Friction Is the Feature

so actually, know, last time we spoke, Jennifer, we were talking about this unique problem of fax machines in the IRS and how they, I think you said they have like 60,000 of them or something. And there's a lot of people that would like to get rid of those fax machines, but that there's some sort of law buried somewhere that no one knows where to locate, but is preventing. these fax machines from being removed and I thought maybe we could start there because I think that really points to a lot of the problems that people in government face when it comes to AI adoption because there are things like I think you called it folk law. We were talking about human hallucinations. Could you set the stage for us a little bit and kind of explain that IRS problem? I think that's a great jumping off point.

Sure. Let me start by saying I do not think that there is a law somewhere that says you can't use fax. What I suspect is that, like many other situations that I've seen like this, back in probably the 1980s, there was a memo or

okay.

a regulation or some kind of rule that might have carried more weight of law or less, you know, that said something like this sort of information can only be transmitted to the IRS through a highly secure means. And at that time, somebody interpreted fax machines to be the most secure. And That got ensconced in what I call folk law or others call folk law, which means that people think that is a law. It is not. And it's very, very hard to figure out where that came from and why people believe that you can't get rid of these fax machines because the law would have to change in order to do that. I am super bullish on AI's ability to us detangle them cascades of rigidity, where what was meant by the lawmaker or policymaker or even regulator was actually a fair bit of flexibility and how it's being interpreted at the lower levels of the organization is extremely rigid and holds us back.

Yeah. Yeah. I think that ties into, this idea of canonical knowledge and we think about it a lot into in an organization, it's maybe a little less complex to figure out like everything that they know, even though the documents might be messy, there's maybe a clear chain of command and most organizations haven't been around as long as the U S government. So there's that problem too. But, that points to this, this real struggle of finding truth.

in government, right? Because if you can't get to the bottom of something, then where do you start

you know, I think the problem is finding where you're supposed to be looking, but if we can get better at that, we can find these answers far quicker than.

we have been able to in the past before. mean, LLMs are uniquely good at searching through very large volumes of documentation and actually understanding it pretty well. And if you need lawyers on the case to prove your point, for instance, that actually these fax machines probably could go, good luck, right? Like they may not be able to, it may take them a long time, they may be inclined to take a

very risk averse approach to it. other words, well, if I can't find any positive evidence saying we don't need fax machines. to tell you that you have to use them, right? But if you're really working in the service of the public interest and you have, you know, LLMs at your fingertips, You can embark on these, you know, some very complex, you know, journeys to understand where this perceived law came from and debunk it yourself. It could be a huge enabler

of government just getting faster and more common sense.

Yeah, isn't that funny? Because on the other end, someone's like, why are you faxing that to the IRS? And they go, they make me do it. And they're like, really? That's so dumb. And the other side is like, why do people keep faxing us this stuff?

They make you do it. that's a great story. so my one of my friends and colleagues, Marina Nitze, who has a new book coming out, by the way, called Crisis Engineering. She tells the story of being I think it was in a sort of I think it was a state DMV somewhere She was going through a process with a public servant that had a whole bunch of stuff in it that just didn't make sense. And one of them was, I kid you not, we're not talking facts here, I'm talking, you know what we used to call ditto machines? those, I don't remember the formal name for them. Turn a crank yet? Or no, no, it was carbon copies. I take it back.

Is that the one where you had to turn a crank? Yeah, yeah.

Carmen, yes.

It was like, literally, like you had to the paper and you would type and then you make the carbon copy. And she kept saying like, why are we doing this? said, we, you know, we have to because it goes over this other agency and they require it. with the sort of the implication that someone had tried for many years, she went over to the other agency and was going through their process and said, so why do you have these carbon copies coming in? they said,

yeah, we hate this. It's just that the agency insists on giving us the information.

Hehehehehe

So much, every single government is just having those conversations and saying nobody actually wants the carbon copy.

Yeah, it's like the they committee. Who's they that make you do this? Nobody knows.

