a16z首位员工谈a16z的历史、风险投资与政府改革｜Scott Kupor

Tell me what it was like starting a16z with Ben and Marc.

It was interesting. Let me give you the backstory. We’d sold our last company to Hewlett-Packard in 2007, and a bunch of us had basically been in indentured servitude to HP for 12 to 18 months. Then it was the summer of ’08. You know your history: in March of ’08, Bear Stearns basically disappeared. So it was the summer of ’08, and Marc and Ben were like, “Yeah, we’re gonna do this.” They’d been angel investing with their own money for a long time, and they said, “Look, it’d be a lot more fun if we institutionalized this and went out and raised money.”

We were literally having this conversation in the summer of ’08, and then in September, of course, Lehman went bankrupt overnight, and the world seemed to be coming to an end. I went home one day to my wife and said, “Good news, I’m gonna quit my job at HP”—probably the most stable job you could have—“and join these two guys.” We were living in North Carolina at the time because we’d acquired our business out there, and Ben and Marc had asked me to move out and run it. I wasn’t even in California anymore.

It took me a long time. I hemmed and hawed, and these guys joke to this day—I don’t know if it’s true or if they just like to razz me—that if I’d agreed to join them sooner, it would’ve been Andreessen-Cooper Horowitz instead of Andreessen Horowitz. I don’t know if that’s true or not. My official title was first employee, not co-founder of the firm.

They started, and then finally, around Christmas, I still have these emails and look back at them. Marc and Ben were like, “Look, you gotta get off the dime. You gotta figure out what you’re doing.” I finally said, “Okay, I’m gonna go do this.” But it took me a long time to figure it out.

What was the job description? What’d they say?

There was no job description. That was part of the problem. My wife was like, “What are you doing?” and I was like, “I have no idea.” Remember, it was just the two of them, and the job description was actually Marc and Ben saying, “Look, we’re gonna invest, and you need to figure out everything else we need to do to build a firm.” I didn’t know what that was.

In the early days, it was literally going to Costco to buy furniture, making sure we had food in the office, and doing things like opening bank accounts. Then we closed the first fund. The way the funds work, as your listeners may know, is that investors sign up for what’s called a commitment. They say, “Great, I’m gonna commit $10 million to a larger fund,” and then we call the capital as we go. It’s literally called a capital call. You do it when you decide to invest in a company and actually need the money to make the investment.

On my very first day on the job, they were like, “Great, we made an investment in this company. Can you please call capital from our LPs to get the money?” I didn’t even have an email yet, so I used my personal Gmail. I sent out this notice to about 50 LPs: “Great news, wire me $1 million, basically, because we need money for this thing.” I immediately got an email back from one of our LPs, who I won’t name, saying, “Who the fuck are you? Why are you sending me this email?” Then he called Marc and Ben and said, “Somebody’s impersonating you guys. Somebody’s literally trying to shake me down for money.”

I finally realized I probably should get an a16z domain set up. That was the impetus I needed to get an a16z domain so I could do stuff. We were figuring it out as we went. These guys knew what they were doing because they were investing, and they knew what their primary job was. For me, the whole goal was figuring out how to create a firm, so most of my initial days were spent hiring people. We had to hire finance people, for example, so I wouldn’t do stupid things like send capital calls to people from my Gmail account.

We always had this theory, which has panned out: How do we build a services component to the business? How do we create a system where we can provide everything an entrepreneur might need, and develop concepts around networks and relationships—talent networks, business development networks? A lot of the early days were spent hiring people who were domain experts in those areas so we could start to build this team. We went from the three of us to close to 100 in the first year, and today I think we have about 700 people at the firm. It’s evolved over time, for sure.

Was Michael Ovitz around a16z at this time?

He was. Michael had been on our board at Loudcloud and Opsware, so that’s how Marc and Ben knew him, and that’s how I got to know him. He was very influential, as you probably know. I’m sure Marc’s told the story about the founding of the firm. Michael was very helpful, so I’d call on him a lot and say, “Look, this is what we’re thinking about doing,” and he’d recount how they did things at CAA.

Essentially, what they did—and a lot of what a16z was built on—was disaggregate the job of what was then the agent. We basically did the same thing for the general partner job. In the old days, and we pitched this at our first LP meeting, we were basically trying to institutionalize John Doerr. For listeners who may not remember John Doerr, he was one of the very, very famous founders of Kleiner Perkins. You’d go to John, and he knew every salesperson, every CFO, and all the political stuff. He was the ultimate connector for entrepreneurs.

Our basic idea was: Could we institutionalize that and not have every GP have to learn all those things? Could we disaggregate that from the GP job and hire professionals to build talent networks, political networks, sales and marketing, and all the other things that ultimately became the platform for a16z?

Was this around Michael Ovitz’s idea of a phalanx? Before that, you had your agent, but really, you had all of CAA—

Yes. Yeah.

—and you essentially tapped into every single resource.

Yeah. His theory—and I know you’re friends with him; you’ve had him on many times—

I’m having dinner with him tonight, as a matter of fact.

Okay. Tell Michael I said hello.

Yeah.

I haven’t seen him in a while. He’ll tell the story much better, but my recollection is that they started CAA in 1975, and the whole industry was very similar to how the venture capital industry was: you had an agent. You’re Robert Redford, and you have an agent who comes to you and says, “Great, I’m gonna help you do this. I’m gonna help you get movie deals, book deals, radio,” whatever the opportunities are. That was great, but the model was basically one-to-one: your agent to actor.

What Michael did at CAA was say, “Look, we’re gonna maintain the relationship context between agent and actor.” When you showed up for a pitch at CAA in 1990, when I think they had about 90% market share, they’d sit around a table, and Michael would introduce you as Robert Redford and say, “Look, meet my colleague who does nothing other than radio deals. Here’s my expert in radio deals. Here’s my expert in international. Here’s my expert in books.” So you had this idea that you were surrounded—you had a phalanx, maybe is the right way to think about it. You had the whole firm, as opposed to an individual who was responsible for your career.

We basically took the same model to the VC world. The VC world was a lot like that: you had a one-to-one relationship with John Doerr, who was fantastic at what he did. The problem was, number one, it just wasn’t scalable. At some point, John has so many board seats that you time out from a GP perspective, because you can only take so many board seats. And second, you had to get everybody to invest the same amount of effort in relationship development across all these domains. That might have worked in the days when John Doerr did it, when we basically invested in enterprise software companies. That was essentially the domain of VC.

