#1004 - Sam Corcos - Inside DOGE, The IRS & How to Scam the US Government

29:00And the process for outsourcing it to a contractor is so cumbersome and onerous that sometimes you just have to do it yourself. Have you spoken to Dominic Cummings? No. You know who he is? He was Boris Johnson's enforcer, right hand during COVID. He was also the guy that Benedict Cumberbatch played in the movie about Cambridge Analytica for Brexit. So technically proficient, has a sub stack. He's like sort of modern DGN, does podcasts wearing a baseball cap. Anyway, I think it'd be interesting for you to speak to him. I'll intro you. But he famously brought up a bunch of stories. I'm going to just give you a few that will maybe make you feel a little bit less bad about your situation.

29:46One was during COVID, there was no way to keep the figures centralized. So they were written on pieces of paper, and then he would write them on a whiteboard. And Boris apparently was just shooting from the hip based on what newspapers were put in front of him that day. And that was where the policies came from. There was a need to find a way to have cloud document storage that people could live work on at the same time. But they didn't know if Google Docs was sufficiently secure, or the same for Microsoft Teams and stuff like that. So they engaged a consultant company who I think spent multiple years and multiple millions of pounds to tell them that we need to do it.

30:35We can build, we'll bootstrap it ourselves. Most impressively up until 2020, a good chunk of the NHS was still run on Windows XP and there was a special ransomware virus, WannaCry or something else, that got released that affected like 1 % of NHS and thousands, tens of thousands of patient data went pages and fax machines still quite heavily used. The rail system in the UK still famously uses fax machines and can't really explain why or how that gets unwound. So I guess it's not just you. We do have a much older country. So perhaps we have a little bit more of an excuse than you do. But when it comes to building tech inside of government, does government do that?

31:23What does building software and tech inside of government look like? Yeah. The interesting challenge about building software in the government that my vantage point is a little bit skewed because the IRS is basically a technology company. In a lot of other government organizations, the technology is very much a secondary consideration for them. But for the IRS, almost everything we do ultimately is downstream of technology. If you want to do enforcement of tax, you have to use computers to figure out who is non-compliant. If you want to pull up somebody's tax information when you're on the phone with them for customer service to resolve an issue, you need the IT system to be good in order to do that.

32:11Virtually everything we do is based on IT systems. So a lot of our systems are internally built. A lot of other systems are built with vendors. And these vendors are, I think the biggest challenge you have is there's this core incentive misalignment issue that you have with the vendors. where because the people making the decisions on whether to sign a contract or not, it's not their money. You have this very large budget. This is one of the other, I'm not sure I would call it a misconception, but it's commonly said. People say, oh, the budgets are so tight. I can tell you firsthand, especially with NIT at the IRS, the budgets are not tight.

33:03We spend so much money on these things, but nobody cares. Nobody's paying attention. We will gladly spend hundreds of millions of dollars with some vendor to build a thing over the next five years, but we won't pay our engineers slightly more money because of some weird statutory thing where you can't compensate engineers at even close to market rates once you reach a certain threshold. Uh, so, uh, I'll give you like a very specific example that I'm dealing with right now. There's a vendor that, uh, we wanted to work with to build a particular system. And when you say vendor, what do you mean?

33:42A vendor is like, uh, Accenture, Deloitte, Booz Allen, Hamilton, Leidos, Palantir. There's like a whole list of companies that are contractors that work commonly with the government. I'm sure there are thousands of them, but there are, there are some that are bigger and more well-known than others. I think Booz Allen Hamilton gets something like 97 % of their revenue from the government. They are like one of the quintessential vendors. One of these that I'm talking to now, there was a previous vendor that we were using and they were really taking advantage of the government on pricing. And so I talked to one of the new vendors and said, can you do this for us and how much would you charge?

34:26And it was a much lower rate. I think it was something like $2 million a year is what they offered. And I said, okay, great, but that's the price, right? I just want to make sure that that's going to be the price that we pay. They said, yep, we're not like those other guys. It's 2 million and there's a couple other projects they're working on as well. The first six months go really well. Great. Then the contract comes up for a renegotiation and now they want$100 million a year. And I'm asking them, I thought we talked about$2 million for this, $250 ,000 for that. It's like, oh, that was our pilot pricing.

35:06It's like, if I had known that you were going to raise your price arbitrarily by 20X for no reason whatsoever, I would not have agreed to this. But now we're in a position where we can't just stop because now they're in our systems. And so now there's going to have to be this whole, we can either, we're either going to come to terms with like a more reasonable price or we're going to have to find a new, we're going to start with a new vendor or we're going to have to bring it in-house. And it's just so tedious. Is that not what happens with all businesses though? That you use a third party, they come in, they pull the wool over your eyes a little bit.

35:45Is this unique to government in some way? So the only reason why it's unique to government is that historically nobody in my seat says no. Right. You're seen as a soft touch. Yeah. This is how you negotiate with government. Yeah. Basically, normally they would come back with a price like$100 million and they just check, oh, is this within our budget? Yes. All right. Sign off. But I'll say, where did you come up with these numbers? How are you defining this price? And they'll say, oh, well, we didn't actually change the price. It was just a change in scope. it's like but no it's the price is 20 times more and then they come back this actually happened like well what if it was like 50 million it's like you can just you can just like arbitrarily change the price like that guys what are we doing this is crazy yeah and uh this is not the thing that they they said directly they said well you know usually when we talk to government they don't negotiate on price people government entities usually prefer a fixed price it's like i don't see, this is completely irrelevant to me.

36:51I want to know what it is that we're paying for, and what we're paying for seems to be something completely arbitrary. So I think they're just used to not dealing with this, because actually this maybe ties into a broader problem, which is that I'm in this role for some limited duration of time as the chief information officer of treasury. Um, this role, I'm responsible for something in the range of like$10 billion of spend per year. This role is paid, I think$160 ,000 a year. And you're not allowed income from any other source. There's like a, there's a rule against like, I can't do consulting.

37:32I can't do anything else. Um, you also have to like publicly disclose your finances. there's all these limitations that come with it um and so historically these people are not like super experienced people who have run large technology organizations before uh it's historically been a role where you get promoted internally from some some role usually a non technical role uh and then you end up the cio and you're responsible for making these decisions And I think over the last several decades at IRS, it just led to a lot of very bad decisions. What have you had to give up to take on this role? Well, the simplest one is money, but you can make money later.

38:22I think the hardest selling point for this is that it's not just that you make less money. That's fine. but for the types of people that they want to bring on, the pitch that you have to tell them is like, I hope you have a lot of money in savings because as a result in taking this job, you'll probably burn down a lot of your savings. And so I've been trying to advocate for, just make it, these are called, a lot of these leadership roles are SES. There's the general GS, which is like most government employees. And then there's the SES, which is the executive leadership group. There's about 8 ,000 SES, if memory serves.

39:02Just make the compensation average. It's like industry average. I think the role that I'm in should probably be compensated something in the range of like a VP of engineering at Walmart would be probably a reasonable comp. But this role, the CIO often makes less than the engineers on the team. And I think that it leads to a lot of challenges when it comes to both keeping people in there who actually care. Like this is, I have found myself struggling with this where there are some contract, there's one that comes to mind that was a couple of weeks ago,$35 million for some set of new toys that we really didn't need, but it's technically in the budget.

39:53And I just said, no, we don't need this. And a lot of people were upset. And I found myself in the back of my head going like, why do I care? This is not my money. I could just say yes. And then this problem goes away and no one will care. There's not a single person that will care. By me saying no, there are only people who are upset. And there's like this ambiguous win for the taxpayer. Of course, I still did say no, but you can feel that in the back of your head. And you're thinking Like you really need people in these positions to care enough to be willing to make those hard calls. A quick aside, if you've been feeling a bit sluggish, your testosterone levels might be the problem.

40:34They play a huge role in your energy, focus and performance. But most people have no idea where those are or what to do if something's off, which is why I partnered with Function because I wanted a smarter and more comprehensive way to understand what's actually happening inside of my body. twice a year. They run lab tests that monitor over 100 biomarkers, and they've got a team of expert physicians that analyze the data and give you actionable advice to improve your health and lifespan. Seeing your testosterone levels and dozens of other biomarkers charted over the course of a year with actionable insights to actually improve them genuinely gives you a clear path to making your life better.

41:08And getting your blood work drawn and analyzed like this would usually be a nightmare and cost thousands, but with function, it's just$499. And right now, you can get an additional 100 bucks off, bringing it down to$399. So get the exact same blood panels that I get and save 100 bucks by going to the link in the description below or heading to functionhealth.com slash modern wisdom. That's functionhealth.com slash modern wisdom. Why should Americans trust Silicon Valley engineers more than career civil servants? Hmm. I'm not sure that that's the framing. I would say, why should they trust people with substantial, relevant industry experience versus people who oftentimes stumble their way into these leadership roles?

42:06I think that's probably the way that I would frame it. And it's not even like trust as in they're going to do anything nefarious. It's just when you work in an organization, you want your leaders to make good decisions. And I think especially within technology, you want people who make good decisions in these roles. and one of the consequences of not having this is just this contractor bloat and like a very specific and relevant example is the amount of money that we were spending at the irs on cyber security contractors was just astounding it was hundreds of millions of dollars a year and we were doing a contract review and we asked all of the leaders of these groups to justify their contracts.

42:55Which of these are mission critical? Which of these are not? Which can we de-scope? The cyber team came back, said 100 % of their spend was mission critical. Not a single thing could be changed. We moved on during that administrative leave portion. We put those leaders on leave. We brought in the cyber engineers who really understood what these contracts do. And we just went line by line. And on, I would guess about a third of them, we hadn't even used this vendor in years. And we were still paying them 10, 20,$30 million a year and just nobody cares. And when it's not your money, there is some non-zero risk that if you cancel a contract, something bad might happen.

43:43And there is zero benefit of canceling it for you at all. There's zero benefit. You can just ask for more money and you will probably get it. And then that's it. So finding people who are willing to make those calls, and this kind of ties into the industry comparison. If the IRS were a private company, it would have gone bankrupt many, many years ago because people would stop buying the service because it's bad. You'd have some feedback loop where people recognize these problems and they they stop funding it. Government contractors have been pretty vilified, I think, by Doge and other people over the last few months.

44:23How come? What is it specifically about the contractors that have got the eye of Sauron turned around? There are many things. I think the biggest one is this core incentive misalignment problem, which is the pricing is often arbitrary and they will claim that it's not. They say, oh, it's firm fixed price. Kind of like the one that I mentioned earlier. They said, oh, we didn't change the price at all. It's firm fixed price. We charge the same price to everyone. It's just a scope change. It's like, okay, but like we're not getting anything different and it costs more. So what's going on here? A lot of the contractors, some of this is not even specifically the fault of the contractors.

45:10So like, I'll give you, I'll give you a specific example of a type of problem that you experience. So if you want to do something simple, like, Hey, I'm going to sign up for this online service. And it's like$10. There might be some threshold where you can do it, but in general, it's like, great. We need to set up a contract vehicle, which could take a year. And you have to go through this whole rigmarole of like a competitive bidding process is what it's called. The procurement systems that we have today, the process that we go through to procure things is totally and utterly broken. And it leads to disastrous and bad outcomes.

45:52So if you wanted to do that, you'd have to do this competitive bid. And this is at least the case for IRS. I know from other discussions is the case a lot of other agencies as well, which is that if you are in technology and you need a specific piece of software, you'll write up the RFP, the proposal for what you need. That then goes to the procurement team. At that point, it's basically a black box. Usually the engineer has no further input into what is selected, who selects it, what they'll often just end up with some random vendor for something that is somewhat related to what they ask for.

46:35And then they just have to make do. The other challenge is that because we have these salary caps for most, for all roles really, but for a lot of the very specialized technical roles, we can't really be competitive with hiring some of these roles. So in the early to mid career, we can actually be reasonably competitive. We can pay engineers something like 160, 180. The cap is 226, I believe. So you can be reasonably competitive for those roles. But once you get into like mid to late career, which is when people really specialize, there's just no way to pay$250 ,000 ever. It just does not exist.