The VA committee, exactly. I mean, in some

ways, though I don't think anybody ever asserted that the carbon copy was required by a law, that sort of does fall in this category of folk law, where people just see it as immutable. And I think that's sort of the big shift I'm trying to push on is changing that. perception that it is immutable into, this is a world we make, as David Graber says, and we can just as easily make it differently.

it's, hard to figure out how to fit AI into government agencies for looking from the top down. It becomes very complex, but what you're describing is maybe public sector workers themselves could be the ones that start opening up the blood flow a little bit, right? Like if, if there's someone motivated to, spend some time digging in with an LLM and trying to find the source of a folklore or whatever, or find the information they need to debunk it. If that starts happening across agencies, that kind of has this bottom up effect of opening things up a bit, right?

Yeah. Yeah, I think it also like referring to Joshua Gans in the microeconomics, like his whole concept of AI is really good at taking out the friction in the middle and enabling the frontline workers. You could almost see it as like a why humans are bad at removing friction in the middle, you know, because because someone has to ask someone they work for, can we get rid of these? carbon copies or fax machines and they, don't know, like, let me ask someone else above them who then has to eventually somebody can make the decision. But that middle human area, replacing that with AI, keeping the line worker in place, but making them much more effective because they don't have this friction, right? which it's kind of makes sense. This kind of falls under his economics viewpoint of where will we affect this most? It'll be enabling frontline workers, reducing friction in the middle, connecting decision makers with the frontline better.

Absolutely. the future, that center of your future sort of depends on public sector workers deciding to use AI not just to make the system as we have conceived it today just a little bit better, but to actually change the system so that it meets people's needs and meets the needs of the moment more. You know, we're not yet really even using it that much to sort of make the current system better. So we've got some work to do. But people are pretty frustrated with how government works inside and outside. And there's a lot of appetite for that. You mentioned keeping the frontline workers employed and keeping their knowledge about the

needs of the people that they're trying to meet, you in play. I think the recipe for this is how do you pivot these civil servants to higher value work?

value meaning like face-to-face work.

Face to face or someone was just talking about how, you know, in New York, Mamdani trying to, bring about big change and someone suggested, you know, you could actually have the subways drive themselves. But that doesn't need to mean you get rid of the conductors. It means that they could be walking up and down the train, which is much needed.

Yeah. my god, absolutely. Yes.

could be checking on people's health and well-being.

our appetite for better service is insatiable. It's bottomless. We're going to keep wanting better and better service. And currently our satisfaction with service is all time low. So it makes no sense that we're going to keep the service levels we have and just automate them. That makes no sense. It makes more sense that companies will improve their service and that's how they'll compete. And face-to-face is one of the best tools to do that. It's not automating face-to-face. I mean, of course things will get automated, but to your point, it's to enable more face-to-face, which is that front line. I guess government though has the most bureaucracy, in the middle, right? A lot of friction. And so a lot of opportunity to remove that friction. a lot of work between now and then, you know, this destination to attack that middle. We won't have solved it for a while, I think. There'll be plenty of people working on taking out friction before it's gone.

Absolutely, and certainly in government, that's true.

like we talk a lot about, in companies, this, this need for kind of a cultural change to take place before AI can really start to reveal its potential, frontline workers might be able to make a lot happen that, that wouldn't otherwise trickle down through government. are there other people somewhere along the chain that you think are uniquely positioned or qualified to help bring about cultural change within something as complex as the government and all its many agencies.

I mean, I've been sort of in and around government for the past 15 years. And you meet people who are quite good at culture change already. they tend to do it from a sort of grounding in the mission, right? Like, this is actually what matters. We are following processes and procedures here for good reasons, but those processes and procedures ought to be in the service of the mission. Are we fulfilling our mission? Are we serving the public? Are we making our nation more safe and secure? Are we preparing for the future? if they start from that place and sort of add in a real general respect for their fellow workers that generally they are trying to do a good job, but also some impatience, right? Like we can't wait forever to serve the public better, It was a pretty good recipe, I think, in the internet era for culture change and can be, you know, a good recipe for culture change in the AI era too. That impatience may take a notch up. as things move even faster.

