your job today where you can get a 5% incremental improvement in your efficiency because you queried something in ChatGPT? We do a ton of rule and regulation writing; everybody in the government does that. What a great use case for AI. Again, we’re not going to have AI write the rules, but AI can do information dissemination. It can help you understand: Is this really statutorily required, or is it not statutorily required?

There’s little stuff like that where I’m like, there’s no risk in that; it’s okay. The good news is almost everything the government has is public information anyway. It’s not like we have to put PII into all these systems. We just have to start to change behavior, one person at a time.

What I worry about is—I think Greg and team will help us with this—but if we say we’ve got this big agenda to get AI, then people are going to build these 5-, 10-, 15-year plans, and quite frankly, nothing will ever happen. Then you can go and do 15 projects, but the traditional mode, as Greg knows, in government is that everything should be top-down—these massive white papers and all the other stuff. I don’t think you adopt new technology through white papers. You do it through actual individual user adoption and just understanding the culture even deeper.

Is your sense that people use ChatGPT for recipes and travel planning already, and they’re like, “Oh, great. Now we get to use it in our job, and we’ll figure out new ways to use it”? Or is this—there are a billion users on ChatGPT globally—is the federal government like, “What is this? How do I use this?” Are there going to need to be training programs on how to do prompts? Maybe talk through if that’s part of the culture shock as well—that people actually don’t use LLMs or haven’t been using them in their lives for the last few years.

I don’t know if I can give you a quantitative answer. My gut is that the world that we all live in every day is still very different from the world that a billion people live in. But we know there are 7 billion people in the world or something, so there are still 80% or more of people who don’t actually use these things. My sense is there’s a little bit of that.

Training is a real issue. I’ll give you another great example. When we first rolled it out—or were about to roll it out—our team wanted to literally put a 3-page memo together on how you should use it and what the risks are, and we were literally going to have people sign this acknowledgment. I was like, “No, time out. That’s the last thing we’ll do to ever get people to adopt it.”

How about you tell me the 3 bullets that you care about? Don’t put PII in this thing if you’re worried about it. Don’t just take its answer for granted, because it might hallucinate. Those are probably good things we should tell people, right? But we’re going to put a 2- or 3-bullet email out, and no one’s signing a document, right?

Number 1 is, again, it goes back to this risk issue: There’s not that much risk in people using ChatGPT, quite frankly, in government. Number 2 is training. The other thing we did is we said, “Okay, great. Here are a bunch of resources.” There are a ton of resources. Coursera has a bunch of free resources, and OpenAI has a bunch of resources.

We just said, “Hey, you know what? Go schedule it with your manager, but everybody—you’re entitled to 2 hours a month of training using all these free resources. You can take time out of work; just clear it with your manager. But we’ve got to give permission for people to actually do this kind of stuff.”

I don’t know the quantitative answer as to whether they’re using it in their private lives. My guess is a lot of people are, but I think they’ve also been told—it’s been a slap on the wrist—which is, “Don’t do this in government because this is bad. It’s dangerous. It hallucinates. Information will leak.” All the kind of parade of horrors that will create a GAO report that nobody wants to respond to, basically.

Yeah. One of the frustrations with attracting talent is, “Why would I use something all day and depend on it in my personal life, and now I can’t use it at work? Why would I want to come and do things in an antiquated fashion?” That’s something we’re thinking about, too.

I can tell you, the people on my team who came from Silicon Valley and are working in my department would just use ChatGPT on their personal phones to help them do stuff. Now, at least hopefully, you can actually use your government laptop to do something that is not nefarious activity.

“Swivel-chairing” is the term? Okay, that’s good. I had heard that. Okay, that’s great. That’s a good one.

So I want to ask one final question, and then we’ll go into a quick lightning round. One of the things I see when we talk to people from our world or from the private sector who come into government is that they’re working 20-hour days. They’re working constantly, right? It is a race against the clock in many cases, where there’s an overwhelming amount of work to do.

A lot of people see it as, “I have 2, 3, 4 years to make my impact. I’m going to do as much as possible, and hopefully I’ll get a little bit done,” right? But it’s such a big world; it’s a big bite to chew. There’s such an overwhelming amount of work to do.

When you look ahead a couple of years from now, how are you going to measure yourselves in terms of coming from the private sector and having an impact? What are the things where you say, “If we get X, Y, and Z done, I will have considered my tour of duty to be extraordinary, and it will have changed the course of my organization”? We’ll start with you, Scott.