47:17And so So there are some, I think there's a special thing where there are a total of 800 people in the government that can exceed that number. And it requires a whole process to do. But the only way you can hire that 250 ,000 person, the way that you would do it is you would hire some contractor, one of these large entities that I mentioned before. You pay them$500 ,000 to funnel$250 ,000 to the person that you actually wanted to hire. And they'll arbitrage off the top. And then they'll arbitrage off the top. And this whole system incentivizes those kinds of things. How important is getting above 225 for the sort of technical talent that you need to run the government?

47:59Well, this is the interesting thing is that for the most part, you don't, but there are some that are crucially important. The IRS, as I mentioned, when I arrived, had about 8 ,500 people in IT. It's now about 6 ,000 plus another, I think 6 ,500 contractors. Yeah. And so the numbers are often misleading and it's hard to really get a handle on who is doing what and how many people are there. There are maybe a hundred of those roles in total where you would really like to be able to pay up to maybe$300 ,000. And it would make such an impact on the organization to be able to retain your really technical, high-context people that are there already and to be able to bring in people who have a lot of experience with this stuff.

48:55If you fix the contractor problem, what would improve?

49:02Well, I'll give you another specific example of like, The contract, the procurement process generally is we have this big initiative to eliminate paper processing at the IRS. We spend something ballpark with labor and all this stuff, ballpark a half a billion dollars a year on paper processing.

49:29We want to use external vendors. This is what people in the commercial space do all the time. There are vendors, they take in paper, they use robots to digitize them, and they basically just send you an API request. So technically, you're still receiving paper, but your organization doesn't even need to know about that. It's just - You don't need a warehouse. You don't need any of that. Yeah. We have this gigantic warehouse in Kansas City. I can show you pictures. There's paper stacked to the ceiling. And it's just some of these things are months, sometimes even years backlogged, just like sitting there because it takes so long to process these things.

50:06So, um, the, uh, when I first started, I said, this is really obvious. It's way less expensive to use the vendor. It scales way better. There are far fewer issues using these vendors. So we should just do that. great. We started the process. We have to write up the proposal. We do a competitive bidding process to get the vendors. We get a bunch of vendors who sign up. Cool. We pick a couple of the vendors. Then it gets contested, which is a thing where if you believe, if you're one of the other vendors and you believe that the process is unfair or that one of the other vendors was selected unfairly, you can contest it.

50:48It costs, I believe,$100 and you can just do it. Anyone can do it. And then the process stops for 30 days. And then there was some situation we had to recompete it, which then takes another, however, 30, 60 days. Then you have to do a new analysis. Then it gets contested again. And then it's just like, we started this process of zero paper in April, we don't even have the contract awarded yet to start on the project. And this costs us more than a million dollars a day that we don't have this implemented and nobody cares because we just have to follow the process. Because if you don't, it can get overturned and it can get contested.

51:35And it's just like the, like, it's just, it's a maddening. It's a million dollars a day. And I'm trying to make this point. We need this done. We need to start. I don't care which vendors we use. We need to start moving on this project. And it's like, well, you know, we have to wait for another 30 days before we can do anything because of this rule. And then, but you can also see how this exists for a reason, which is you don't want, you don't want people to just choose their buddy's company, right? That's, I'm sure the checks and balances, you need some checks and balances and you can kind of see how you got here.

52:11which is I'm sure there was a time when as soon as people got into government, they just like traded favors with all the people in the past, or they traded future favors of like, hey, company X, I'm going to get you all these contracts. And then as soon as I leave government, you're going to give me this cushy job. So you need some checks and balances there. It feels like we are way over indexed on process and on like competitiveness to the point where I'll give you another specific and absurd example, value-added resellers. So this is a thing because everything has to be a competitive process for the most part.

52:53If you want licenses for some generic product, the kind of thing where you could just go online and sign up for right now, be it like, I don't think licenses for Adobe or licenses for Figma or licenses for any web application that you would use, like any SaaS product. When you're looking for the thing, you're not looking for any similar tool. You need this one specific license, right? But you can't compete that because you might end up with some different license, which doesn't actually serve your purpose. So what you have to do, this is a common thing, is you have to create what's effectively a fake auction where you find a value added, you find three or four value added resellers who are all reselling that license and they're competing on the price for that license.

53:44So they all just take a cut off the top. They don't do anything, but they're there to provide cover for it looking like a competitive process. This is ubiquitous. And I think this is one of the things that the people over at GSA are looking to fix. General Services Administration is responsible for the procurement process broadly. This is one of the big things they're looking to fix because they just take money right off the top and they almost always do nothing. There's sort of a strange contrast going on here because people say the budgets are tight in government. But you seem to be saying the opposite at the same time.

54:26Which one's true? Are there misconceptions about how much money government spends? I think the, if I'm a, if I'm going to steal man the way that like when, when people are in the government and they say that budgets are tight, I think probably the fairest way to describe it is they're saying that for the things that are important that we want to spend money on, we are not allowed to. I think that's probably the fairest way to describe it. It's not that there's a lack of money overall. The overall budget, yeah, the overall budgets are astoundingly large. But when they say, hey, we'd like to spend$10 ,000 on this thing, it's like, well, which contract vehicle are you going to use?

55:10Oh, well, we don't have one. Okay, well, you need a competitive bidding process. Maybe in a year and a half, you can get that. And we will have spent a million dollars to try to come up with some contract vehicle so that you could spend$10 ,000. And so the amount of overhead, I've described it to a friend of mine recently as in government, everything is at least medium hard. There is no like, just go, oh yeah, that's simple. Just go, you can handle that. Everything requires so much time and overhead to do, especially with all this contract nonsense that it's the, if you want to do something very simple, if you want to hire good people, the problem is you can't spend money on things that are useful because there's so much overhead to do all these things that you're trapped.

55:59And so the amount of, I don't want to call it discretionary, but it's really like the, if you know what's important, like the paper initiative that I mentioned, I could argue budgets are tight because like we can't get money for this thing, but we have money for this thing. We just can't do it. We're stuck. So legislation is tight. Budgets are wide. you could say that or you could say the rules are tight the rules are cumbersome and onerous because most of the stuff for it's called far um federal acquisitions regulation i think is what it stands for i don't think hardly any of that involves legislation it's just like compounded rulemaking from gsa and other groups that just make it really hard to do these things When it comes to the big spending government, how much of this is negligence, fraud, incompetence, waste?

56:55How culpable is anybody here? Because it sounds like everyone has plausible deniability. Yeah, I think one of the things that's been a strange observation is this term waste, fraud, and abuse.

57:16Trump in his first campaign ran on drain the swamp. This has been a longstanding campaign goal. I think what was most surprising was discovering how gray the boundaries are between these concepts of what is waste, what is fraud, and what is abuse. I'll give you a very specific example. This was not at our agency, but I was talking to another CIO who's dealing with this. They were doing contract review and they found a contract for a piece of software that's like$20 million a year that has been on the books for like five to 10 years. And one of the Doge initiatives was matching the number of licenses that we pay for with the number of licenses that are in use.

58:08and they discovered that we'd been paying for these licenses. And not only was nobody using it now, no one had ever used it. Not a single license had ever been activated for this product. And it was set to just effectively auto renew. And every time it would come up for renewal, people would say, oh, yes, mission critical. We have to keep it because nobody really knew what it was for, who used it. So is that waste? Is it fraud? Is it abuse? I think we could say it's wasteful at a minimum, but you start looking into like, oh, well, who runs the company? It's like, oh, that's interesting because that guy used to be in the government.

58:48I wonder if there's anything there, but nobody really looks into it. These boundaries get very fuzzy and it's really hard to know what's going on. I would say in terms of like abuse as well, this ties into some of the other procurement problems. Like, uh, there's this whole program within, uh, the small business administration that has some requirements on how many of your contracts go to small businesses. And there's this, it's the eight a program and there's something called, uh, the small disadvantaged businesses. And frankly, it's just a huge scam. The. If you give a contract to a small disadvantaged business, you can skip the whole procurement process.

59:37You can skip the competitive bidding process. You can pick the vendor that you want. It's like you get the fast lane, but it's capped. For some of these groups, it's capped at$25 million. So this is one of the things when I was doing contract review, and you see all these things like$24.9 million,$24.9 million,$24.9 million. You're like, okay, something's going on here, and you can skip it. And these contractors or this workaround, they will often take 10, 20, sometimes 50 % off the top of a contract to do actually nothing. On the books, they're supposed to do 51 % of the work is what it says, but they never do 51 % of the work.

1:00:19I shouldn't say never, almost never. I have yet to find an example, having looked for this. I have yet to find an example where they do any of the work, really, and they just funnel money. I think when I was talking to somebody who's investigating this, they estimate that this might be as much as$80 billion a year in just graft. Yeah. My estimate puts it at about$30 billion. We'll get back to talking in one minute, but first, Momentus make the highest quality supplements on the planet, and I have fallen in love with their whey protein powder. It is grass-fed and each batch is exclusively sourced from EU dairy farmers whose cows are free of hormones, steroids, and all of the other junk.

1:01:03So if you have been struggling to hit your protein goals, which let's face it, is everybody, Momentous' grass-fed way is a super convenient way to get super convenient way to get extra protein into your diet, lean muscle mass, muscle growth, supports recovery, post-exercise, etc. But it's also NSF certified for sport, meaning that even Olympians can use it. And there is a 30-day money-back guarantee. so you can buy either or both of these, try them, and if you don't like it for any reason, you will get your money back, plus they ship internationally. Right now, you can get 35 % off your first subscription and that 30-day money-back guarantee by going to the link in the description below or heading to livemomentous.com slash modernwisdom and using the code modernwisdom at checkout.

1:01:43That's L-I-V-E-M-O-M-E-N-T-O-U-S dot com slash modernwisdom and modernwisdom at checkout.

1:01:56a increasingly large group of people are getting increasingly rich off of this yeah that's right how's that feel from the inside and how's that feel as an american i would say from the inside it's uh it's obviously very frustrating i think it's in some ways being in the role that i'm in i feel encouraged because you can see how to fix it look at these big glaring problems. Yeah, it's both concerning and encouraging at the same time because you can see how bad it is, but also you have such an impact on being able to change these because you have the willingness and the authority to do it. And you can have$10,$100 million swings with very little effort because people just haven't really looked at these things before.

1:02:50So I think the biggest surprise to me has been for so many of these things, just the scale of how problematic a lot of these contracts are. I think everyone kind of knows that the government is inefficient and that's kind of okay. We all know that. But we all think these things might be 2x or 3x inefficient. But you find examples where things are 100x, 1 ,000x beyond what you pay in the commercial sector. And it's just, there's just no excuse for it. It's crazy. Is there anything in addition to the technical talent problem that's contributing to dysfunction of IT systems in the government? I think upstream of all of these things is leadership.

1:03:40And I would separate out the, call it political leadership from administration to administration and much more the career civil servant leadership. Um, there are some examples of, uh, entities that are within my purview that are pretty well run. Um, FinCEN, which does a lot of, uh, they do a lot of, uh, international finance, uh, investigations. Uh, they have their CIO, Amy Taylor, who's technical. She's been there for a long time and she runs a tight ship and you can see downstream of good leadership. tends to lead to good teams. They tend to have technical talent. And you see some of these organizations that have had, like Tim Gribbins over at Bureau of the Fiscal Service, he's been running that for six years.

1:04:35He's had stable leadership. He is very intentional about who's on his team. They do performance management. They take it seriously. And they have a pretty good team. So if you have good leadership, things like what is the technical talent of the people on the ground tends to work itself out because the people at the top care enough to follow through on ensuring that you have that. They continue to find people until they get the right technical talent. Yep. Yeah. It's a rule of business that A grade players hire A grade players, but B grade players hire C and D grade players. That's right. And so many of these things.

1:05:14So at the IRS, which is to be blunt has had poor technical leadership for roughly 40 years. It's just this compounding issue. So we at the IRS, the policy for a long time has been, we don't do technical interviews. We just do resume review. So if you were to apply and you just wrote the word Java on your resume, we will just assume that you know how to do this. And so the, it's a sort of a joke, but it's not really a joke. Somebody told me, they said, historically, if you want to hire a good engineer, you have to hire five because you just don't know what you're going to get. So if you need five engineers, you got to put out 30 roles to find five or six good people.