Yeah, it feels like within organizations, there's competitions created by not wanting to get outpaced by a competitor where does the pressure come from in government? Is that pressure existing if people are just expecting higher level service from everyone because AI has delivered across a few industries where now that's just the expectation? What are the things that would drive the government,

Thank

to feel the pressure to move faster.

I think people have always felt it to a degree, the barriers to change are so much higher, right? So the math comes out a little differently. Because you can't compare yourself to a competitor. I think what's taking that place at the moment is the sense that. We have done such a poor job of meeting people's needs and at great cost, right? That a lot of the public wants to blow it all up. They no longer have patience for the system and they are throwing their, you know, their votes and their will behind, political actors and others who say, yeah, that whole thing, that's all broken, none of that works.

Let's do something really, really radical. I actually believe in something radical, but the systems matter and that what government does is incredibly important. I want to push for radical change while not taking for granted that if some of these things break, really bad things happen. But it does feel to me like.

While that is an ongoing criticism of government, that it just without the profit motive, without the competitors sort of nipping at your heels, you will move slowly. We now have just societal forces that are helping people want to move faster. the American public has to believe that government is worth it, right? That it will not be perfect, but that it is worth supporting in

evolving instead of like erasing it and starting over, That done chaotically is truly devastating for society

We talked a little bit last time. This has been rattling around my brain since then. This concept of friction as a tool, meaning politicians may, create a program and make the requirements for qualifying just difficult enough. and, and this is, you know, sometimes by accident, but sometimes by design and what happens in, in this world where anybody can apply for any program that they qualify for with a click of a button. how do we have to rethink programs? But like complexity seems like what happens next is now there's all these conditional elements that have to be in place to throttle it like truth versus just laziness.

I mean, you have to have the courage to say this is what we could afford in this, or this is what we think is effective towards our goals, and set the conditions that allocate scarce resources the way you want to allocate them. So if your programs, for instance, are designed to help the needy and you are rationing those resources by friction, you are, by definition, giving the resources to those who need it least.

And it's just, think it's really honestly just we have to be clear-eyed about that and say, so it's not popular to say we're lowering, you know, we're lowering benefit amounts or we're raising the, you know, the means tested limit so that fewer people

Yeah, or those who want it most.

will be eligible, and I'm not saying we should do that, but that is a better response than saying, here's a program that's, you know, that X number of people are eligible for because, you know, the income cutoff is here or whatever. But in order to actually get the benefit, you may qualify, but in order to actually get it, you have to jump through a lot of hoops. Most current example of that is the Medicare work requirements that came in the One Big Beautiful Bill Act. That means a lot of people that are eligible for that benefit won't get them because they don't, they can't get through the paper, I call it paperwork, though it's not all paperwork, the paperwork hurdles of showing that you've been looking for a job that every two weeks or every month or whatever it is, and documenting that such that the eligibility worker can say, yeah, you've checked all the boxes. So that means that people who have the most support the most time, the fewer demands on them. And sometimes, you know, even like help, right? Like we'll actually overcome those hurdles and those who have, many people they're taking care of in their home. You know, the the stress of poverty and all of those things. They're not gonna get the benefit. And.

Yeah. Like moms.

the folks who have the most chaotic lives, like all of the wrong things will select you out of that program. you know, unfortunately this rationing by friction is also wildly expensive. I believe in the state of Georgia, we could check this, but I have.

I've heard this stat, Luke Farrell published this the other day, he must have fact-checked it. But when they did work requirements in Georgia the last time, they gave more money to Deloitte for implementation of work requirements in the Medicare Medicaid system than they gave out in benefits that year.

my, yeah. That doesn't surprise me,

Ouch.

but wow.

As I said before, and I don't mean to pick particularly on Deloitte, but in this case, a statutory requirement like work requirements to prove that you are looking for work on a regular basis is welfare for Deloitte.

Right, for Deloitte workers.

takes money directly out of the purpose for which it was suited and puts it into the machinery of government that is burdensome, that is intrusive, that is ineffective and inefficient. And it's not good policy.

That's an interesting point. I don't even know how I would go around measuring the cost of friction, but that sounds like a super valuable thing to do.