But if you believe, as I do, that Mark’s software “ate the world” or “is eating the world,” then as the domains grew, the expertise needed just continued to grow. The fundamental theory Mark and Ben had was: Let’s hire GPs who are functional domain experts in the particular areas they want to invest in, and then surround them with this phalanx of people. Have someone who all they do all day long is manage CFO relationships or CEO relationships or relationships with VPs of engineering; someone whose entire job is government relations, business development, or other things like that.

So again, our pitch to an entrepreneur when they came in—we used to do this thing we called the reverse pitch. An entrepreneur would pitch us, and then we would bring the team in and do the reverse pitch to them: “Hey, look, if you work with Andreessen Horowitz, not only do you get me, Ben Horowitz, as your board member, but you get this entire table of people whose job it is, day in and day out, to figure out how to institutionalize a set of relationships and networks that’ll make you successful.” That was the key application of what Ovitz did and how Mark and Ben adopted it to the VC world.

Before you left a16z to go to government, how many years were you there?

16 and a half years, from 2009 until—

What are some of your most memorable stories from the very early days, the first year or two or three?

Well, there were some fun ones. We had a little bit of our Mad Men episode at the firm, where in the evenings we would sit around and drink bourbon. This was when Mark used to drink. I think he doesn’t drink anymore.

No, he said that. I think he was on this show earlier.

Yeah, yeah.

And I think he said the best day was 12 hours of caffeine followed by 6 hours of alcohol or something.

Exactly. He was great, right? Mark would drink a lot of coffee. He always had sushi for lunch. Mark gets very deep into things, so if he gets excited about sushi, quite frankly, he’ll eat it for every meal for 6 months at a time. We would sit around his office at the end of the day and drink. I remember very distinctly that we had to buy a special ice cube maker that made the square ice cubes.

Yeah.

And Arsho, who’s his assistant, was incredibly generous with us. She would come in at the end of the day and make sure the ice machine was right. So that was fun.

Hold on. You gotta bring me—

I have no idea what we’re doing.

So let’s get over this. Bring me to this. Okay, so Mad Men style: we’re working all day. Now we’re hanging out—

We didn’t drink during the day.

No, of course.

So I’d say Mad Men style is maybe overrated because—

No. So then—

Nobody was smoking cigarettes.

So now it’s nighttime.

Now it’s nighttime.

Nighttime.

Yes, it’s evening time.

Workday’s kind of over.

Right.

But we’re gonna keep it going, and we’re gonna drink.

Exactly.

What’s happening?

Right, yeah. In our original office, and actually in our new office too, Mark and Ben had these very spacious offices, which was very nice. Mark literally had a bar. He had one of the most amazing bourbon and whiskey collections I’ve ever seen. He probably still has it, although he’s been generous with my birthdays over time. He sends things to me now since he’s less of a drinker, so I pick up most of his stuff.

We would sit around and talk. It was the one part of the day where there was no formal agenda. Sometimes we just bullshitted about whatever was going on; sometimes we were talking about an investment or something in the company. It was fun. For someone like me, even though I’d been around Ben and Mark for almost 10 years already, they’re the kinds of people where, if you have time to sit down with them and just shoot the shit outside of work hours, you always learn something.

It felt like we had this sense of camaraderie, that we were trying to build something. There was no agenda. Sometimes we’d sit around for 15 minutes and then we’d all disappear; sometimes we’d sit there for an hour. It never got raucous. We all had families and kids to go home to, so it wasn’t like we were going from there to a bar somewhere else. It was just a relaxing way to end the day and have some fun or intellectual conversation about the things we were doing.

What do you think those versions of you three back then would think about the outcome that played out over the next 16, 17 years?

When I did the math for myself and was trying to convince my wife why, in the middle of the global financial crisis, we should pick up our family from North Carolina, go back to the Bay Area, and do this thing, my math was: “Okay, I’m pretty confident they’ll raise the first fund.” I had no idea, quite frankly, what would happen after that. My general calculus was that, worst case, these funds would take a long time, we’d weather the financial crisis, and these guys had always been very good to me in terms of taking care of me. I thought, “Look, I’ll find another job if this thing doesn’t pan out.” That was my general back-of-the-envelope calculus.

I never imagined we’d get to where we were. Nor did I imagine we’d get there as fast as we did. Mark’s talked about this, but in literally the first couple of months after we closed the first fund, we invested in Skype. If you remember, Skype was part of eBay. It was bought by eBay and completely neglected inside of eBay, and yet amazingly, it had this incredible network effect. If you look at the numbers, the number of subscribers just kept growing and growing.

We joined with Silver Lake at the time. I don’t know if you know Egon Dourvan, who runs Silver Lake.

Mm-hmm. Michael Dell—I just saw him a few nights ago.

Yeah.

He speaks very highly of him. He actually brought him up the other night.

Exactly, yeah.

Yeah.

And Michael Dell was part of that too. We essentially did a buyout, which was a crazy thing to do. We had just gone on the road and told all of our investors we were doing early-stage venture investing. We hinted that this Skype thing was going to happen. We pretty much said it was going to happen, but we couldn’t say it was going to happen because it hadn’t been done yet.

It wasn’t actually a surprise for LPs, but it was clearly very different from the strategy we had laid out. So we did this thing, and immediately, of course, everybody attacked us and said, “These guys are crazy. What the hell are they doing doing a buyout? It’ll never work.” There was all kinds of stuff at the time. If you go back and read it, you’ll see. There was this big IP dispute with the original founders, like Janus and Nicholas, and eBay about who actually owned the fundamental underlying IP.

And so a bunch of people looked at buying Skype, but nobody could get comfortable because they were like, “Gee, if we buy it, what if Janus and Nicholas come out of the woodwork and say they own it? Do you really own the IP?” Egon’s genius was to figure out a way to work with these guys. Our bids came together, and we worked collaboratively with them as a way to moderate this IP question.

But anyway, everyone was like, “Look, this is crazy.” The beauty of it was that literally 18 months later, we sold to Microsoft for 4 times what we paid 18 months earlier. The news cycles were unbelievable. Part of this business is that momentum begets momentum, which is true of so many businesses.