1:05:58Because the hiring process is not sufficiently scrutinous. Yeah, and this, I'll give you, again, this is a thing that I'm dealing with right now. I'm loving the examples. It sounds sort of very conceptual and fluffy until you bring it into land, with real stories. Totally. And this is the benefit of just being in the weeds trying to solve these problems. Cold face. Yeah. The specific thing I'm working through right now is we're in the process of recomposing the engineering org in the IRS, which is we have too many people within the engineering function who are not engineers. And where they end up, I think it's going to be in chief operating officer, some of them in product and design.

1:06:46It doesn't really matter. But I have a memo that I'm still drafting that I'm probably going to publish at some point. The memo is titled, Engineering is for Engineers. And it argues that if you're in engineering, you should be an engineer. And if you're not an engineer, you should not be in engineering. Anyone from the private sector would view that as a tautology. Because why would you ever have somebody who's not an engineer in engineering? But in government, this is going to be very controversial because most people in these roles are not engineers. So the goal is let's find who our engineers are.

1:07:31Let's move the people who are not into some other function. And then we're going to bring in more engineers. I think we currently have 100 engineering roles open right now at the IRS, which if anyone listening wants to apply, it's at USA Jobs. That's the website. We have 100 roles open. The challenge, though, and I'll just walk you through the process why this is such a problem. At the IRS, we have six job openings, right? And as a point of comparison, if you wanted to hire somebody, you would write up the job description. This is what you do. You post it. You interview those people. You decide if you want them.

1:08:13You hire them or you don't hire them. It's not like that here. So let's say you want an engineer who specifically does this one thing. Well, you send it to the HR team. They decide if it matches one of the existing, they call them PDs. position descriptions, I think, if it matches one of the existing approved PDs, which in the case of our current hiring pipeline, many of them did not. And so they ended up getting posted with some fairly generic thing. But by the time you've written the position description, the engineers are no longer in the loop on this process. So it gets handed off to a mid-level person in HR.

1:08:58They decide if your job description is accurate or not. They then post it. You get a bunch of people apply. The mid-level in HR decides whether that person is qualified. The engineer is not in this loop at all, by the way. Yeah, the HR person does not have technical capability. They can't assess whether or not this person can actually write in Java or whatever. They just look at a resume. They make the determination. Is this person qualified for this role? and then eventually this person, the engineer, will finally get a packet of PDFs to review. It's like, here's 10 people that we think are qualified for this role.

1:09:35And they don't get to see the people coming into the pipeline. They don't get any of the stuff. I'm currently in the process of trying to figure out, is this a law that says that the engineers are not allowed in this process? Is it just fiefdom building where somebody is trying to just exclude them so that they have a role that can't be disintermediated. How do we get the people who are the hiring manager, the engineers, the engineering leaders who need more engineers, how do we get them back in the process of actually hiring our engineers? I'm actively working on this right now. I'm working with the Office of Personnel Management to try to figure out what the issue is there.

1:10:22I'm working with IRS HR to try to figure out what the issue is there. And this kind of ties into the other set of problems that I've found. One thing that is fairly unique about what's happening right now is the amount of collaboration between agencies is probably the highest it's ever been. So when you have an issue with the Office of Personnel management, I can just call and I know legal counsel there and I can call them and I can say, this is what we're seeing. Is this an OPM policy or not? He'll look into it. He'll get back to me. Usually these things are done in such information silos that, uh, that sort of collaboration never happens.

1:11:05The, the people at OPM might say, oh, it's not my problem. That's a treasury problem. So they'll just ignore it. But there's a, there's a level of collaboration and mission orientation to get these problems solved. That's probably the biggest difference from what we're seeing now, from what's been the case in the past, that I think really is leading to a lot of these changes. Doge became famous for hiring young people, infants, some are accused. Why is the federal government behind the private sector typically when it comes to hiring young people? So yeah, I think the biggest reason is there were actually some quite successful programs.

1:11:46There's one called Pathways. I think it was maybe before Pathways, but there was a program that used to exist where you could recruit people out of colleges and there was an internship program. But this actually maybe ties into what we were talking about before of like so many of these things are just unforced errors where we used to have this internship program. We'd get people out of college or grad school. They would be in the job. They'd see if they like it. And then we'd want to hire them. And then we could hire them. That worked. This was like early 2010s. Then there was a change that said, well, you can't just hire them because that, I believe it It has something to do with like they thought it violated something around veterans preference or something.

1:12:34There was some reason why they didn't want you to just be able to hire people normally anymore through that. So they had to go through a competitive process, finger quotes, right, which is the one that I basically just described before. And so it was often, and some of the people who have been there for a long time told me, it was usually the case that those interns who are actually really good, who they really wanted, would not be the ones that could be selected because they were deemed unqualified or they were deemed insufficiently experienced or they did not have enough of a veteran's preference to make it to the top of the stack.

1:13:14A lot of the hiring processes that we have are they, for a lot of these roles, you are specifically instructed to hire the minimally qualified candidate, which means you are instructed to hire effectively the first person who meets the minimum threshold. Okay. That sounds like it would be quick though. It is quick. All right. It's quick if you don't care who you get. Right. But it's quick to get people in who are often underqualified and not fit for purpose and slow to get them out. Yes. Effectively impossible to get them out. So you hire quickly, fire slowly. Yeah. Or never. Fire never. Yeah.

1:13:59You can hire slowly. There are different hiring authorities. There are some roles where you have something called direct hire authority, which just allows you to pick the people who you want for specific roles. There's a very limited number of roles that have direct hire authority. Most of these are in this competitive process. And the competitive process often means that you will end up with candidates that are not the ones that you want. You are painting a picture that is almost exactly what I thought it was going to be. You know, as a immigrant foreigner who doesn't really pay much attention or care about what goes on inside of the US government, this is precisely what I thought it would be.

1:14:38it is ossified you're right it's big it's inefficient um and i have to assume that it is causing even well-meaning people who really want to try and make change to sort of regress back to the mean of government so even the high agency you know upwardly mobile talented people that want to drive things forward learn over time that their efforts are kind of wasted and actually when they do try and move things forward they rock the boat so much they're probably disincentivized from doing that too so change changes highly uh highly push back against yeah and i would say there's the call it 100 or 200 people that i've really come to rely on at the irs out of 12 000 13 15 ,000 including contractors.

1:15:39Yeah. They're really quite good. And I've had conversations with them about like, why are you still here? Because if I was in their position, I don't think I would have lasted three months. And every one of them has said like, oh, believe me, I've thought about leaving. But several of them have said, I just, I care enough about the mission. I recognize that if I left, very bad things would happen and they have the sense of responsibility and that's why they're still here. And, uh, I, I really admire that and I'm glad that they're still there. I would, I would not be able to do this if they weren't there.

1:16:21What's the federal government missing that the private sector isn't? Is it a competition, incentives, accountability? I think there's one thing that it is always missing and will always miss. And this is just a,

1:16:40I would describe it as an assumption that you have to make when you're thinking about designing any system. And it's kind of how we got here to begin with, which is there is no feedback loop when it's the government. It's kind of like why FDR was so opposed to public sector unions is because in the private sector, you can unionize, but if you demand too much from the company, eventually the company will just die. But the government won't. You can just demand more. You can just work less. There is no counterbalance to the unions in the public sector. You can just always say yes and nothing happens.

1:17:14The debt just gets larger. The amount of money just goes up. And there is no force that pushes against it. So the assumption that you always have to make and why we end up with so many of these rules around personnel, around contracts, is that there is no natural incentive structure that prevents government from doing bad things. I think it doesn't surprise me that this is the sort of challenge that people are facing. but also 16 % only 16 % of Americans trust the federal government. That was some Pew study from 2023, I think. Is that surprising to you given that you, what you've seen from the inside?

1:18:01I don't know what trusting the government means, but you would trust somebody to look after your dog. Okay. So what are the things that they need where they need to be driven? They need to be competent. They need to be conscientious and aware. They need to have the skill set that's required to be able to look after a dog. They need to not be malicious against dogs. So there's, you know, the fraud, waste, abuse, like insider trade, all of those things. There's a number of different levels that you need to get through in order to be able to think, I trust this person. And not all of them are malicious and not all of them are competence based.

1:18:36But 16 % of Americans trust the federal government. Doesn't seem surprising. Yeah. I think it's generally good for people to be skeptical of the government. It is the entity with a monopoly on violence. That is a thing that one should inherently be skeptical of. I do wonder how much of that is just an overexposure to the news. I think that's, there are a lot of things that are either completely not real that people believe are true, or they are sufficiently non-representative of reality that they are effectively not real. And if you go way back, people cared a lot more. I'm sure you know all this stuff.

1:19:34People cared a lot more about their local community. There was no 24-7 national news coverage. But over time, it's basically become a team sport. And people watch it like they're watching football. And these things that have always happened sort of on the edges of the government are now front and center. And they look like they are representative of reality when they're not. Right. So I don't really know. And it's been interesting to observe from the inside how the tail can wag the dog in a lot of these things. Like a specific example is at the IRS, we hosted for the first time ever, we took 50 of our best engineers who have been there many for 10, 15 years across all the different teams.

1:20:20We brought them together for a strategic planning session at Treasury to discuss across all these teams, which rarely talk to each other. What are you seeing? How do we land the plane on this modernization program? It was incredibly useful. People were finding data sources that we didn't even know about. We were finding all kinds of issues. problems that were just a coordination problem that have been stuck for years got resolved on the spot because people could just talk to each other. Because historically, when you have this many layers, for an engineer to talk to another engineer, they've got to tell their supervisor, who tells their director, who tells the subject matter expert, who goes across to the other team.

1:21:06And then it goes through like eight layers of filtering, none of whom know anything about this topic. You can't just Slack them. You can't just like communicate with them. What is communication? What's the platform that you guys use? Teams typically or email. Microsoft Teams and email. Yeah. I haven't used Teams apart from - It's very similar to Slack. Okay, right. How many people are in your Teams? That's a good question. I don't know if I have any actual direct reports. I've just sort of co-opted a bunch of other teams. You're floating in. You're like a mercenary. Basically. Yeah, a freeloader.

1:21:46So there's these communication gaps that are a huge problem. We hosted an event of strategic planning. Wired Magazine put out a piece that says that we hosted a hackathon that Doge. Oh, this was what that was? Yeah. Ah. Doge, they said Doge hosted a hackathon with a bunch of Doge people. What's a hackathon for the people that aren't familiar? Well, so typically it's like you do 48 hours of a bunch of coding and you have some deliverable at the end. And what's... Navy SEAL Hell Week for coders. Yeah, for software developers. Yeah. And what's weird about it is they requested like, hey, we heard you guys are hosting a hackathon.

1:22:25And we said, no. Most of these people didn't even have their computers open. This is a strategic planning session with 50 career IRS engineers to talk about IRS modernization. And they invented the story out of whole cloth that it's a Doge hackathon. It's completely made up. And this is sort of the tail wagging the dog thing where I still get requests from Congress asking about this hackathon. Would that be a bad thing if you had hosted a hackathon? I'm not even sure it would be, but like none of this is real. What do you think is the subtext of the hackathon? This is Silicon Valley degeneracy, caffeine fueled, Adderall addicted, young people eating their Chick-fil-A and coding up the matrix.

1:23:14I think what this story intentionally implied, which is not real, but what they want people to think is that the young Doge engineers are hacking around with taxpayer data and we're doing something nefarious. Yeah, I suppose the word hackathon, it's not got great optics. Yeah, they're using these words intentionally and they suggest certain things that... You ever read Ryan Holiday's book, Trust Me, I'm Lying? Yeah. He has this great concept called trading up the chain, which is about how you start with something that's like a half truth that you suggest, which then, like, it started out, I think it's even in that article.