Friction is. It's expensive in a lot of ways. We spend shocking amounts of money on the systems designed to administer benefits and other systems because of the legacy ways of building and buying technology in government that desperately need to change. They have needed to change and now they need to change even more with AI.

Yeah, it seems like some of those technology partners should bear some of the onus too, right? the government might be kind of low hanging fruit in that way, right? Like, we know they'll overpay for this or that this will be really hard to implement. And so this is a good place to set ourselves in the center of, right? Instead of. Yeah, there's there's not as much motivation for like, let's. Let's see how streamlined we can make all of this. Like, let's all band together and see how much friction we can remove when there, yeah, there seems to be, money lining up with friction and it's not going to the people who necessarily need it most.

There are some vendors that I think are very mission driven and trying to make the benefits system a lot better. And there are those who are just playing the game that the government has laid out. And it's hard to blame the companies in the sense that government sets the rules.

You can certainly, and I do, wish for them to act in more pro-social ways, but that's not a theory of change. The way to change it is to change the way that government procures and operates these systems.

I'm Right. Yeah. There's a company out there we partner with and, you know, most of the revenue they receive from the government is performance based. And so there's alignment. It's there's clear alignment in that case versus hourly services revenue. And to your point, who can change that?

not Deloitte, the government has to change how they take bids and who they give it to and to require this sort of approach. then of course the market will follow the consumer. Yeah, I'm still stuck on this. How could we measure the cost of friction in the government? Who would go about measuring that? would it cost more to measure it? Is there any organization within the government that's job it would be to measure that?

No, not friction exactly, but obviously we have built a government with a lot of different, what I'll call watchers, right? We have the, one of the problems in government is the balance between watchers and doers, but watchers are very important. So the government accountability office, the inspectors general and agencies, auditors, et cetera, could be charged with friction audits. But they tend more to look at fidelity to existing procedure and they very much value safeguards against,

sort of in the case of benefits, improper payments, for instance. But they tend to be very neutral, right? So what they'll say is, this is we're to do. Have they done that? If the one big, beautiful bill requires work requirements, will make sure

Hmm. you

that they are doing it exactly right. It's a sort of procedural obsession rather than what is the outcome? Are people getting the benefit? Are

Right. In a sense, it's almost shifting fewer benefits, even though it's the wrong ones, to more friction versus like cutting the friction and offering more benefits to the right people. It's, look, we've saved X amount of money by providing fewer benefits to fewer people by adding friction and more people and more oversight versus

Well, the oversight buddies don't think that way. They just think, did you follow procedure? They are not really often looking as their identity really is much more as neutral. So they're not saying, great, you paid fewer people out. They're saying, or you paid more people out. They're just saying, these are all the procedures.

You paid the wrong people out.

And that creates this incentive. And I talked about this in my book a lot at the beginning, when I was working on unemployment insurance during the pandemic. you had.

you know, these fraud prevention tactics that were in play, that people were sort of doubling down on, that really didn't prevent fraud at all, but did unfortunately keep legitimate claimants from getting their unemployment insurance benefits. And it was very difficult for the leadership to get behind changing those practices because the oversight bodies care about, you know, whether you're doing what was sort of on the

to be done rather than what works, right? This is a core dysfunction of government is the desire to be faithful to process at the expense even of the stated outcome.

Right. Right. Is it fair to say that, we've possibly swung the pendulum from, one end to the other where now it's in order to cut friction, they cut benefits. And it's kind of like, you know, chopping off your arm because your elbow hurts. It's this idea like, well, fewer benefits means fewer friction. So we're saving money. Whereas there's this now option with AI to say, no, we could actually keep the same benefits, possibly increase the benefits, save a lot of money on the friction side, still cut the overall costs. But it requires somebody to think in a little bit more of a complex way to solve the problem than to just say, the budget's too high, cut the benefits, we cut everything. It just seems very. crude and it seems like that's what's changing with AI is we can use a scalpel Like we can actually do surgery.

Well, first of all, as much as the costs of friction and costs of these systems administering these systems is really, really high, they're still small. They're still small relative to the benefits that Georgia examples of an outlier. I mean, especially if you include Medicare and Medicaid, which are just in a giant portion of our federal snap is much, much smaller piece of it.