We were able to raise our second fund well ahead of what we had planned, in part because the Skype thing worked out so well. We returned literally half of our fund to our LPs within 18 months, which these days is completely unheard of. Even back then, it was unheard of, and it completely changed the narrative. We also had a company called Nicira, which was in the networking space. Martin Casado, if you know Martin, is one of the partners there today. His company got sold to Cisco early on in the first fund.

We had these very quick early wins, which potentially could have gone the other way, but they went in our favor and dramatically accelerated our progress. Key to our business model—to do the Michael Ovitz business model—we had to have a big enough fee stream to hire all these other non-investing people.

Yeah.

Mark’s talked about this before, but what Mark and Ben did was very magnanimous. The two of them took no salary at the beginning of the company because they wanted to take as much of the fees as they could and give them to me to hire other people and build this thing out.

To continue that build-out, the second fund gave us a stacking fee stream that allowed us to say, “Okay, now we can double down on this strategy.” We had early success. It was also 2009, in the middle of the global financial crisis. You know this, but the Sequoia guys put out this “RIP Good Times” presentation. Have you seen this?

Was it “RIP Good Times”?

“RIP Good Times,” right. It was literally like the world was coming to an end. We were all gonna die. It sounds a little bit like today, actually, now that I say it. No one would ever raise any money. Put all your money under your mattress.

We were this shiny gold object in 2009, when everyone else was hunkered down and worried about their portfolio companies. We had an enormous advantage: We put a shingle out and said, “Hey, we’re open for business.” Every other venture firm—and rightly so, I understand it—had an enormous number of portfolio companies and was very worried about whether they would survive.

So much of that early momentum accrued to our benefit. It enabled us to raise the second fund, which gave us the funding to build out the services strategy, and that really compounded in a way I don’t think we expected. Mark and Ben may tell the story differently: They always knew it would be a success. I believe that. These guys are entrepreneurs for a reason.

Obviously, entrepreneurs can be self-confident, and they don’t wake up thinking they’re gonna lose, but they’re always surprised. There’s no fucking way—

Yeah.

—if you build something that—

Yeah.

They’re confident they’re gonna succeed, but they—

Yeah.

—can never predict the degree to which they do succeed. It’s funny that you say they might tell the story differently, because one thing I love talking about with Mark—and it’s obvious if you watch the episode he and I did together—is the amount of business history that’s in that guy’s head.

It’s unbelievable.

I’m constantly prompting him, and it’s like he—I don’t tell people, you know this, we don’t know what we’re gonna talk about before we talk about it.

Right.

There’s not a list of questions.

Right.

I surely didn’t say anything to him other than that we’re both psychopathically obsessed with business history, entrepreneurs, and all this other stuff.

Right.

It’s impossible not to have an interesting conversation. But I’m like, “Damn it, I should get Mark on.” Instead of telling the history of other businesses, I should get the history of a16z from him while you’re speaking.

Yeah. Yeah.

It’s amazing, and as you know, the guy is encyclopedic on so many things. That’s the other thing that he and Ben were really smart about: the concept of marketing in the venture business. It was considered verboten to market in the venture capital business. I don’t know why that was. It was probably mostly historical, because in the old days, capital was scarce.

The source of power for venture capital firms from the late 1960s and early 1970s until probably close to the 2000s was that money was scarce. The VCs had the money, so the balance of power was with the VCs, and the entrepreneurs were down here. That was sufficient. A lot of great firms, many of which still exist today, were incredible in that era. Sequoia has been and continues to be incredible.

All the existing firms at the top of the game—Sequoia, Accel, Greylock, even Benchmark, which was a spinout of Merrill Pickard—all lived in that era when nobody marketed. I believe that was partly because they didn’t have to. You went to them because they had the money and you needed the money. That was the marketing.

When we came along, we didn’t know anything, and we didn’t know anybody. Mark and Ben rightly said, “We came out of the enterprise software business, and we know that in the enterprise software business, just because you have a good product, nobody gives a shit if you don’t tell them what it is.” They said, “Look, we’ve got to tell people who we are and what we stand for.”

So we told the story: We were going to do things differently. We were going to build this whole service as part of our business, and we were going to market on that basis. We didn’t have a long history of a track record in the business. To this day, if you go to Sequoia, which is an amazing player in the space, you’ll see posters of Cisco, Apple, Google, and all these other companies in their conference rooms. We had none of that.

Mark and Ben had done their own angel investing, and of course the two of them had had success as entrepreneurs, but we couldn’t compete against that. What they rightly did was say, “Look, if we can’t compete on what people are competing on today, we have to change the rules of the road and compete on something else.”

They created this category: The new venture capital firm wasn’t going to be just capital, but capital plus. It was going to be capital plus a whole network of people whose job it is to help the entrepreneur—

It’s in service of the entrepreneur.

That’s exactly right. That was a huge difference in what they did, and we got a lot of shit from all the other VCs on Sand Hill. They were like, “Look, it’s verboten to market,” and all this other stuff.

I remember talking to all of our LPs every day, and they’d say, “Oh, so-and-so, we heard this from one of your competitors.” It was always, “Look, these guys talk a big game, and they tell people stuff before they’ve done it.” We were doing exactly—

That sounds like an entrepreneur to me.

Yeah. We were doing exactly what any entrepreneur would do: We were trying to create a category. You have to tell people about the category; you have to tell your narrative. And look, we had to deliver on it at the end of the day, which we did.

It was such a unique way to approach the business—not just the foundations of the business model, but the idea that we were going to create a marketing strategy around it. That enabled the firm to grow at a pace I don’t think any of us ever anticipated.

So it sounds like at the beginning you were much more operational.

Yes.

As your tenure continued—

Yeah.

—did you switch over to investing, or were you always in operations? How did you do it?

I did, yeah. When I first came, I was like, “What do you want me to do?” And they were like, “Well, we’re going to invest, and you’re going to do everything else.” I have to admit, my ego was hurt for a little while. I was like, “Okay, well, why can’t I also invest?” What I came to realize, particularly because at the time they were doing early-stage investing, is that I’m not a deep technologist like these guys are. So much of that game, I believe, is your ability to actually understand the technology at a much more granular level.