1:23:56They say that I am SpaceX affiliated because my brother interned there many years ago. And then, if you look at a lot of news stories today, they say, Sam Korkos, SpaceX engineer. I have never worked at SpaceX. Right. The closest affiliation I have is that my brother interned there. Like, I have nothing to do with SpaceX at all. But they just like, They imply something which suggests something which then just becomes a reality. And then all of these news stories start to say that. It's like a slippery Motten-Bailey type thing. Yeah, exactly. And it becomes the new narrative and they just build on top of it.

1:24:35How big of a deal was trying to deal with this hackathon story? Was that a huge pain in the ass? It was a huge pain in the ass. Yeah, because people kept asking. This is the tail wagging the dog thing, right? I started getting emails from IRS engineers saying, hey, can I join the next hackathon? It's like, guys, there is no hackathon. There was never a plan to do a hackathon. None of this is real. And it's just like, I still - I have enough to do already. I don't need to be pushing back. That is fake news. It's 100 % fake. Fake news. In other news, if you thought that lemonade couldn't get better, it just did.

1:25:12Element has taken the classic taste of lemonade and actually made it useful and functional, giving your body back the sodium, potassium, and magnesium. It actually loses when you sweat. No sugar, no junk, just the stuff that works. And that's why I'm such a huge fan. That was horrendous. You're nowhere near me. Lemonade in an ice cold glass of water. Just, okay, that was better. Hits different. Hits different. And they back it with the best policy for money back that I've ever seen. No questions asked. Unlimited duration. If you don't like it, they will give you your money back and you don't even need to return the box.

1:25:44That's how confident they are that you'll love it. I keep harping on about it because it genuinely makes a difference. When I use it, I feel the difference. And proper hydration is not just about drinking enough water. It's about having the electrolytes to allow your body to use those fluids. And they offer free shipping within the US. Plus, right now you can get a free sample pack of Elements' most popular flavors with your first purchase and that money back guarantee by going to the link in the description below or heading to drinklmnt.com slash modern wisdom that's drinklmnt.com slash modern wisdom right in the face yeah you thought you could hide back there didn't you no no no i see you man obviously a lot of people associate doge with elon elon left recently yeah how big of an impact well actually even before that how much of a driving force was elon in doge um In the stuff that I was doing, not a lot.

1:26:40I had very little interaction with him.

1:26:45We primarily worked with the secretary's team and within our agency. I think the biggest effect was there was more... How would I describe it? Yeah.

1:27:06I think the biggest impact was when there was the initial blow up, there was a week of real uncertainty on what it all meant. But over time, the teams had already been so embedded with the agencies and the departments. And since we really take our instructions from them, it didn't have that much of an impact. I think some of the impact is that the interagency communication weakened a little bit just because there's less of a central connecting force for these things. But for the stuff that I do, it has not had a particularly large impact. I imagine that in some other teams where they really needed Elon to push something from the top, that probably was more impactful.

1:27:57But I think it's very possible that part of the reason why it hasn't impacted us at Treasury that much is that Secretary Besant is able to get a lot of these things done himself. Right. So you don't need that big hammer, that big Elon-shaped hammer coming in to fix things. Okay. I guess Doge has kind of taken on a meme all of its own.

1:28:26What do you wish more people knew? about what doge is as a movement, as an organization. I don't know what you call it, an entity of loosely collected fucking group of morals. Yeah, it's like a loosely connected group of mission-oriented people who want to fix a lot of the things that are broken in our government and stop the ever-growing national debt from collapsing the country. That's probably the main driving force that we all share and why we joined. Why we joined the government is to get this solved. I would say the biggest thing is that these are some very smart people, some very high agency people.

1:29:18That's something that I've... It's a word that I've taken to using a lot more. I didn't... In the private sector... And so many people are high agency that it's never been a thing that I've thought. You don't need to select for it. Yeah. But in government, there are certain people that I've come to rely on. And when I say I rely on them, they are the people where when I say that this is important and I need somebody to do it, they will actually do it. And far fewer people will do that than you would expect. The number of people where you say, hey, I need this by Thursday. Thursday happens and you're like, Hey, did you do that thing?

1:29:58Oh yeah, I'm on it. I'll do it right now. Next week goes by. Hey, did you do that thing? Oh yeah, I'll get right on it. And it's just, I don't know if it's malice. I don't know if it's incompetence. I don't know what it is. It might just be the fact that they know that they can't get fired. I have no idea. I don't know genuinely for a lot of these people, why I cannot seem to get any deliverables out of them. But once you find the people who can actually deliver, you've got to hold on to them for dear life. Doge website says that you've saved$200 billion or so? Maybe, yeah. I've heard a lot of people say that's not accurate.

1:30:39What's the truth there? Well, so this is a long and nuanced conversation about how you calculate savings.

1:30:50so the i've actually had many internal debates on how we calculate this stuff so the way that we've been tracking it is the most common type of contract is a firm fixed price or basically just like think of it as a normal contract we're going to buy this thing from you for this amount of money um there's several other types of contracts bpas which is a blanket purchase agreement. There's an ID, IQ, indefinite duration, indefinite quantity. The way that savings are generally calculated is when you have a contract, let's say you're my contractor, I'm going to pay you a million dollars to do this thing.

1:31:32If I have already paid you 500 ,000 of that and then I cancel the remainder of the contract, then we will book that$500 ,000 as a saving. the problem is that doesn't fully capture the amount of savings because that contract probably would have been auto-renewed. The other challenge is for like a blanket purchase agreement, it's really more like a credit card limit. An IDIQ is basically the same thing. They tend to be for different stuff. So these BPAs can be very large. I think there are some IDIQs that are like a trillion dollars for something like nine millimeter ammunition for the military.

1:32:13It's basically just saying, if you need it, just buy it. Like just, this is who you buy it from. This is how much it costs. Just buy however much you need. Um, a blanket purchase agreement says like you can buy up to a hundred million dollars of services from this company and it's fairly open ended. Um, the, the challenge comes down to how do you decide what a savings is? If you have a $2 billion blanket purchase agreement and you cancel it, how much do you book as a saving? Well, we've tried to do some math of like, well, all right, how much was the average spend on that BPA over the last five years?

1:32:52And however much we haven't spent so far this year, we can book the remainder of that as a savings. But the data quality in the government is so poor that it is effectively impossible to do that. And so when you're trying to figure out how you use this calculation, how much are you actually saving? It's really hard. And I can even give you a specific example at the IRS, right? So one of the first things that I did at the IRS is I said, hey, all right, we've been brought in to do contract review and to look at our budget. What is our budget? I heard$22.5 billion. I heard$10.8 billion. I heard$12 billion.

1:33:32I said, guys, what is our budget? it. How is it possible that I cannot get a straight answer to this question? It took six weeks for me to get a clear answer to what our budget even is. Part of the reason is because of these continuing resolutions where I think at that point, we still didn't even have a 2025 budget defined or something like that. This is all kind of outside of my wheelhouse. Part of it is the fact that a lot of the money from the IRS came from the Inflation Reduction Act, which had$80 billion that was given to the IRS specifically. And so accounting for that was very challenging.

1:34:10So we eventually figured it out. It was something like$22.5 billion if you include Inflation Reduction Act money. Great. My scope was in IT. How much of that is IT? I think it was$10.8 billion for FY2025. So then we worked from that and we said, all right, what are we spending on now? Which things do we count? Which things don't we count? I think we would conservatively say that we've saved$2 billion in contract cancellations. It's roughly correct. The reason why I say it's conservative is I think we're on track to spend something like$4.5 billion. And our budget was$10.8 billion within IT. And so you would naturally say, wouldn't that mean you saved$6 billion?

1:34:56But when you count savings, you don't count canceled solicitations. You only count the way that it's currently calculated. You only count money where a contract was signed, but the money had not yet been sent, which is like this very narrow window of contracts. So for all of the things that we said, no, we're not spending 400 million on that. No, we're not spending a half a billion on that. That doesn't even show up. And so this is one of those tricky determinations of what is saved. Is it one-time savings? Is it annual savings? The Inflation Reduction Act money has a time duration. It isn't an indefinite amount of money.

1:35:35It's for some set number of years. So how much of that do you count? Because you couldn't count it as annual because it's not permanently recurring. It's only recurring for a certain number of years. So all of this to say, it's a lot more complicated. And so when there are these journalists who say, well, we did our math and it only saved X, it fundamentally misunderstands what savings even means. they're using some of their own calculations to try to come up with like the most conservative, least accurate version of how much was saved. And so I think my opinion on this, and I know people are still going back and forth on this, is really the amount of money you saved is however much you were planning to spend minus how much you actually spent.

1:36:22But you're not going to know that until the end of the year because these things get reconciled at the end of the year. Well, you're certainly not going to know it if the budget's never actually defined. How do you know that you're not cutting stuff that's important? If the priority is we need to have cutbacks, presumably one of the huge concerns is cutting shit that you actually needed. Yeah. I mean, in some organizations, they are higher risk than others. In some organizations, you can cut things that are important and just put it right back. and nobody even knows. That's not the case for the IRS.

1:36:59That's not the case for something like the SEC. If you mess something up, it can have significant market moving effects. Like when I was talking to the team about some of the changes we wanted to make at the IRS, we were calculating the cost of really bad decisions. If we totally blew something up, the amount of money that we could cost the American public would far outstrip any of the savings we can make in doge just because we collect. Do you have to be gentle? You have to be very thoughtful. It's a scalpel. It's not a chainsaw. And so we went through really diligently line by line on every one of these.

1:37:38What is it for? What does it do? What is the risk? And a lot of the people, as I mentioned before, especially the initial leadership team just said, everything is critical. Everything, you can't cut anything. In fact, we need more. Nothing can be cut and we need more. And when we swap them out for people who were more in the weeds, who knew what these things were, we found actually quite a lot that we could cut. And it's not just cutting. It's really this reallocation. There's this historic problem that we have where we assume the vendors are going to save us and we just keep repeating the same cycle, which is we give vendor X a billion dollars to finally modernize this thing.

1:38:16It's now years, it's years behind schedule. it's all just gone up in smoke and we say oh wow that was bad maybe this other vendor will fix it and then we do that for five years and then they just blow it up in smoke and we just keep repeating that cycle over and over again why is it important to cut spending in the government well the most tactical reason is that if you run out of money bad things happen the amount of leverage on that is pretty significant especially if you can do this across all of the departments i think the other uh the other thing that we've observed just specifically within the irs is that uh by spending less money the organization is running much better because we have fewer of these competing vendors who are building totally disparate systems that don't communicate with each other.

1:39:12This kind of ties back to the earlier incentive problem we discussed, which is if you're a vendor, your incentive is to lock in. You want to create a custom something and you want to make sure that we can - Only you know how to work it. And I see the archeological evidence of that everywhere in the IRS. The compliance team told me early on they really need help with their software systems. They said we have at least 60 different systems that do not communicate with each other at all. I assume this was an exaggeration. I spent two full days now at the compliance facility just watching how these systems work.

1:39:52If anything, they're underestimating how many of these systems there are. In software, when you're building these types of systems, one of the just core principles is you want to have a single source of truth for any piece of data. This is foundational. Sometimes you might have two sources of truth if you're doing like a really messy migration or you have some caching layer and you're kind of balancing those two. But really one is the goal. as of last count, which I think is last week, the IRS has about 108 competing sources of truth. I'm not talking about data sources. This is like the most complex Rube Goldberg I've ever seen in my life.

1:40:40It's just these systems don't communicate with each other. So there are some specific examples that the customer service team showed me where if you want to know what somebody's current address is, you have to check a whole bunch of different places and they just have to use some judgment on which of these is actually the current address because there are so many different places where we store this information that don't connect to each other. So this is our foundational number one priority is data quality. It's just we need to get to the point where all of our systems are using the same information under the hood.

1:41:17We just lose stuff. We have all these data integrity problems. This should have been a thing that we did 20 years ago, but we're finally prioritizing it now. If Doge is saving billions, because you're one cog in a big Dogey wheel, if Doge is saving billions, how come Americans aren't feeling these savings in their everyday lives? I mean, the tax rate's the same, so i don't really know it's uh if the doge dividend does eventually happen they will see something from that um but i don't i don't know the status of that that's not really in my in my wheelhouse i think the uh my hope is that if we are successful specifically in what i'm focused on at the irs that when you call the irs somebody will answer you can go to the internet to solve problems.