True.

So I don't want to overstate that piece of it, but I do think, and I think that decisions about. you know, what we're spending money on in our government are largely made for a set of political reasons that aren't that relevant to this conversation. I just want to be, I don't want to overstate it. But if you look at something like the pandemic fraud that happened on unemployment insurance or the, paycheck protection program. The fraud in those really was significant. It was quite high. We saw it in the state of California. And I think that's an area where we could have done far, far better. was pre the explosion of chat GPT, obviously. And yet there were certainly AI tools available at that time to help states catch fraud.

than to be able to stop the payments, right? Sometimes the tools are there to see the fraud, but the authorities to actually not pay that out once it's determined that it might be fraudulent are often very weak. That seems to be what has happened in Minnesota with this big scandal. This big scandal there is that in fact, people still got to do a better job of catching the fraud. No, they did know that it was fraudulent. They had to continue paying it.

because it goes to the court system and the court says while we're adjudicating this you will continue to pay that out. I do think that these stories of fraud in the press have terrible implications and consequences for a social safety net that works. And we're really talking about social safety net here, not government more broadly. But we could do a far, far better job paying out legitimate claimants and have more political support, both in Congress and amongst the American people, for a functioning safety net.

we had talked about like just simplifying regulations, right? Can lead to better service delivery. But as we're discussing here, like the, these regulations sort of are many tendreled and running all over the place. our state governments maybe in a better position to move faster with some of this stuff than, than federal government, ostensibly there's like fewer moving pieces, but also could that like provide, we were talking also about like kind of this competition, this idea of being motivated to make these changes, right? Like if certain states really start moving the needle and finding ways to make this stuff work, that put pressure on other states? And then does that move up to the federal level?

Yes, I am very bullish on states doing good work that kind of shows. the dynamic that we want to see in play now. So largely I would call it subtraction rather than addition. I'm not saying you don't ever regulate anything new, but we have had this habit of just adding rules and regulations and never subtracting them, which is how you get these giant volumes. I noted that in New Jersey during the pandemic, they had a backlog like everybody else. And the commissioner there, Robisaro Angela, when he was called,

in front of the state legislature to get yelled at for the backlog. He brought boxes and labeled them 7,119 pages of active UI regulations just in his state and said, you want to blame the COBOL systems, you want to blame our frontline workers, you want to blame, you know, all these things that are really not the problem. The problem is you can't administer a system robustly and scalably. with that degree of policy and process, cruft, and accumulation. And UI is good example. It's got more of them in part because it dates from the 1935 Social Security Act. So it's 91 years old now.

people think, okay, a law was written and then the program goes, no, the law is written and then there's this sort of, you know, plaque essentially that builds up if it was like your teeth. you're never taking that off becomes, you it's very burdened. I think states are in a good position to do that. The caveat slash bright spot I would add to that though is that just, you UI is a perfect example of this. Very often, whatever outcome it is that you are trying to improve may look like a state function, but in fact is shared by states. federal, state, and local. And so the complexity of these things is just sort of made, of algorithmically.

harder because you've got federal state and local and you've got executive legislative and judicial right like all those together and and it can be very hard so you you want to think about these things as states taking the lead

But pushing up to the federal government and saying, hey, guys, need DOL, for instance, Department of Labor. If we're going to get to a 200-page description of the rules and regulations, which is absurd. It wouldn't be that short. If you're get to 200 pages that would really help you tackle this in a far better way, you are going to have to work with the Department of Labor. You can't just do this yourself, but you can tee those things up and say, what trade-offs are we making here when we say that this regulation still needs to be active or this? the safeguard still needs to be there.