Yeah, I’d say for me, there are really 2 things, and these are my primary goals for our organization. Number 1 is I want operational efficiency to actually be a first-class metric in the government. You probably know this, but today everybody’s power in government is measured by how big their budget is and how many people they have. Again, we get the exact behavior we incentivize, which is that everybody wants to grow their budget and grow their people.

What I want people to do is say, “You know what? We’re stewards of taxpayer dollars. Every single dollar we spend is someone else’s money.” We have this acronym inside of OPM: OPM stands for “other people’s money.” That means your job is to deliver a high-quality service, but your job is always to look at what continuous improvement you can make.

We should actually reward people for doing high-quality service at lower cost, as opposed to rewarding people who do the service by growing their budget, basically. It’s very simple.

Number 2 for me goes directly to the talent conversation we’re having. It doesn’t need to be exclusive, but if we can get early-career people—whether in tech or otherwise—we have, as I said, a youth problem broadly across government. Anybody who wants to be a tech person, a financial analyst, or an HR specialist—if we can get early-career people to say, “Going to government is awesome. It’s a great mission. I tackle really hard problems, I learn, I develop my career, and, oh, by the way, if I want to then go to the private sector, the private sector will recognize and reward that service,” then I think—look, I would be the most optimistic person in the world.

We will have the most amazing Pax Americana for the next 50 years, basically.

Absolutely. I think it’s an amazing opportunity now, too, for young people. You look at the youth unemployment rates and what’s happening, even in the private sector, where it does seem like it’s much harder for college students to find their first job now. If they view it as a 2-, 3-, or 4-year career and then they can go into the private sector with some skills, that actually feels like a major opportunity to bring young people in. I’d love to hear your thoughts as well.

Absolutely. Making work in government prestigious to private-sector employees and employers, and focusing on the taxpayer and citizen experience, are my technical goals. The web of government websites is insane. That is a passion project of mine. We're taking this seriously.

The concept that you have to tell multiple different government agencies the same thing about you—your birth date, your address—in a bunch of different places, that's crazy. A lot of other countries have solved this in a first-class way. So we're thinking about how we could responsibly implement consent-based, citizen-centric data sharing between agencies, which would lead to a portal you could go to and see everything in one place: my benefits information, my tax return status, when my passport expires. This is one government. It's all the executive branch. Let's serve things to the taxpayer in a way that makes sense for them.

That's very important to me. Also, just going back to this concept of one government: with my background in intelligence and my background in Palantir, how do we take a bunch of disparate data sets and use them to drive decision-making? Right now, we have a bunch of silos. The president signed an executive order about breaking down these silos. So using data to drive decisions and turning it into intelligence, we could really make these intelligence- and data-driven decisions. That's something I'm focused on.

Okay. It would be great to end with a lightning round. A myth about government and tech that you would like to retire, or a piece of government tech that you would like to retire. Scott, let's start with you.

Look, I think the myth that I'd like to retire is that the government actually cannot and should not be a leading innovator in the deployment of technology. I just think it's silly. Again, this whole risk mindset has totally warped everybody's views.

Yeah, right.

The myth that the tools you'll get are terrible: I have a great laptop. It's fast, the battery works, and it lasts a long time. I have an iPhone, so you're not going to be on some antiquated hardware.

Yeah. So we've talked a lot about this, but what's a small process or policy tweak that you think would have the greatest outsized impact?

I think it's very simple: just get people to forget about risk as a pass-fail thing and get people to think about risk as a spectrum of things. Also, as we talked about, get people to think about what's the upside benefit of what we're doing.

I think we'd make a big difference.

Rethinking the compliance requirements that came out before our computers existed.

One metric you track that you think others should.

I think it's operational efficiency, but—

Rework percentage: how often are we doing things over and over?

A book or resource you'd recommend to new builders in the public sector?

My favorite is still Only the Paranoid Survive. I think that's still applicable to everybody, whether private or public sector.

The Goal and The Phoenix Project, taking manufacturing lessons and applying them to technology.

And then, finally, finish the sentence: The US will win the future if we blank.

Win the AI race.

Remain undistracted.

Awesome. Well, Scott, Greg, thank you so much for taking the time to talk to us about making government cool again and building for the public sector. We so appreciate the service that you're doing for our country. Thank you.

Thank you for having us.