1:42:13One of the big priorities that we have is digital correspondence. This is something where right now, when you get a letter in the mail from the IRS, oftentimes we don't, you can't just go to the internet and like see that letter. The only evidence that you got it is the piece of paper. And then to respond to it, you either need to fax it to us or you need to send it by certified mail and then we may or may not receive it. And you have no, the feedback loop is like months. And so this whole thing, it's just such a mess. And so one of the big focuses right now is you still get a letter in the mail and it says, go to irs.gov to respond.

1:42:55You can go to the internet, you can log in, you can see the message, you can respond to it. You can upload documents on the internet. This is basic stuff. We don't currently have this. So instead of that, what we have today is the call to actions. Give us a call. You sit on hold for some number of hours. And then they say, great. Yeah, we need some more documents. Can you fax them to us? It's like this whole thing is such a mess. And so I think if we're successful, a lot of those things will start to feel like the government is able to solve basic problems. Is there a fear that modernization or updating is just code for breaking stuff that already works but imperfectly so uh i i have one of the other cio memos that i'm writing is about um why we need to stop doing modernization just as a concept, because I think it stems from a general lack of understanding of the software development process.

1:44:03This whole concept of like, we need to modernize, we need to throw a bunch of money one time at solving this problem and then never touch it again. I've analogized this to some friends of mine who are not in the software development space, where if you have two ends of the spectrum, one end is like building a house where almost all of the work goes up front. And then there's like this trickle of maintenance that comes afterwards. On the other hand, it's like getting a puppy where there's some extra work at the beginning during the puppy phase, but you can't just not feed the puppy for like a year.

1:44:37You can do that for a house. You can just like forego maintenance and it's not just going to immediately collapse. But software systems are effectively living systems that require constant maintenance and updates where you get feedback from the people using it. And they say, hey, this thing doesn't work. Can you fix it? And then you fix it. And they say, hey, what if this thing was over here? What if it added this thing? What if it pulled in data from here? And you're adding to it over time. Our current software development process in the IRS, and this is another one of those things that I'm spending a lot of my time working with the team to try to reorient our thinking on this actively is normally we'll spend a billion dollars, five billion dollars, some number of billions of dollars one time on building a software tool.

1:45:23And then the explicit directive is never touch it again. It is now done. Now we only do new stuff. That is now a legacy product. Now we're going to build a new system to replace that old system. And then, of course, what you end up with is just two systems. And then you do this again. Now you have three systems. And that's how you end up with 108. Yeah. Yeah. And it's just a total mess. And it's not the way we should do this. So I think the question that you had of, like, is modernization ultimately even a good thing, I think is totally valid. And it's not one that gets asked enough. the question about like mainframes are not strictly speaking bad.

1:46:01And the way that we define legacy, we actually changed this a few months ago, but historically the definition of legacy at the IRS is just anything that's 25 years old or more. And if it's legacy, you've got to redo it. And we have, there are systems that are more than 25 years old that are fine. Just leave them alone. In fact, they're more than fine. They have been battle tested. I was going to say, if they've stuck about for this long, we have a rule with my friends that, um, we always get excited about new life hacks. Oh, this is the new, you need to try obsidian dude. The best new note taking, whatever the, this new meditation app, this is a way to make a toasted sandwich in a fucking toaster.

1:46:41Um, and if the person that's proposing it hasn't been using it for at least six months, it doesn't count because yeah, you've got shiny object syndrome about this new toasted sandwich process. But what you didn't account for is that after five toasted sandwiches, it breaks the toaster. Yep, absolutely. Yeah. And it's absolutely true in this case. Like for some of these things, these mainframes are totally fine. In fact, they are maybe the optimal solution for this problem. The thing that mainframes have been optimized for for half a century is batch data processing. And you know what we do a lot of?

1:47:13Batch data processing. It is maybe the perfect solution for that problem. the challenge is when you have other things that you need to do where that gets in the way of it and so for individual master file and for a lot of the functionality we want at the irs the fact that it's based on a mainframe does cause problems because there's a there's a latency and there's a delay in getting data because a lot of these things you end up getting weekly batches and you can't have a week delay on these things and it's really hard it's when you have such complicated tax logic, it's really hard to get data in and out of the mainframe.

1:47:49And so you have these issues where this is how we ended up with what we often refer to it as shadow IT. You have all these different IT orgs that have basically created their own universe within IT that doesn't connect to anywhere else. And so it's very hard to get that information to places where they need it. shadow IT is also this is an interesting symptom of the type of dysfunction and I think this might also exist at very large corporations where because IT as a function has failed to deliver for so long each of these subcomponents like the compliance team the customer service team they've basically lost confidence that whenever they need something done they can ask IT to do it So what they do is they create their own secret IT group with their own contract, with their own team, and they build their own product that doesn't connect to the rest of the universe.

1:48:44And then compliance does that over here. And then automated underreporter does it over here. And everyone ends up building their own shadow IT org. And you can see how it happens because if you're responsible for delivering on something and your counterparty cannot be relied on, the only option you have is to do it yourself. But then the cost of that is hyperfragmentation, which then causes all these other downstream problems. And so figuring out how to make the IT org functional, I think, is front and center for all of these other problems we need to solve. i want to talk about security you know doge has had a lot of criticisms there's 300 million american citizens bits of data that have been put up in the cloud this is contributed to by stories about hackathons which have got scary names and make people think that it's yeah being flippant with sensitive data and stuff like that um is there a risk that move fast break things in the US government is a strategy that is too dangerous to do?

1:49:56Basically, does speed really matter if it risks security? I think that, so we're, even in the current implementation, pretty far from move fast and break things. Almost all of the work is being done by people who've been there for 10 or 15 years. It's certainly moving faster than has been done historically, but the bar is so low for how fast things have been done historically that making them go faster is, I would say, pretty low risk. I think there's a lot of fear mongering about these things, about how this data is used. It's going to the cloud. It's already in the cloud. It was in the cloud before we got here.

1:50:42A lot of these things come from a misunderstanding of what it is. I've had to answer many, many questions. Part of this fragmentation problem we have at the IRS of our data quality issue is we have something like 15 different fragments of APIs. APIs are basically just how two computers talk to each other. And we have all these different systems that don't communicate. And so we're building a unified API. So there's a central place within the IRS where if you are building an interface for one of the compliance teams, you can see the same data that customer service sees. So when they're on the phone with that person, they can have a conversation about the same piece of information.

1:51:26But because most people don't know what APIs are, there's been a whole bunch of fear-mongering articles saying they're building a mega API to do something. It's like APIs are how computers communicate. Every company has API. They're using scary words, and I don't think they know what the words mean, because if they did, they would just know this is normal. This is so normal, it's a complete non-story for any software engineer. How secure were the systems when you first got in? Did you have any concerns around security? It's very dependent on the organization. So IRS, pretty good. treasury pretty good some of the other agencies where they don't really think of it as their core competency um not as good um i i would say it definitely could use a higher priority the the core challenge with a lot of this is that cyber security engineers are in very high demand and It is super hard to attract senior cybersecurity leadership.

1:52:39With a salary cap. Yeah. There was a really famous job listing. I think it was the UK head of cybersecurity. And the annual salary was, I want to say, about 65 ,000 pounds. Something to be about$85 ,000. Yeah. You're talking about a rounding error on countries that want to get access to our cybersecurity. or people that just got Bitcoin at a hundred bucks. Yeah, yeah, yeah. That is not a difficult amount of money for you to counterbalance if you want to try and turn somebody. Yeah, that's exactly right. And the core challenge there is I've been trying to recruit people to be in the chief information security officer role across a lot of these departments or agencies within Treasury.

1:53:26And the challenge is that it's a massive pay cut The other thing is this ties back to people say, all right, great. You know, I've got, I've got 10 people on my team that I think would come with me and we can get this done. It's like, oh, you can't do that. And also if you don't like the people on your team, there's nothing you can do about it. And the, just the, the number of handcuffs that we put on them, even when they get here. The other thing that's, I would say, especially challenging within cyber for a lot of these agencies, especially the ones that need it, is the best way to destroy your career as a chief information security officer is to have a major cybersecurity breach under your watch.

1:54:15The problem is if this was the private sector and you came in, people will care about reality. So if you came in and you changed around the systems, you started observing things maybe for the first time, and you discover, oh, the Russians are in our system. And they've been in here for three years. You would then immediately ring the alarm bells in your company and you would get them out and you would be praised as a hero for having found it and removed them. But because this is politics, what's going to happen is they're going to say, oh, it was under your watch that the Russians were in there.

1:54:51And you can say, well, what do you mean? They were in here before I got here. Doesn't matter because it's politics. That doesn't make the news story. It's like there was a massive cyber breach under such and such administration. No, they discovered a massive cyber breach under such and such administration. Yep. But that's not, the tail wags the dog here because that's not, that's not the news story. So it's very challenging to convince them to be in this role because especially if they're coming into some of these smaller agencies who have had real challenges with this, you're rolling the dice on it.

1:55:24In fact, if you are successful and you find these things and you stop them, you better hope you're like at your retirement age because it might be a lot harder for you to. All right. That's the encore to your career. What do you say to people who accuse Doge of being unvetted and unchecked?

1:56:11I mean, they've said that about me. able to enact change and make decisions, hire, fire, spend, save, cut in a manner that other people bypasses some of the Chesterton's fence that probably should have been left there? I can really speak mostly just to Treasury and IRS because that's where I have the most visibility, pretty much anything that you want to get done, you have to go through the career staff. Like there is no, it's very challenging to like find the button to push because usually by rule, it's done by a different person. And so there are definitely times when there are some normal set of clearances or steps that you'd have to take that are not actually required, but they've just been done sort of by ceremony or something like that.

1:57:04And you can shortcut them. But typically, at least in our case at Treasury, this is done with the Treasury leadership team bought into this. They say, you guys are responsible for looking over this, and then that's what we do. So it's very hard to fire people. The only way that you can really reduce the size of government is through the reduction in force process, which is a big thing that the Doge team is pushing for. What is that? So a reduction force is the, it's really the only mechanism historically to have like a sizable reduction in the number of people employed by the government. Tech companies are doing riffs right now, a reduction force.

1:57:51And the most common methodology you would use in a riff is you find who your low performers are or maybe your relative low performers and then you cut them. unfortunately in the government, it is almost the opposite of that. Because of all these extra rules, the people who stay are there based on tenure and veterans preference. And so you're almost guaranteed, and this is to your earlier point, why do we have so few young people in the government? I think it's 7 % are under 30, whereas in the private sector, it's more like 20. wow okay so it's way less it's like a third of what you typically see um a big part of the reason for that is when you do these reductions in force it's basically tenure so it doesn't matter who your top performers are it's effectively irrelevant to a riff the only way you can do it is you take if you want to do a 20 riff you take your 20 youngest people who are often your top performers and then you got to remove them.

1:59:00And it is a really painful process. And there's this whole thing, you have to create a, you have to look for competitive areas. You can do a 100 % riff of a competitive area. It is this bizarre, cumbersome process, but kind of tying back to the steel man, which is you don't want riffs to be that easy if your goal is to retain civil service because you don't want new administration. You don't want to revert back to the spoil system. But I think most people would probably say that performance-based rifts is probably good instead of tenure-based. I think, I'm sure that's something people are pushing for, but that would make a really big impact.

1:59:48Doge removed at least some non-zero number of government employees. Yeah. Yeah. Talk to me about the process there and the fallout. So some of that was through reduction in force. Some of that was through DRP, which I think stands for Deferred Resignation Program. It's basically asking people to voluntarily resign for severance. we had I think about 25 ,000 people at the IRS take DRP out of I think we had 110 ,000 when we started something in that range the in in the IRS we haven't seen that much fallout yet because really we had increased our staff to a historic levels. I think the IRS historically has something like 80 ,000 staff and we had increased it well over 100 ,000 in the last administration.