I can't help but overthink it and go, I love to overcomplicate things. but with that in mind, can I create a system that would go through all of those laws or whatever they are, policies, assign a cost, a friction cost to each one of them because some of them are there, but they're not costing anything. So getting rid of them isn't going to do anything except cost money to get rid of them because no one's enforcing it. So it doesn't matter that they're there, but go through, assign a cost to each one and then say, this is the cost of each policy. Now let's start doing this on a cost basis. Like take the ones that cost the most to enforce and then attack it one at a time, reducing friction. from a cost standpoint, because a person's not gonna do that, but AI can do that. It can look across the organization, look at the cost, look at different people's roles, It couldn't be to the penny, but it has a shot at assigning some sort of liability, because each one of those laws are essentially a liability. Putting them on the books, what is the liability? And then saying, okay, your job is to cut. the liability, to do that you gotta start with the ones that cost the most and do the least. Almost like a P &L for every policy. And that sounds crazy because as humans we go, my God, who would do that? We're not gonna do that. But when that's like a button push, like maybe we just need to think outside the box.

till we are we are doing that. This is the. team at Recoding America, working with two states. I won't name them because they're not ready for us to talk about it, but probably add more states soon to do just that. It is right now sort of the baby steps of it. So you can find things like where in statute or regulation is a wet signature required where it doesn't need to be. all up for change. That'll be good. Like that'll have, you know, one less person who's like, God damn, I gotta go like fax this. I gotta go take this in person and you know, one less time where a transaction gets delayed because of that. That will help. But that is a very different ball of wax from say, you know, the interconnected regulations. And you see in something like unemployment insurance,

where it's, and where these things are contested, right? And there are real values at play here that we're gonna have to work out. I mean, I'm somebody who believes in safeguards for sure, but the right number and size of them. And at some point you have to say, Is each of these individually have value? Yes. Do they collectively though overburden a process that we are no longer getting the thing that we intended? And if so, then we have to make choices.

Right. you

The perception is that they'll all be like, know, wet signatures. Like no one will care and this is just stuff that go away. Some of it will be like that and some of it will be stuff where we have to make judgment calls and people and reasonable people will disagree. and to bring it back, I think AI can help us with those tradeoffs, right? Like we have but we have start thinking of them as tradeoffs instead of each one of these things. is obviously good because standing alone, that is the kind of thing we want, right? Like trade-off denial is one of the biggest things that holds us back from the kind of government that we want.

it kind of highlights in a way, like how flawed the thinking was behind, Doge putting forward this idea that you're going to streamline the government by getting rid of a bunch of people, which is really just the opposite of what it should be. Right. Like it's not about getting rid of people. It's about going through and doing what we're talking about. It's about mapping all these inefficiencies. and figuring out how to get rid of that and make the various people working for you more powerful and like better at their jobs in essence.

Yes, the irony is also that, what we're left with is that we've cut the workforce, but we haven't cut the work.

What you want to do is cut the work and then you can alter the workforce, right? But when you just randomly cut the workforce without changing, know, which procedural things are you, do you no longer now have to do? That question remains unanswered, right? You have not created a more muscular, robust, and effective government. You've just created chaos. I mean, one of my favorite examples of this is the Paperwork Reduction Act, which I always refer to as the comically misnamed Paperwork Reduction Act, because it creates enormous amounts of paperwork that have, I think, very low value. And it keeps agencies from collecting information from the public. keeps them from doing user research to test forms and applications to make sure that they work for people. It was a good idea in 1980 when it was written and it's a bad idea now. I think we calculated at some point that tens of millions of people hours that go into paperwork reduction act compliance for, you know, for no real good. Well, that's an easy thing to, I mean, it's not easy, but like you go to Congress. And you say, like, we don't need this, and people are still doing it. And let's focus people on the meaningful work, right? this is very low value work, and we need to focus them on even higher value work than they are today. I wish that the Trump administration had come in with that agenda. and then workforce cuts would have been much different The irony is that this also, from what I understand, is what happened in the 90s with the reinventing government agenda. Very different, did not happen with such suddenness.

and not with the kind of chaos, but John Kamensky, who was around for that and helped lead that effort, has this great line where he says that Congress ate dessert first, right? They did the same thing. They went around and cut the workforce, but didn't do the procedural work to remove that work that public servants have to do. And it's kind of ironic that we've made the same mistake, but this time, like sort of even boulder. that work remains to be done and it can be done in a thoughtful manner where we are really having the appropriate dialogue about these trade-offs. And I don't want to have, you know, years of dialogue about these trade-offs. I think we need to tee them up, have a dialogue and make a decision, make a change and move on. And that's also not what government is great at right now. But in some ways, the fact that Doge took the cut, you know, chainsaw to the workforce approach instead of the Vivek Ramaswamy chainsaw to regulations approach means that

regulatory rightsizing can be done in a more thoughtful manner should we suddenly develop the political will to do so. We are in the business of developing that political will, so I hope we succeed.