We sit in pitches all day long, and if you don’t understand databases, for example, you can hear 10 pitches about database companies, and unless you have that depth of knowledge, the honest answer is you have no fucking clue which one is the right one to invest in. I ultimately assuaged my ego by saying, “Okay, look, that’s not my strength, and so I need to focus on the stuff that I think I can do.”

What happened over time was that we changed the nature of the firm. This is another thing I heard Ben talking about on another podcast not too long ago: We started with a centralized firm, so everything reported to me. We grew that for about 10 years, to about 500 people or something. All of the operating functions reported to me, and even the investors essentially reported to me. Mark and Ben didn’t really have any direct reports.

What they realized over time—which was brilliant—was that you can’t scale a venture firm in that fashion. They said, “Look, we can’t have 20 people sitting around when we’re doing software eating the world, and we’ve got crypto and healthcare and bio and American Dynamism and all these other things going on. You have to specialize.” So, 10 years into the firm, we started a process where we decentralized it and said, “Look, we’re going to have smaller investing teams that are domain-specific. We’re going to have a crypto fund that Chris Dixon runs, an American Dynamism fund that David Ulevitch runs, and an AI infrastructure fund that Martin Casado runs,” and so on.

Then we said, “Look, in order to do that right, the needs of the entrepreneurs are different. All of these ancillary operating services we built also need to feather into these different verticals so that we can do the things that are appropriate for the entrepreneurs.” What a gaming entrepreneur needs in terms of business development support is totally different from what a life sciences entrepreneur needs. So we went through this process of disaggregating everything into these structures.

We kept a few things centralized that made sense, like legal and finance—core compliance functions. You don’t necessarily want people doing those separately on their own, so let’s keep those centralized. Investor relations, too: from an LP’s perspective, the view is, “Look, I want to interact with Andreessen Horowitz as a firm. I don’t want to have to build completely independent relationships.” So we kept a few things centralized.

What that did for me—and I’ll admit, Mark and Ben know this—I fought it initially. When you go from owning everything, you’re like, “Wait a second, I suddenly feel like my domain got a lot smaller.” Ben talked about this on a podcast, and he said exactly what I felt: Anytime you do a reorg in a company, you always get a lot of hurt feelings, because by definition, somebody in a reorg has fewer things than they used to have. That’s just the nature of it.

It took me a while. Again, this was another time my ego got hurt. I was like, “Wait a second, I thought I was doing a pretty damn good job managing these things.” Mark and Ben, to their credit, were like, “Look, we’re not saying you’re not doing a good job. We’re saying the needs of the business have changed. The needs of the business demand that we do this, so let’s go find other ways you can add value to the business.”

That was when I finally got an opportunity to start doing some investing. What I recognized again was, look, I can’t do early-stage tech investing because I’m just not a domain expert. So we started a later-stage fund. We call it Later Stage Venture, for lack of a more creative word, and we started that in 2019. We brought in a guy named David George from GA, General Atlantic, who’d been doing this and was great.

I basically started working with David and carved out more of a role doing stuff on the healthcare side. By the time I left for government, I had invested in about 15 companies, about 10 of which I was sitting on the board of. I was doing stuff where the evaluation of the company was more about the operational successes and the opportunity for the company from 1 to infinity, as opposed to the zero-to-one side of investing, which I just didn’t feel like I had the right skill set to do.

So you just mentioned it: You decide to leave after 16 or 17 years—

Yeah.

—and you take a hard turn—

You think so? I actually don’t think so.

—into government. Why?

Yeah. My wife asks me that a lot, too. My wife is way smarter than I am, and her job is mostly to question, “What the hell am I doing?”

This seems to be a recurring theme, with your wife questioning your decisions.

Yes. The historical reason was that I had literally done policy stuff starting in high school. I worked for a group back in high school that used to do healthcare policy research. We were trying to figure out the economics of breast cancer screening. When I went to Stanford, I worked for a professor named Alan Garber, who’s actually now the president of Harvard, doing stuff around prostate cancer. The question was whether it made sense to screen large populations of men for prostate cancer.

I won’t go into details, but there are weird nuances to prostate cancer. The basic answer is that most men die with some form of prostate cancer; very few men die of prostate cancer. Screening broad populations can be problematic because you’ll catch a lot of things that never would have affected health outcomes.

I’d been doing that, and then when I went to Stanford Law School, I worked for another guy whose name is in the news today, Jay Bhattacharya, who now runs the NIH, if you follow that. Jay was a researcher at Stanford. I was his grunt. I used to do modeling for him for a policy group he worked on.

I always had these interests in policy, and then life intervened, and I ended up going in a different direction and into the private sector. But at some point, doing policy stuff was always on my bucket list. The reason for doing it now was really a couple of things. From a personal perspective, it was the right time in my life. My kids were all out of the house at that point.

Mark and Ben were incredibly gracious and said, “Hey, look, if this is what you want to do, we’re happy to support it.” They enabled me to transition my stuff off and gave me the freedom to do that. I’ve also been very lucky financially along the way to have the means to take a break from the business. The second thing was that I really believe what I learned in venture capital might have some applicability to what I would do in government.

So the job I have is this: I run this group called the Office of Personnel Management. Essentially, you could think of us as the HR and talent agency for the 2 million civilian federal employees. We’re responsible for making sure we have the right people coming into government, and for all the policies that govern talent in government.

Why did you think those skills were transferable from your previous job to this one, though?

Because what I learned, particularly through all the work I’d done with Mark and Ben in companies and managing people, is something very simple. It may sound obvious, but I think it’s really true: organizations are pretty simple. It doesn’t matter if you’re a government organization, a private organization, or a not-for-profit. They’re all the same. An organization is a collection of people who are aligned around a similar set of objectives. It’s no more complicated than that.

The question is: do you have the right set of objectives? Do you have the right people? And do you have a set of incentives that track to those objectives, so that when people do the things you want them to do, they get rewarded and recognized, and you can then do 10 more great objectives? Or, when they don’t do that, do you have a mechanism to correct that behavior and/or remove them from the organization?