2:00:50So this isn't like a huge change compared to that. I think the biggest thing that we're focused on primarily within IT is this recomposition. And so trying to make sure that our technical staff is technical and the more we can do to encourage the people who are ostensibly engineers but are not actually capable of doing their jobs to move to a different part of the organization find a job elsewhere i think that would be uh that would be a net benefit what are some of the errors or biggest blunders that you think doge made like if you could go back and if you were king of doge what would you go back and suggest that they change?

2:01:37Are there any sort of faceplant moments or real sort of unforced errors that you wish hadn't happened? I know of some secondhand of like specific things that were done that upset some people that were just unnecessary. But I really think the biggest one fundamentally is that at a certain stage, we really should have brought in more people into the conversation, there became this media narrative. Them and us. Yeah, them and us. And like Doge is doing these nefarious things. But anytime I would bring somebody in and say, all right, these are my priorities. They're like, whoa, that's amazing. Like these are all of the things that we want done.

2:02:16I didn't realize that you were involved in making that happen. But there was a degree of secrecy that I don't think was useful. yeah i mean it just plays into bad optics yeah right that yeah it is young adhd you know super intelligent infants coming in and and vibe coding their way through very precarious government systems and it plays into a broader narrative that's kind of a skepticism about the rich at the moment um i was having a conversation with george mack yesterday and we were talking about how um elon for all that people have huge criticisms about him he continues to deliver largely and his most recent one trillion dollar compensation package thing um sounds absurd and obviously is but is basically predicated on him reaching another stretch goal that is so insane you know it's autonomous robots in every home within this time and it's 10xing the value of tesla and it's this and this and this and this and this you think well if the guy does it then that's that's fine and you know i think a lot of criticisms can be hidden uh rightly if a lot of concerns around someone can be hidden rightly if that person continues to deliver which is why if kanye drops a new slamming album people will probably still listen to it and in this very bizarre way even though this shouldn't be the way it is he's redeemed for transgressions of the past simply because he's good at the thing that he's supposed to do it's like kind of just a bias that humans have got in this way so you know there's this broader narrative at the moment peter teal i think has maybe taken the new mantle of kind of like tech oligarch billionaire man that is the scary lizard person accusation type thing.

2:04:15Marc Andreessen's in there, but he's not a big enough name. If when you say this person, the person that you're explaining this to goes, who? That doesn't work. And if you're not sufficiently VC-pilled, you don't know who Andreessen or Naval or whatever these people are. My point being, there is this broader narrative at the moment that's kind of like the tech billionaire critique 3.0. You know, you sort of had the first wave and then you had kind of Zuckerberg, but he's had a rebrand. He's got curly hair and an oversized t-shirt now. So he kind of doesn't really work. And like, what's he really do?

2:04:49He just spends loads of money on Apple engineers and steals them. Like, we don't really understand that. And then, well, like Bezos is just kind of like in his Conor McGregor arc now, it seems. You know, he's got a hot new wife and a wedding in Tuscany and he spends all his time on a yacht. And like, and I like Amazon, like even though it's kind of, you know, a bit dodgy and stuff. and Bill Gates, you know, like fucking, what's he even do? Like, we're not really too sure. Like, so I think we've left that world behind and now we've got this new one. And the protection that at least as far as I can see, Elon has had and has is he keeps delivering on stuff.

2:05:23Like lots of people like Teslas and they are the best electric car as far as I can see. And he keeps on doing stuff that kind of is good for the planet, at least in some pocket. Lots of blast radius of this stuff that I'm super uneducated about. But there is an argument to be made that his push for EV has done an awful lot to improve climate change, even as people sort of throw the usual accusations of blah, blah, blah. My point being, if this is the world that you're coming into, the optics of we welcome the input of people that exist here already, we will stick by the rules of decorum to some degree.

2:06:08And I understand that if you come from the world of business, I don't have a uniform that's required for the guys that work for me. I want them to be comfortable and to just maximize the way that they operate. And they understand, like if they're going to turn up on set, they're not going to be in their pajamas, right? But like largely that doesn't really matter. When you get to government, there is a sense of sort of pomp and circumstance that kind of people need to, the hallowed halls of that run this great nation, et cetera, et cetera. And I think that these two worlds clashing, this sort of skepticism around move fast, break things, tech oligarch, billionaire, tendrils going into, this is part of a tech-enabled shadow government run by adult infants on stimulants.

2:06:51With, we need to, there are checks, balances, rules and procedures that exist for a reason, like Chesterton's fence at scale, right, at governmental scale. And I think that, at least from as far as I can see, the optics game has not been played as well as it could have done by Doge. And what that's resulted in is people that say, like an Elon approach, like, don't look at what I say or what other people say about what I do, look at the outcomes that I get. um but unfortunately politics what is it like politics is different yeah some politics is an optics game and how what is it some insane percentage of uh congress people's time is spent fundraising and going to dinners and lobbying it's like more than 50 or maybe even like 75 of their time isn't doing the thing it's doing the thing that enables people to pay them to do the thing or support the thing or whatever the bullshit is sure and um yeah i get how we just need to focus on this is a big fucking mess this is a big octopus and i just a gaudian knot and i just need to take a a scalpel to it a flaming scalpel and slide through the middle of it um but yeah at least my total you know bro science off the reservation uh perception of what's going on at least that contributed to it in a pretty significant way i think the optics have been quite poor and you think am i more concerned with how i look or what i do i only have so many hours in the day and so many resources that i can deploy why should i put it um just you know if you could run it back i'd be like hey let's have a director of doge communications and and brand and have like a little team of people that go maybe we don't say that maybe we shouldn't tweet tweet that maybe big balls maybe that's not you know can we scrape that from the internet somehow like is that a way that we can, you know what I mean?

2:08:46Yeah. Yeah. I think the, the interesting thing, having spent obviously a lot of time with these, the young Doge engineers, including big balls, including big balls. Yeah. These are extremely smart people. I think this is, uh, the, the perception of these, that they, I I've seen some people, some politicians call them infants as well. And it's, it's interesting because the founding fathers were roughly the same age when they wrote the constitution and led the revolution against the British. What age? In their early twenties, typically. Wow. Early to mid twenties. These are not children. These are adults who are very capable people.

2:09:29I think they might be perceived as infants because our governing class is typically in their seventies and eighties. And so to them, it seems like they're infants, but, uh, they're generally very thoughtful people. who are challenging a lot of the institutions and systems that have been calcifying over many, many, many decades. And that can be uncomfortable. And one of the things that we often will push on for people is, if we go back to some of these examples, like the hiring process that I described, it's like, where does it say in the law that this is what we have to do? And it usually doesn't, but everyone sort of thinks that it does.

2:10:14Same with like the fax machine thing. Where in the law does it say we have to do faxes? And nobody can find it because it probably isn't. It probably isn't actually a law. It's just many, many, many decades of inertia that has been built up that requires somebody to challenge it. I can't imagine that being associated with Doge in DC gets you a particularly warm welcome. You know, it's an interesting, I haven't had any issues. Even like within the IRS, I really, I've had a couple of circumstances where my first interaction, they react in a sort of frightened way because of everything they've heard, you know, the hackathons, all this stuff.

2:10:58And then usually pretty quickly, they realize that none of that stuff is actually real and that our goals are largely the same on how to solve these problems. Yeah. It's just, it's been, it is interesting to see how different perception can be from reality when you're in government. Was there actually a challenge in moving Doge from being a meme about a problem to actually solving a problem? Or was it solutions first sort of from the beginning as far as you could see? I mean, that's I think a meta question that's like outside of what I've been working on. I think in my particular case, like we were told, look at these contracts, figure out for me personally, why is this modernization program so bad?

2:11:46How do we fix this? How do we land the plane before the end of this administration? And so when you go in and you talk to people, a lot of the people, especially an engineer, the engineers on the team, they want to solve this problem. They don't feel good about the fact that this thing has been ongoing for 35 years and will probably never get done. They actually want to solve these problems. I think from the outside, what a lot of people will assume about Doge is that you guys have come in and are at odds at loggerheads. There is this small pocket of people drinking new tonic in the fucking corner of the office, the new bit that they're allowed into.

2:12:27and they are in conflict with the rest of government. How much from your vantage point, and then I guess anyone else is outside of the treasury too, how much has that been, wow, this is enabling and this is good versus this is threatening and can you fuck off, please? I think it very much depends on the agency. Have you heard of any that are particularly cantankerous? Oh yeah, for sure. Can you say? I think the ones that know deep down that their entity serves no purpose and they know that if we found out what was going on, we would seek to eliminate it. Those are the ones who are particularly - Taking a clock.

2:13:12Yeah. Like they know, from what I understand, I was not here when the USAID thing happened, but from the people I've talked to on the ground who are actually looking into it, there was no intention to shut down USAID when it started. They just knew that there was a bunch of problems. They knew, which is better grammar. They just knew there were a bunch of problems and they came in, they started talking to people, they started looking at the books. And then over time, they realized this whole thing is a giant ball of worms. Like they were, they first came in saying like, all right, what do we cut?

2:13:46What do we retain? how do we fix this? And very quickly they realized like this whole organization is a problem. And they've seen that at a few other agencies as well, where, uh, I'm, this is kind of obvious, but it's very different in my experience at the IRS. The IRS serves a very obvious purpose. Um, and so there was never a question of like, do we just blow the whole thing up or not? Right. Yeah, yeah, yeah, yeah. I understand. Okay. What does the IRS actually do? Yeah. Well, so this is an interesting misconception. So the IRS, functionally, it's basically a software company. Almost everything we do involves a lot of software.

2:14:31Very little of what we do actually involves collecting taxes. The Bureau of the Fiscal Service collects most of the taxes. The best way to think about it is BFS, Bureau of the Fiscal Service, is more like Stripe. They do payments. The IRS is more like QuickBooks. It says it all that you're having to use. Yeah. Those analogies. We do very little in terms of - You're a big ledger. It's a big ledger. Yeah. And we check to make sure that the amount of money that you say that you made is the same as the amount that we know you made. And if those numbers are different, then we send you a strongly worded letter.

2:15:16And if they continue to be different for very much longer than that, then there's an enforcement action taken against that person. Are you guys in charge of the enforcement thing too? Yes, to an extent. We have a criminal investigations group. There's a whole compliance team. The compliance team is quite large. I don't actually know at what point like DOJ maybe gets involved, but there's a pretty significant compliance arm. It's one of the major functions of the IRS. How does the IRS stack up against other parts of the government in terms of efficiency and efficacy? Does it do what it says it can do and how easily does it do that thing?

2:15:54So some of the things that it does very well are things like compliance because there's so much leverage in it. There's effectively an infinite amount of money to make sure that we are collecting the proper amount of taxes. We have some very sophisticated teams and people. Part of the reason why I think we do is also that if you are interested in solving these kinds of problems, there's really only one place that you get like$5 trillion worth of leverage on this stuff. So for like a really good data scientist, this is a really, really interesting problem. People will take a pay cut to work on this kind of problem because it's very, very interesting.

2:16:32Where we struggle is in a lot of the software systems on, especially with customer service, individual online accounts, a lot of that taxpayer facing stuff has just generally not been very good. Are you in charge of corporate tax too, or just individuals? Yeah, all taxes. Wow. Okay. So this is huge. So you guys, I mean - The obvious challenge, the obvious interesting thing here is, I think, many commentators online, many political parties run on increasing or reducing taxes. Tax is a massive part of a lot of different political messaging and also the sort of thrust of on the ground activists. at some point that needs to come into contact with reality so we want to raise taxes for the most wealthy people we want to cut taxes for the middle class we want to streamline the tax system just have a flat tax and you know cut out fucking 95 of the people that have to work for you because it's all super super simple yeah but at some point that has to be enforced so you can have all the policy that you want and then it arrives presumably at your desk or one of the guys in your very small cohort of C-level people's desks.

2:17:58And they go, this is what's happening with tax rates. Change the code on the ledger so that we know that that's happening. So just dig into the specifics of tax policy, the talking points that people get to play around with, when that meets the road. Yeah, I can actually give a very specific and timely example. So in the one big beautiful bill, which is typically referred to as OB-3 internally, when OB-3 was passed, it has a bunch of changes to the tax law. The first step is it has to go to the Office of Tax Policy and a bunch of other lawyers to interpret what it means and how it will actually work in practice.