Yeah, I wonder if there's almost like maybe this role exists, but like a new type of lobbyist, right? His job is to go around talking about how much money they can save by filtering through regulations and removing all of this clog.

I mean, that's definitely happening, especially in red states. If you look at these, some number of states have declared themselves to have their own doges. And largely those have been the Vivek Ramaswami version of what doge was going to be, not the Elon Musk. Now, again, that can mean getting rid of constraints that hold back government from acting effectively. people may also hear that as removing constraints on what companies can do such that they could pollute the environment more quickly. This is a very complicated issue. And if you say deregulation, people will think of wildly different things, some of which I would be very much in support of and others not so much.

Good point. Yeah.

Yeah, it strikes me that, with this new technology as it sits, it puts a heavier burden on objective, like understanding and knowing what your objective is. Because the mathematics that are calculated behind it have to drive to some solution and that solution is your objective. it. feels like we can get lost. We can have things without objectives just sort of floating out there. And one of the things that I keep sort of thinking about is, you know, running the government like a business can make sense on the surface. But then you look at like what's a business's objective to make money and to charge more and to be more profitable and to deliver to the shareholders. So you take those rules and you apply it to government. if the government's job is to make money, then we wouldn't have any programs. how do we thread that needle between understanding that efficiency? What we're talking about when we say run it like a company, we say run it more efficiently, not run it from its objectives. as I'm, looking at attempts to make the government more efficient, they kind of miss the fact that you start mimicking how companies are run, you get lost in what the objective is.

You know, one thing that has taken me a long time to really understand is that when I look at a program, I think, OK, the obvious objective of this program is unemployment insurance, for instance. We're trying to help people who have a temporary, hopefully temporary gap in employment not fall into greater poverty so that they can get another job, right? the truth is that we don't really often negotiate the goal. We negotiate the program itself and the procedures that will run it. Because sometimes when there's disagreement, someone else might say, a key goal of the unemployment insurance program is not to give money to people who don't deserve it. And we don't actually try necessarily to. get those to use to mesh. What we say is, you want that, I want this. Here are program design rules that we can agree on. And that really puts public servants in a really tough position. Having said that, I think the leadership that I see that is most effective, both in terms of administering within our existing system and changing the system, is the leadership that comes to the table and rallies everybody around a meaningful outcome goal. everybody agrees on that in certain times, right? Like, hate to keep going back to unemployment insurance during the pandemic, but like, left, right, and center, in those first couple of months of COVID, everybody wanted the checks to go out the door. Everybody wanted the benefit to the people. That pendulum swung later, right? Back to what they call program integrity. Let's not give money to people who don't deserve it, because in fact, a lot of people who didn't deserve it and were scamming the government made lots and lots of money off of these things. I don't have a real answer for that. other than. real leadership within bureaucracies saying, yes, we are going to be held accountable to different outcomes because different people have voted yes for this policy or this law for different reasons. But at the end of the day, like, let's make this a goal and then inspire people to do what's necessary to achieve that goal, which is getting you more outcomes-oriented

behavior and less procedural fidelity behavior, which tends to have really perverse consequences.

it's so interesting to think how much detail is written down into the how and not the why. And where's the canonical truth of why any program exists? Like, why does the fax machine exist versus some rule that says it exists or doesn't say it exists? Determining whether they have them or don't have them is. shouldn't be based on the rule, should be based on the objective. is, does this help meet the objective or not? anyway, great conversation. I really enjoyed this. Yeah, thanks for having it with us.

Definitely. Yeah, thanks, Jennifer.

Thank you, it's been a pleasure.