When I looked at that, I thought, “Okay, I’m going to come to government, and it’s a totally different set of objectives, which I agree with. But the fundamental question is: do we have the right people, and do we have the right incentive structure to actually enforce the behavior we want?” The short answer, particularly on the latter side, is that the incentive structure is completely screwed up from my perspective.

Before you move on—

Yeah.

Let’s describe this. Why is the incentive structure completely screwed up from your perspective?

So, a couple things. Number one is, it’s an incredibly legal, regulatory, compliance-related culture. Everything that happens in government is basically driven by fear of doing something wrong, so we do the same thing we’ve been doing all along. The idea of innovation, any kind of innovation, is very difficult because the better thing to do is keep doing what you’re doing and not take risks.

When I first came in, one of my reports said, “Look, our job is to keep you out of the news, basically.” I said, “Well, tell me more about that.” They said, “We don’t want you to get hauled in front of Congress for doing something they don’t want you to do. We don’t want you to get a report from the GAO,” which is basically a compliance organization. “We don’t want the inspector general doing stuff.” I said, “Okay.”

Okay, so then let me just sit here and do nothing.

That’s easy. Yeah, it’s pretty easy to solve that, right? We’ll just do nothing. And that’s exactly right; that’s what I said. I was like, “That’s not your job. Your job is to help us figure out what we can do that’s good. And yes, if we have a good reason for doing something and we screw it up, I’m fine getting hauled in front of some congressional committee and explaining that to them. Obviously, if our process was crappy, that sucks. We’ve got to fix that.”

But what it means is that it doesn’t allow for any innovation. I come from this crazy world in venture capital where we take obscene amounts of risk, which is fine because you’ve got upside that’s potentially uncapped. If you find Coinbase in 2012, as Chris Dixon and Mark did, and it works, it can generate 300 or 500 times your money. That’s a good thing to happen, and that can cover a lot of mistakes.

We don’t have that in government. Instead, the answer has been to take zero risk, which also means there’s no consideration of the upside potential of anything. Part of the incentive structure we have to put in place, and are trying to put in place, is what I’ve been calling “permissioned risk.” That’s deliberately different from the permissionless risk I think exists in Silicon Valley. We’ve got to have this conversation, and we’ve got to be willing to let people take some risk. That risk needs to be appropriately sized for the organization.

If you’re in the Department of War and the risk is that people die, maybe we ought to have a really low tolerance for that. But what I do, and what most people in government do, isn’t life or death. We want to get people their Social Security checks. We want to get people their IRS refunds. If it happens a day or two earlier or later than we expect, in most cases nobody’s going to die.

But if we could automate a system and figure out how to make it dead simple for people to get their Social Security checks, retirement checks, or anything like that, that has massive upside potential. The incentive system has never allowed that. That’s part of the reason the system is screwed up: there’s zero opportunity for innovation. As a result, you have slow, plodding, not-great customer service from a lot of these organizations.

Well, not only that: if there’s zero possibility for any kind of innovation, what kind of talent are you going to attract?

Yeah. So here’s the other side of this: you have a performance management system that’s also screwed up. I’ll give you a story. I think you and I may have talked about this a long time ago, when we first met. Everybody generally gets ranked from 1 through 5 at the end of the year in terms of their employee rating. Some organizations have a 4-point rating, but let’s just say 1 through 5. One is the worst—you definitely should not be here. Five is like you’re knocking the cover off the ball. And 3 means, “Hey, you know what? You’re good. You’re doing what we asked you to do.” That’s all fine; every organization needs a lot of people like that.

Historically in government, 99.7% of people are a 3 or above, meaning literally 0.3% fall below expectations. Now, in a population of 2 million people—

Is that even possible?

It’s not even mathematically possible.

Okay.

It’s not. So you have one problem, which is that we don’t do anything on low-end accountability in the government. I’ll tell you how we’re going to change this in a minute. But that’s a serious problem, to your point: how do I attract and retain great people when I’m sitting here busting my ass, and you’re dialing it in, and everyone knows you suck, but nobody actually wants to do anything about it? That’s a terrible place for high-performing people to be.

To leave.

Yes, exactly. And on the other end, 4s and 5s are the top ratings. Those mean not just that you’re doing your stuff, but that you’re extraordinary or outstanding, that you’re doing great stuff. Just to give you an example, on the most senior side of the government, which is about 7,000 senior executives, something like 90% of people get ranked a 4 or a 5. Literally everybody’s awesome. And again—

In an organization of 7,000 people, impossible.

Yeah, it’s not possible. What it means is that we’ve gamed the system. As a result, if I’m a high performer, I want to be a high performer, and there ought to be smaller numbers of people who are high performers. By the way, I ought to disproportionately get money for that.

When 90% of the people get ranked at the top of the class, the bonuses make no sense. I give you $1,000, and it has no incentive value or retentive value. Instead, we should have a system—and this is what we’re doing—we’ve put a rule in place that no more than 40% of the population can be at that top ranking, and 60-plus percent of the compensation pool needs to go to those 40% of people. Personally, I’d like to do something even more extreme, but we’ve got to crawl before we walk and run on this stuff.

If you want to attract great people, you have to hold people accountable on the bottom end, and you have to allow for differentiation. The third thing, by the way—you’ve got me on my soapbox—is that almost every promotion opportunity in government is gated by tenure. You have to be in the job for a certain period of time, typically at least 1 year, before I can even consider you for a promotion.

If I hire someone who’s awesome in software development and they knock the cover off the ball, and within 6 months they demonstrate that they’re better than I thought—

They’re essentially stuck in purgatory for 6 months.

They’re stuck, yeah. And then, by the way, if you promote them—

And then they have to—

They have to wait another 12 months, right? So we’re eliminating stuff like that.

These are all policy changes we’re making. We’re going to eliminate time-in-grade. We want a merit-based culture where you come in and we level you appropriately for what you can do.

One of the programs we have is called Tech Force. We’re trying to build an elite corps of 1,000 software engineers, AI experts, and data scientists to serve two-year terms in government. We’re hiring them at what’s called the GS-14 level, which is pretty high. That’s the second-highest individual contributor level in government.

We’re hiring that way because we’re giving them assessments that technically analyze their skill sets. They come in and have to take an actual code development test administered by a third party. The government has never done this before. Instead, I would give you a questionnaire, and you would self-rate your skills. You would say, “I’m a really—”

It’s one through five? Turns out I’m a five.