2:18:42So there's this whole step of the lawyers have to say, all right, what is no tax on tips? What does that mean? How does that impact these? The thing that's really interesting about a lot of the stuff that IT does is that a lot of what we do, the first 80 % of the work is usually pretty easy. It's that second 80%. Everything gets very challenging. It's those edge cases. There's sort of an internal asterisk to everything, which is like, this is all fine, but what about the Amish? It's like, well, they're not going to send it on computers. So we have to have some backup for that. And it's like, they're more of like a placeholder for all of the edge cases that you have to be aware of for people.

2:19:30Everyone, it needs to be, I'm reminded of a friend of mine who worked at Google. He's pretty high up at Google. And I was asking about like, why does it take so long to get things shipped there? And he said, he had a conversation with Eric Schmidt. And Eric said that, remember that every problem we have is a problem at scale. There is no small problem at Google. And it's the same thing at the IRS. Every change to the tax code we make affects a vast number of highly diverse people. And you have to account for every one of those possible edge cases. And so that's the core challenge and why it's slower to do these things, because you have to account for all of these things.

2:20:15There are people who are blind. There are people who are deaf. There are people who don't use computers. There are people who just aren't capable of using computers. There are so many different edge cases you have to handle for. There are people who don't have a social security number. People don't have a driver's license. There's every possible edge case, you have to figure out how to account for it. A lot of that gets done in the policy realm where they say, for all of these things, how are we going to handle those things? What is the cap for how much is exempt for no tax on tips? What is the interpretation of the statute?

2:20:49How will we actually implement it in practice? At the end of the day, for almost everything where the rubber meets the road is in technology. And so this is an interesting thing that I did not really expect in my role. I thought it would be like IRS modernization is the one and only thing. But But what I've discovered is that in almost everything that you want to actually do in the government in terms of full implementation, almost everything touches technology at least a little bit. So I end up engaged in conversations on the Trump accounts, which are coming up, which is$1 ,000 to every newborn American.

2:21:31Okay. Yep. That's another thing. There's a whole bunch of policy stuff, but eventually computers are going to be the ones doing it. We've got to figure out how to do it. Almost every policy change or anything you want to do eventually touches an IT system. So if you take OB3, the tax policy folks are going to come up with their interpretation. They're going to work with us on the implementation of how do we actually do it. Then we have to turn that into changes to the core mainframe system because that's where a lot of the business logic lives. Then you have to figure out how do you change whatever interface you have?

2:22:09Does this require change in the tax forms? That also requires a change in technology, both to print the forms, to send out the forms, to receive the forms, to have the digital forms. How does that connect to all the other things? The tax code is very complex. And so there are so many intermingled parts that you have to solve for. So that's like a thing right now that we're working on. It's mostly sitting right now with the tax policy folks, but we know enough about where it will likely land that we're starting on the implementation now. I suppose the interesting thing about tax as a person who is a super fucking noob about how this occurs, tax has to account for a big amount of income for the government.

2:22:55Do you have any idea how much that is as a proportion of where money comes from? It's almost all of it. Right. Yeah. Okay. So you are the guy who is technically, technically in charge of the systems that acquire the revenue that the country needs to run. You could put it that way. Yeah.

2:23:20that seems like a big deal yeah yeah it's uh the amount of leverage in the role is very interesting in both directions in both directions yeah it's pretty high stakes um the good news is that there are a lot of people working on this who have been working on this for a long time and so i've only been there for six months but this is maybe to give credit to the civil service system that we have, a lot of these people have been there for a long time. The trains are going to keep running and things are going to keep getting done. And even though I've only been there for six months, I can have leverage one way or the other, but there's a very low probability that any one person can come in and really cause major problems in the organization.

2:24:14Mostly so far you've spoken about like implementation, it's been a little bit more dirt than clouds, I would say. Yeah. When it comes to the cloud side, at least in IRS, is there an argument to be made for simplifying the tax code? What would you do? Let's say that you were in charge, which you kind of are, but not in charge of that bit. What do you think would be some good solutions? I mean, everyone largely agrees that a simplified tax code is better for... How complex is it currently? Maybe the most. It might be the most complex tax system in the world. I would not be shocked if that was the case.

2:24:57I spent some time in Estonia a while back, and they have probably the simplest tax code in the world. It's a flat tax. If you make this amount of money, this is how much you pay. If you make this amount of money, that's how much you pay. I think the average Estonian spends two minutes per year doing taxes. It's like, how much did you make? Multiply that by 0.2. That's how much you pay. That's it. That's the whole system. Our system is massively more complex than that. The problem is it's very hard to do that. The IRS also doesn't do policy. So it really has to come from Congress. They have to be the ones who are willing to do it.

2:25:37Yeah, but you're the guys that have got to work out how to code it Right. Into the ledger. If they could simplify it. They've added another bylaw. This is going to take ages. Oh yeah. You talk to the people who've been here for a long time. That is a thing they talk about a lot. Like whenever there's a change to the tax code, because this often comes pretty late in the game and you have like a very short timeline because you have to deliver it by roughly February 1st. And so if you hear in September, like, Hey, by the way, massive change to the tax code. good luck it's uh it's a struggle to get it done but i think historically they've done a pretty good job of getting those things done it's so funny i just i keep thinking about the um sort of flaming sword wielding politician or online commentator talking about we need to tax the rich yeah i need to tax the poor or we need to simplify the whatever and um there is just this total ignorance around you and what you do and the 12 000 people that in some form or another have to make that happen yeah and uh that's one of the main realizations that i'm taking away thus far is that uh yeah the rubber really does meet the fucking road at some point and like what does this mean to enforce what does this mean to implement how do we change the systems on this what's the interpretation of this where i went to uh the moody center uh the big uh arena thing downtown here in austin and there was a automated uh corner shop type scenario you can get drinks and snacks and stuff like that.

2:27:26And you go and you do it yourself and you scan thing and then you pay. There's no staff. There might be one member of staff stood at the front who is just ensuring people like generally seem to be going toward the till once they've got stuff in their hands. But it asks for a tip. Yeah. This automated system asked for a tip. And I think to myself, well, is that exam? Is that a special type of tip? It's an interesting question. I don't know. Because who the fuck am I tipping? Am I tipping the coder? that made this system? Am I tipping the guy that stood at the front, the only member of staff that's physically here?

2:28:00Is this being fed into the servers so that the AI agents get a day off? What is this? I think this is exactly why these things are so complex is that is one of the edge cases that I'm sure the Office of Tax Policy folks had to think about. It's like, who is this tax exempt? For whom? Who knows? Where's it going? Where is it going? Eventually we have to figure it out. And I think your analogy of like in the clouds versus the dirt, I probably spend 90 % of my time in the dirt. Most of my conversations are with the engineers on how do we implement this stuff? How do we get it done? How do we change our systems?

2:28:40If we need to hire, why is our hiring process bad? We have this other thing. it's called an ATO process. I believe it means authority to operate. It's basically, imagine you wanted to use an iPad or an app in your iPad, whatever it is, you want to use a tool. Somebody needs to check to make sure that it is secure, that it meets our standards. That's the ATO process. The IRS has probably the worst ATO process in the entire government. I've cataloged a lot of other systems and how other agencies do it. Ours, I think, is the worst. And it takes oftentimes multiple years to get simple tools authorized to use.

2:29:25And so as a result, we're using code editors from 20 years ago. We're using - They're the only ones that have been proven secure previously. Yep. Yeah. I mean, I think about trying to speak to my accountant in America. And if I want to send documents, I can't do it through a ton of different services because she is at the mercy of her organization's ATO equivalent. And I think, oh, fuck, this is clunky. But, you know, at least it's secure. And I suppose there is a tradeoff between speed and security in this way. and uh yeah the systems do need to be i mean fuck me what was it like boxing day playstation in 2014 or something got hacked and like hundreds of thousands of people's data got diddled and then microsoft had one and recently in the uk marks and spencer that are sort of a up range supermarket um their online ordering seemed to stop almost immediately and then the shelves started to become empty and it looked like some sort of ransomware i don't even know if that ever got revealed i spoke to the bbc cyber correspondent the same week it was happening and he sort of like i can neither confirm nor deny uh but basically said like from what i've heard it's fucked like you've just got this big ransomware thing that's going on all of this shit's held you go oh it's impacted the supply chain like ordering logistics thing oh that's held for ransom and now you're just seeing there's fewer sandwiches on the shelves they're just dwindling and people are like, I need to go to Tesco because Marcus and Spencer doesn't have this thing anymore.

2:30:59And then the same week, Co-op got popped. And I think the same week as well, like Harrods got popped too or some other big high street, you know, department store luxury thing. So yeah, when you get this wrong, it's not like, oh, let's just look to the private sector. They nail this all the time. Like they fuck it too. Yeah, yeah, yeah. And the stakes are pretty high for a lot of these things in government. So I think there are things that exist right now that have helped with a lot of these processes. So there's something called FedRAMP, which is a set of standards that give broad authority to work within the government.

2:31:37So in the past, I've heard stories because I wasn't here in the pre-FedRAMP days, anything you wanted, you had to go through like your own custom process to like check to make sure. And then if you as the vendor, if you had a software product, if you wanted to go over to a different department. You had to go through this whole thing and it was just a major pain. I think it was GSA or maybe it was OMB, Office of Management Budget, created this centralized concept called FedRAMP, which is kind of like SOC 2, if you know what that is. It's a security standard. It's basically saying, we're going to verify that you do all these things.

2:32:11And if you do, you can be used across government at that level of security.

2:32:20FedRAMP comes in several different flavors, the ones that most of them use, FedRamp moderate, is not that bad. It's not that cumbersome for a company to use. So it takes some time, which I think is okay. But talking to people before FedRamp existed and how incredibly painful it was shows that there's some progress being made in these things. And that's existed for some time. Do you think it's healthy that taxpayers are skeptical of the federal government? Yeah, for sure. I mean, I think the

2:32:57general skepticism of how the power is used against people, I know that some of the skepticism comes from a world that I have really no exposure to, which is like the intelligence world. I think when it comes to the stuff that I have the most exposure to, which is within Treasury and IRS and how the government uses our money, I think peak skepticism is probably warranted. The, uh, there's an old joke of, uh, what's, uh, what's the best kind of military aircraft? Okay. And, uh, the answer is one that has a factory in every state and a part made in every district. Right. And, uh, I now understand that joke a lot better.

2:33:55How so? How does that relate to what you do? Well, I don't have as much direct exposure to it in the treasury, but I see it like friends of mine who are in DOD or friends of mine who are in all these different agencies. The, uh, all of the stuff that gets crammed into these bills from Congress are really where a lot of the challenges stem from. Um, it's, there's a, there's a classic example of this, the littoral combat ship, which is a failed initiative by the U S Navy to build a type of ship. It's pretty widely recognized as a major failure. Um, the S the simplified version of this problem was they, there was a simulation done that said, okay, at the end of the summit, we recognize that we need a very lightweight ship that can go in shallow.

2:34:48waters. Very inexpensive. Because in the future, instead of having a few very expensive ships that can be major targets, we need lots of very inexpensive ships that can be in shallow waters. Okay, cool. Makes sense. During the process of development, this is what I've heard, and I think this is all in government accountability office reports and things like that. They said, oh, well, my district makes missiles. We should put missiles on it. My district makes this thing. We should put that on it. Next thing you know, you have this completely pointless outcome, which is this actually quite large and very expensive ship that no longer serves a purpose for why it was originally commissioned.

2:35:32but many, many billions of dollars have been spent on it, which for some people was the goal, but for other people, we still don't have that ship that we wanted. So that's where a lot of these things end up getting stuck is in that congressional churn.

2:35:52How long would it take, do you think, if you were to overhaul the whole IT system for the government, like to fix this legacy behemoth and actually get it up to speed? I'll start by specifically talking about IRS, where I have the most exposure. I think the core, the primary fix is going to be solving the data integrity problem that we have. That is the first and most important deliverable that I would say that we have. We should be able to solve that within, call it three years, just based on the timelines that we're looking at. Assuming that we don't run into problems with procurement, assuming we don't run into all these other kinds of issues, we should be able to do it within three-ish years.