You’re a five. You know this language, you know that language.

Uh.

It’s crazy, and there’s a whole historical reason for it: there was a lawsuit about disparate impact on these exams. But the point is, if we have a merit-based culture, we’ll bring people in at the right level and promote them at the right pace. I believe they’ll want to stay because they’ll feel like they’re actually able to do what they want to do.

Until the changes we’ve made, historically we’ve had a system where it’s very hard to break out of the pack. That’s just not conducive to having a high-performance culture.

In the simplest description you can muster, what is the main problem as you see it that made you feel the need to change your entire life, move across the country again, and dedicate two, three, four years of your life to doing this?

Yeah.

Is it the demographic problem? Is it that we just don’t—It doesn’t matter; it’s not a political thing.

No, no.

But if you care deeply about America—

Yeah.

—you want the most talented people—

Yes.

—in government.

Yeah. Three things motivated me, which we’re trying to address. One is: Do we have the right talent? Do we have the right people in the right seats? This is where, again—

And the answer’s clearly no.

So—

You can’t say this, but—

Yeah. Look, I want to be fair. There are a lot of very good people in the federal government. I think a lot of this is a systems-and-incentives problem, not necessarily a talent problem. Now, there is a talent problem, which you and I talked about a little bit: we’re bad at getting early-career people into government.

Explain the demographics of who’s working in the government right now, because this is a wild stat.

Today, 7%—seven—of the federal workforce is under the age of 30. Just so your listeners have a comparison, 23% of the total U.S. workforce is under the age of 30. So, by a factor of three to one or more, the federal government is way under-indexed on early-career people. If you go to the other side, 44% of our population is over the age of 50.

In the government workforce.

In government, sorry. So, 44% in government is over the age of 50.

Yeah.

That compares with about 33% in the total U.S. workforce. So we’re three to one under-indexed, and we’re 11 percentage points over-indexed in the over-50 category in the federal government. We’ve talked about this, but look, I’m a small-government guy. I won’t presuppose your ideological views, but—

I’m the son of a Cuban immigrant.

Okay, so you’re more small-government than big-government.

Yeah.

Okay, that explains a lot. If we did nothing else, half of our workforce will retire over the next five or 10 years, and we’ll be stuck with nobody to actually do the work. We need something more than zero people working in the federal government.

The first thing that motivated me is: Can we solve that problem? I think you solve it in two ways. One is that we actually have to tell a story about why it’s interesting to go to government. One of the things I’m doing after I go here is starting a college tour, which we’ve never done at OPM. We’ve never gone out to colleges, where the actual young people are, and said, “Let’s go find you. Let’s tell you what’s compelling. Let’s actually create programs around this.” So we’re going to do that.

You may not have seen this, but I do these stupid social media things on Instagram and WhatsApp because I’ve hired younger people who understand how to get a message to people who are actually looking at this stuff.

And your message is what? We need young, technically talented people to just—

Here’s my message—

—do a two-year stint.

Yeah.

And you don’t?

Exactly.

Somebody was saying, when you first came in, “Oh, you’re kind of screwing things up because our pitch here is, ‘You have a job for life.’” And you’re like, “Yeah, but young people don’t think that way.”

Yeah. I had a manager who told me that. She said, “You guys suck because our whole pitch to people is, ‘Come here and you have lifetime employment.’” And I told her, “Look, that is the worst pitch I’ve ever heard.” There’s not a single person under the age of 30 today who is thinking about lifetime employment. And if they are, I hate to say this, but that’s not the kind of person we want. That’s not the kind of self-motivated, driven person we want in government.

My pitch to people is, “Look, come here. Do something that’s good for your country, but do something that’s actually good for yourself. Come spend two or three years working on some of the most complex, hardest problems out there.” If you’re interested in space, go work at NASA. What Jared Isaacman is doing at NASA is fucking awesome, and you should be part of that. Or if you want to work on energy, go to the Department of Energy. There’s all kinds of cool supercomputer stuff and all kinds of things. Work on these hard problems.

After that, quite frankly, if you want to go to the private sector, that’s amazing. Let’s help you go to the private sector. This Tech Force program I mentioned is for two-year stints, and we’ve partnered with 40 private companies. All the names you would know: Alphabet, Meta, Databricks, and all these guys. During the two years, they’re going to provide training and career development for the people in this program.

At the end of the two years, we’re going to have a massive job fair. If we’ve done our job right in the government, we’ll have trained these people so well that I want every private company to be fawning over them. It’s a little bit like what you and I were mentioning about Israel and the IDF. They have this cyber corps, where the best and brightest kids coming out of high school go. When they come out, every venture capital firm is falling all over itself to fund them, and every cyber company wants to hire them. That’s the kind of thing we want for the federal government.

If people love being in government, great. God bless them; they can stay in government. But we don’t need to ask people to make 30- or 40-year commitments.

I think appealing to people’s interests is way smarter than saying, “Hey, here’s the government. You’re a citizen. Come try to do something good for the government.” That will get some percentage of people, but you’re going to get a lot more if it’s, “No, this is actually good for you—

Yes.

—and your country at the same time.” It’s aligning the incentives. And look what could happen. This is where you’re starting when you come in. This is what we can do, this is what you’ll learn during this, and this is what happens after that.

That’s exactly right. And then while you’re here—which comes to the second point—I have to create a high-performance culture for people like that to succeed. If I can convince young, aggressive, career-minded people to come to government, they can’t come into the environment I walked into, where you have to sit in a job for a year before I consider you for promotion, or I can’t hire you at a GS-14 level because you have experience.

I’ll tell you a story somebody reminded me of today. If you follow the DOGE stuff, there’s a guy named Big Balls. Have you ever heard of Big Balls?

I’ve heard of Big Balls.

That guy.

I don’t know anything else but his name.

It’s very hard to say it on air. But anyway, as I mentioned to somebody I talked to this morning, he actually has a name. His name is Ed, so apologies to his mom.

Yeah.

Big Balls has become his name. He joined the government. He joined DOGE. He joined Elon’s team. I think he was 19 years old. He went to Northwestern, and I think he dropped out as a sophomore. He’s one of the most brilliant software developers around, basically. He’s like, “Look, I got motivated by Elon, and I want to do this stuff.” And he’s still there, by the way. He works for Joe Gebbia now as part of this National Design Studio, which is redesigning all of the websites for the government.

A few months after he joined, I saw this article come out. This was at the height of the DOGE stuff, when of course everyone was thinking the world had come to an end. There was a quote from a government employee that said, “We would never hire…” He didn’t say Big Balls. He said Ed. He said, “We would never hire Ed in government because he didn’t have a college degree, and he didn’t have 5 years of work experience.”

When I saw that, I literally wanted to pull my fucking hair out, because I was like, this is exactly the problem. This guy’s awesome. We should hire him because he actually has the skills to help us, and almost every job description in government has a college degree requirement. In my mind, that’s a proxy for skill that may be indicative of something, but may not be. The fact that Ed didn’t have a college degree had no bearing on his software engineering skill set, but it would have been used to keep him out of a job because he wouldn’t have qualified. Or we would have hired him at this low level, where we would have paid him $40,000 a year, which he never would do, of course. He would go to Silicon Valley and get paid 10 times that.

Part of this is, if we’re going to get great people, we have to be able to have them walk into a culture that believes in performance, believes in accountability, allows them to progress the organization, and pays them at the level that’s appropriate for them. Look, we’re never going to pay what Silicon Valley does, and that’s fine. But we ought to at least level people appropriately. Ed should not come in at a GS-5, which is the most entry-level for government. He should come in at a GS-14 or GS-15, which are the highest levels, because his merit shows that he’s capable of performing at that level. But the system as it existed would not have allowed that to happen.

This is what we’ve been doing from a policy perspective: fixing all that stuff and saying, “Merit matters. Take a test, take an assessment, and everything else, quite frankly, we shouldn’t worry about.”

So Big Balls kind of gives you the blueprint where he says, “I’m super talented—”

Yes.

“—I dropped out, maybe I never would have qualified for the set of requirements in federal government, but I heard this person I admire tell a compelling story.”

Exactly. Right.

You came in before, during, or after DOGE? What was—

Yeah. I came in with the DOGE guy, so it was obviously Elon, but there were guys like Steve Davis, who was kind of the day-to-day person running it. I had the opportunity to spend a bunch of time with them before inauguration. I got nominated the first week of December or something, so it was kind of December and the first part of January leading up to inauguration.

Then they tell you, “Okay, until you get through your nomination process, you basically have to be persona non grata and stop communicating with folks,” because otherwise it can foul your nomination process. We had developed a bunch of plans and ideas for what we would do once the administration turned over, but I wasn’t there on day one to actually execute those things.

But you were observing a lot of what was going on.

That’s exactly right. And they were executing it literally out of my office. It turns out my office was ground zero for everything that was happening.

So regardless of controversy over—

Yeah.

—anything that happened there, I’m not even interested in that. I’m interested in whether there are any ideas you saw coming out of that that you thought were good and valuable and should be spread further.

Here’s my overall view. Part of the issue is—just to give you a sense of the numbers—there were roughly 2.3 million federal civilian employees on day one when President Trump took office. Today, we’re just about 2 million. So we’ve reduced the size of the federal government by close to 300,000 people, which puts the federal government at the lowest level it’s been, as a percentage of the workforce, since about 1960. You literally have to go back to Clinton-Gore before you have anywhere even close to that. The numbers the president has are much higher, but still, that was the last time you had any serious effort.

The entire apparatus of government travels in one direction only: budgets just get bigger every year. Every time an agency wants to do something new, they go back to Congress and say, “Great, you want us to do that thing? Give us more headcount to do it.” That’s been the path of travel for the last 50 years or so. So the way I think about what Elon did is that Elon had to come in and change the direction of a 2.3-million-person oil tanker. You can’t do gradual, incremental stuff. You have to have a big-bang approach.

People can argue and debate about his tactics, and that’s fair. I think there are fair arguments on both sides. But I don’t think there’s any argument that you could ever make the change we did without a big-bang strategy. Look, he got a lot of shit for it. Lots of people got a lot of shit for it, and history will write the story about that one.

When I came in, I was like, great, they did all this work. Now we have to institutionalize this. That’s how I think about my role: Elon, Steve Davis, and all these guys did an amazing job of breaking glass and saying, “This is how things are going to be.” Now we have to do the long-term institutionalization of it, which means we have to go through and say, “Great, what are you actually working on? Tell me the 20 things you’re working on. Which ones are really important? Which ones should we not be doing?” We never take stuff off the plate in government. We just keep adding.

This inertia—

It is.

—is insane.

It’s crazy.

You oversee retirement benefits for all, right?

Yeah.

We were on the phone the other day, and you told me a story I just could not believe. I was like, “Do you understand?” We were talking about the fact that America is full of dynamic, really smart, high-agency young people who don’t even know this is a path and an opportunity. Maybe we should just shine a light on this—

Yeah.

—and it’d be beneficial regardless of what you think. If you have the choice between a better-functioning government or not, of course you’re going to choose that, and it—

Sure.

—it has to do with the people that are in it and an organization. But then you’re like, “Do you know about the mine full of 4 million pages of documents?” And—

400 million.

I’m sorry.

400 million.

Sorry.

Yes.

400 million. And I’m like, “Scott, what the hell are you talking about?”

I know.

Can you tell this story?

It’s a great story. And credit to Elon, because Elon made this story a real one. We, OPM, manage retirement. Roughly 100,000 to 200,000 people retire every year from the government, and when you retire, there’s a whole process. You have to fill out a bunch of paperwork. It has to go to your HR department, which has to confirm it.

One of the literally crazy problems in government, by the way, is that we do not have the canonical history of your government service until you decide to retire. Most people to this day are in government for 30-plus years. They go to different agencies over time.

Each of those agencies has different HR systems and different payroll systems, so I can’t look at David’s record and say, “Hey, you spent 5 years here, and you spent this much time there.” Literally, part of the retirement process is creating that record.

We’re changing that, by the way. One of the other big projects—which I’ll come back to in a minute—is consolidating the entire HR IT system in government from 120 different systems into a single system. We’re going to go from spending $2 billion a year to $40 million a year. We’re literally going to save 95% by getting to a single system, and we’ll now have a record of David Senra that you can look at every day and see all the details.