2:36:47Just about in time. Yeah, right. but there will be obviously incremental deliverables along the way that will solve a lot of problems I say like absolutely fully resolved by that point I think the fix if you want to talk about how do we actually fix it it's less about a specific fix to the software and it's much more a fix to the organization we don't have a budget problem in IT at IRS we really have a management and leadership challenge within the scope of that organization and have for many many decades if we can change the mindset from pay vendors a whole bunch of money to solve our problems and we can learn how to listen to our internal stakeholders, this has been my main focus for the last several weeks, is just because a system is old does not mean that it's bad.

2:37:56So it is hubris to think that a system that has had quite literally a million hours of labor put into it to make it good at this thing, that you can rewrite it from scratch and then cut over one day, and then it will suddenly start working. It's just completely and utterly unrealistic. And so how do we take the existing systems? How do we incrementally improve them? How do we change the culture to where we are building software in the way that software is really meant to be used, which is this, it's more like the puppy in the analogy earlier, where you're feeding it, you're training it, you're evolving it over time.

2:38:34You're not just throwing a bunch of money at it and then just ignoring it for 20 years. We need to figure out how to get into a software development lifecycle that's actually relevant for software. And what about the government outside of the IRS? Different organizations have different challenges. And so I mentioned earlier some of the data quality problems. I wish maybe we can do this afterwards. I can call my friend Clark. He's the CIO over at HHS. and he has been trying to figure out how to fix their payroll system which is i'll go into the specifics just because it's it's indicative of the kinds of problems that you see all over the place when you're actually like in the dirt trying to solve these problems um hhs they have 30-ish people full-time whose job it is is to run payroll and most private sector company.

2:39:35This is done by a company like ADP or Rippling or whatnot. There are companies that just do your payroll. Here, they have to request all of their records from all of the different subcomponents. I think there's roughly 40 of them. They then send over spreadsheets of all their employees and how much they should get paid. Those 30 people have to reconcile those to make sure they're all formatted properly. They then send that spreadsheet to a mainframe at the FAA to run a bunch of COBOL scripts, which I don't think anyone even knows what they do anymore because they've just been running. They're all on these calls.

2:40:13Just every time there's an error, they just manually reconcile it. They then take the output of that FAA mainframe. They then send that over to a DOD computer for processing. They then get the result from the DOD computer. They then chop it up and then they send that information to each of those groups and then they manually enter it into their financial system. That is how they do payroll. That is like the full-time job of dozens of people. And these things are not that hard to do with computers, but it's just always been that way. And so there hasn't really been a push to fix it because it technically works.

2:40:51Like they are able to do payroll, but that's sort of indicative of the kinds of problems that you see all across the government. There's just, there are so few people who are thinking upstream of why, why are we doing it like this? Why are we doing this at all? It's like the fax problem I was mentioning. It's like, why are we, why are we getting faxes still? What are we doing? It's like, I don't know. It's the way we've always done it. It's like, right. But these other ways are so much better. It's like, well, I don't know. it's too much of a challenge to change it, or maybe we don't think we can.

2:41:25It's not really clear. So to take that as like a specific example from HHS of your question of how do you fix it in government? It's not really about fixing specific systems. It's about changing leadership within IT, making it so that more of the people who are leading, more of the chief information officers, making them technical so that they know how to build software. That would be a huge improvement. Having more people in engineering who are engineers, that would be a huge improvement. If you have people whose jobs are to, if they are non-technical people whose jobs, which is a huge chunk of the population at IRS, if your job is to manage a vendor, but you don't know what the vendor's doing, you're probably not going to do a great job at managing that vendor.

2:42:20So I think that's the core shift that we need to have is in the same way that we treat the legal function, counsel's office, as having some set of standards and some expectation of what it is their job does, I think we need to apply that same logic to engineering across the government. So I guess I'm interested in, personally for you, you were famous, are famous for a lot of time tracking, complex system of executive assistant, virtual assistants, helping you do the things at levels and all the other shit you were doing in your life. How much of that has been able to transition over to this new life?

2:43:06Less of it than I would have liked. I think the biggest change in my current role in the government, historically, I really optimized for having a lot of deep focus time. My phone, for the better part of 10 years, just defaulted to do not disturb. and I would have consistently five, six, eight hours of just uninterrupted focus time. I am now almost as far in the opposite direction of that as can be imagined. I have multiple phones. They ring constantly and I've largely just accepted that I am not going to get very much focused time. I have set a goal that I want to try to contribute production COBOL code before the end of this year because I've never really been in a system like that before so I'm going to try to find a way to block off some time before the end of the year to do that but generally speaking very little of it translates Some of it is helpful in just being able to block off time for when I know I need to review things, making sure I have capacity to resolve the things that need to get resolved.

2:44:30Are you still tracking your time in the same way? Generally speaking, I track it in my government calendar. It's like my days now are mostly meetings. It's like meetings, phone calls. You were famous for not. For not that. Yeah. Yeah. For trying to minimize the number. And actually still, I tried to minimize the number of meetings already. And I probably have a solid six to eight hours of meetings every day. What about the army of assistants you had? Have you been able to bring those over with you? Nope. Nope. So you're a one man band now, largely. Yep. And I have, I do have somebody who helps me with scheduling, which interestingly is like the one thing I didn't use an assistant for before because I used Calendly or some similar scheduling tool.

2:45:19But - That's not OAP. It's not ATO'd. Exactly. For fuck's sake. Yeah. So you're telling me that inside of the government, there is not a calendar scheduling, this is what I'm available. And there might be one in government, but we don't have one at Treasury as far as I know. Presumably it would be APO'd for you if it was, right? Yeah, I think that we have one. I think that the GSA has one because they're a Google shop. We're a Microsoft shop. And I think Google has one pre-built into their calendar. Okay. Yeah. Microsoft needs to add that in, please, if that's okay. What's happening with levels?

2:45:57It's still going. Yeah. I tried. I was a little bit too ambitious with how I thought I could manage my time. and I thought I could be the CEO of Levels and take on this role at the same time. And after a couple months of frankly doing a very poor job as the CEO of Levels, I recommend to the board that we put my co-founder in as the CEO. And so he's been running it as the CEO. So I'm on sabbatical and he's been doing a good job. So I feel much better about it. It was really stressing me out because I would schedule time. It's like you schedule three hours to do my Levels stuff. And I'm in a meeting with the Levels leadership team and it's like the White House is calling me.

2:46:45It's like, shit, what do I do? It's like, you got to take the White House call. Like, what are you going to do? What was it like? You know, you've got this business, co-founder, CEO, people working, technology, new stuff, exciting. Isn't that cool? Guys, I'm going to leave to take 160 grand a year position inside of the treasury. Yeah. How'd that go down? I was... So part of the reason why... Because people had been asking me to consider this as far back as December, after the president won the election. And I said, no, because I'm running this company. We have a baby. We have another one coming in this December.

2:47:25There's so many things, so many reasons not to do this. and eventually with my wife's encouragement and then really actually the encouragement of our board, they said like, if there was ever a time to do this, this is it. We have enough cash. The major decisions that need to be made by me have largely already been made. The major decisions that need to be made outside of my scope, those people are still there. so if there was ever a time to take a sabbatical to do this this is it I think we also made some false assumptions about how much capacity I would have to still contribute to levels which I'm still I still have regular conversations with my co-founder now CEO so we're in touch pretty regularly on this stuff but it was originally going to be six months it's looking like it's probably going to extend through at least the end of the year because there's some there are a handful of things that we really need to get delivered before i i go back but that's that's really where it stands now how long do you think you're going to stay in this position the current plan is still january um what's the likelihood of you extending that again i think if you know if the treasury secretary or the president says like we need you to do this you'd probably say yes.

2:48:52Um, if it's just, if, if I no longer feel that I can be effective in this role, if it's just like trickling off in terms of my effectiveness, then I, I think that's fine. I think we've already, we've already made some very significant changes that would be long lasting. I think the longer I can do it, the better, but a big part of it is like, how long can you tolerate burning through your savings? There is a factor for that. And there's opportunity costs. That's part of the reason why I thought this would be a thing I do when I'm retired, not in the peak earning years. Right. Yeah. Because it's not just burning through savings.

2:49:35It's the loss in potential additional revenue for you. So I mean it is, you know, government servant. There is a big sort of arc of service here, especially when you compare it to being CEO of fast-growing, exciting tech company that can do wearables and health and stuff like that.

2:49:57It is a sacrifice in order to be able to do that. Should it be that way? Should it be a sacrifice to work for that? Should it not be a – it should stand on its own two feet as a position in and of itself? I think on the salary that you can make in these roles, you can make it work, but it's certainly a lot more challenging than it would be if you took an equivalent role in the private sector. That wasn't in D.C.? That wasn't in D.C., yeah. I wonder whether just from an influence perspective, you know, birth rates, not great in America, not great anywhere, but specifically in America at the moment.

2:50:35And I wonder what the downstream impact of the selection effect of having a bunch of people who really find it hard to have families being in government, who are the ones that are contributing to and thinking about these problems. because you are selecting out anybody who does have a family, or if not, they're having a family in a very particular sort of unique kind of way.

2:51:03When you think about sort of elected officials, they are representatives of the constituents, but they are unrepresentative of the sort of life that so many people lead. So you have this kind of detachment between the two. I wonder if you've thought about that. Yeah. It's definitely a real impact. And you see this especially in the leadership roles. That's where this divergence gets quite large, is once you get past this early to mid-career where the government can actually be reasonably competitive, people start to leave for the private sector. And those tend to be your most capable people, the people with the most context knowledge of the systems that they're working on.

2:51:51I think there's also this general issue where when I'm looking to hire right now, when I'm trying to bring people on for these appointed roles or these leadership roles, the only two categories of people that are in a position where I've been able to get them converted are either sufficiently young that they have no expenses or they are either at or near retirement. where their kids are off. The kids are already in college or off beyond college. They have enough money saved up and they want to do this to do public service. And it's effectively their retirement project. Those are really the only two camps that I've been able to convert on these things because the amount of effort that you have to put in, the opportunity cost of people who are in these peak earning years is too high.

2:52:52They have to relocate to DC for the most part. It's just too much of a lift to get people to be able to do that. How do you ensure that the changes that you implement are permanent because you leave, teachers got out of the classroom and the slow regression back to the mean begins all over again? Yeah. We talk a lot about how do we make sure the weeds don't grow back. That's a big part of the reason why I've agreed to stay for longer is some of the weeds would grow back. I can already see it happening. You haven't pulled the roots. Yeah, exactly. I already see it happening on some of the things that I've tried to change.

2:53:34And so I need to stick around for long enough to make sure that it gets solved. I think procurement reform is going to be a big one, a lot of personnel reform. This is one of the other interesting observations, because I've had many of these conversations, people talk about how it feels like the momentum is slowing. It feels like the amount of money being saved is all slowing down. And it's a surreal experience because from the inside, it feels like the exact opposite. We're just now starting to figure out how this actually works. It feels like we're only just now starting to figure out how do we actually fix the procurement system at its root?

2:54:18How do we actually make changes? How do we implement a better performance review system? How do we fix the hiring process across all of treasury? We're thinking more fundamentally instead of just cutting contracts, which is good, but it really feels like, from my perspective, somebody who's in the dirt for pretty much all day every day we're only just now starting to understand how to affect these systems in a way that will be permanent so it feels like we're uh there there are a lot of reforms some of these things require uh a a legislative fix requires an act of congress um i haven't really seen a lot of movement there but i know there are now people thinking about, as we make more progress within the executive, how do we start to look towards changes in legislation that will change these things in a permanent way that are even outside of what's available in the executive branch?

2:55:22Well, Godspeed over the next few months, man. It was not a predicted pivot. I think we went for coffee maybe eight months ago, something like that. And if you'd said, yeah, I'm going to be the secretary, I mean, the CIO of the fucking treasury department would not have been on my 2025 bingo card. Yeah, same. I hope you survive it and I hope that the next kid thing comes through and I look forward to doing the postmortem with you as well when you come back up for a... Oh yeah, let's do it. I appreciate you, man. Thank you. Thank you.