So anyway, you want to retire. What you did prior to last year was literally fill out paperwork, put a fucking stamp on it, and mail it to your HR department. They did their stuff, put a stamp on it, and mailed it to a payroll department, which is different from the HR department in the federal government.

Yes.

Payroll is a totally separate thing. The final piece is that they mail it to an underground mine in Boyers, Pennsylvania. Nobody here has probably ever heard of Boyers, Pennsylvania. It happens to be in Butler County, where the president’s assassination attempt happened in July 2024, I guess. It’s what you’d expect rural middle America to look like.

The federal government happens to be one of the largest employers out there. People go to work every day in an underground mine that used to be an actual working limestone mine. It was later turned into a storage facility, but you can’t get in there without going through massive security. You can go down into the mine, where people are literally driving golf carts around among the limestone walls.

Then the paperwork shows up, and we have 800 people in my retirement service organization who take it, read it, make a bunch of manual calculations, and give you a retirement check. Historically, that process takes anywhere from 3 months at the low end to more than 12 months, partly because we go back and forth. If you forgot to sign your paper, I literally snail-mail it back to you and ask you to sign it, introducing another month-long delay while it sits there waiting.

Elon identified this, rightly so. He was like, “This is insane.” He teamed up with Joe Gebbia, a co-founder of Airbnb who’s now running the National Design Studio. They started this project at OPM before I got there, and then I picked it up: How do we automate this process?

Today, we’ve fully automated the process, so you never actually send a piece of paper. In fact, we declared June 30 the last day of paper in the mine. We’ll never receive another piece of incoming paper there. Everything’s online. We should be able to do this in 2026. This is not rocket science, right?

Mm-hmm.

You should be able to fill out an online application for your retirement. As a result, with the first iteration of this, we can reduce the processing time by more than 50%. My goal is to reduce it by 90%. Our commitment is that if you give us a complete electronic application, you should literally have a full retirement check within 15 days. It should not be any harder than that.

We have 400 million existing pieces of paper from people who retired prior to last year. Sometimes they call us and say, “Hey, my spouse died,” or, “My child needs to be added to my insurance,” or whatever. If you call our call center today, we send what we call a runner. That’s actually a job classification in the mine. The runner goes into the mine and pulls a file out of a physical file cabinet. Some of those files date back to the 1950s. It’s great—I’ll show you some pictures after.

The runner brings the file to the person who adjudicates it, and then, when they’re done, returns it to its place. We’ve just purchased 2 massive digitization machines—I don’t know what you call them—literally scanners that take up an entire room. They’re like a mainframe computer.

Mm-hmm.

From this day forward, every single one of those 400 million pieces of paper is going to be digitized. That will allow us, at some point, to shut down the storage component of the mine, but it will also dramatically change how we serve people. Our service isn’t good for people, in part because we haven’t given the people there the tools at their fingertips to resolve problems. They literally have to wait for somebody to deliver a file from a file cabinet. It’s insane.

There are problems like this all over the government. They’re so obvious that we should fix them. But this comes back to what we talked about with risk. I didn’t discover this, and Elon, to his credit, didn’t discover it either. You can go back to the 1970s and 1980s and find articles about how fucking ridiculous this mine is. Nobody has ever tried to solve this because God forbid some congressional person says, “Oh, you screwed it up,” or, “You didn’t do this or that.”

I was like, “Look, this is ridiculous. We have to fix this.” But the incentive structure is so aligned against fixing these things. It’s way better to provide shitty service and not risk rocking the boat than to convert these things into a modern-day experience. This is the fundamental incentive problem we have to solve.

That ties to your main mission: if you don’t change this incentive structure, you’re not going to be able to recruit high-quality, talented people. Let’s play this out 10 years from now. It might not matter to you that 44% of the workforce, whatever the number is, is 50 or over. Well, it’s going to matter when they’re all gone and we have nobody to replace them.

Yes.

We have nobody to replace them.

Yes, that’s exactly right. Everything ties together, right? I’ve got to go tell people a good story to get them here. When they get here, I have to have an environment they’re excited to work in. If they’re excited to do the work, then we’ll actually get modernization because they won’t be trapped in a system that prevents them from trying new things.

We have to accept the fact that we’re going to screw some stuff up. Mark always gets excited about some things. There’s this whole meme in Silicon Valley that Silicon Valley loves failure, right? Nothing could be farther from the truth. Nobody fucking likes failure. But what Silicon Valley does recognize is that if we’re going to innovate, we’re not going to get it right all the time.

If we’re talking about life and death for people, we should be pretty careful to get it right all the time. But if I can make the retirement experience 10 times better, and there’s some reasonable risk that I fuck it up and it doesn’t work, quite frankly, we’ve got to incentivize people to take that risk. It’s crazy that we’re not doing that.

Historically, nobody’s been willing to do that because the person in my seat wanted some other political job afterward, and therefore a bad congressional hearing would stop it. I told the president this when he interviewed me in December, after he asked why I thought I was qualified for the job. I said, “Look, I’ll tell you the other reason why I think I’m qualified: there’s not some other job I’m trying to get out of this.”

I’ve been incredibly blessed and lucky in my private life. I didn’t come to this job because I’m competing for some other private-sector or public-sector job. I don’t really care. I’m going to do what I think is the right thing, and if I get in trouble, I’m either going to fight if I screwed up, or, if it turns out I did something stupid, that’s fine. Fire me. I’m okay with that.

But so many people are in these jobs because they’re a stepping stone to something else. That prevents them from being willing to take the appropriate level of risk to try to get stuff done. To me, that just makes no sense. This is why we’ve had stasis for so long.

All right, so let’s say somebody’s listening to this right now. They want to help. They’re young and want to do a few-year stint. Where do they go? How do they get in touch with you?

We have a whole website, believe it or not: earlycareers.gov. That’s one place they can go. We’re about to be on about 40 different college campuses. If they’re there, they can show up and see us. They can follow me on X. I’m just @scooper. OPM is @USOPM.

Any of those places. But literally, send me a DM if you’re interested, and I guarantee you we’ll figure out a way to make something work.

I love how much you’re on this. Thanks for taking the time to talk to me today.

Thanks, David. Appreciate it.

Appreciate it.

Yep.