Understanding and wielding power in local government, with Daniel Golliher
This week I'm joined by Daniel Golliher, founder of Maximum New York. We talk about his experience doing (and teaching) civic engagement, about the realities of policymaking (particularly at local and state levels), on how policy engagement is a skillset that one can choose to simply be effective at, and about a bit about inculcating agency.
Sponsor:
Vanta automates security compliance and builds trust, helping companies streamline ISO, SOC 2, and AI framework certifications. Learn more at https://vanta.com/complex
Timestamps
(00:00) Intro
(00:33) The reality of political science education
(02:45) The apprenticeship model in government
(04:24) Challenges in government training
(07:06) The role of legislative aides
(10:33) Local government and housing policies
(14:24) Effective political engagement
(16:15) The power of communication in policy
(20:26) Sponsor: Vanta
(21:44) Witnessing government in action
(31:29) Learning government and politics
(32:21) LLMs and policy
(35:10) Engaging with local politics
(37:58) Influencing policy
(43:25) Running for office
(47:10) Blue tape
(58:39) Wrap
Transcript
Patrick McKenzie: Hideho everybody. My name is Patrick McKenzie, better known as Patio11 on the Internet. I'm here with my buddy Daniel Golliher, who is a political entrepreneur who founded a school which teaches people how to do political engagement on a city and state-based level. Daniel, thanks very much for your time.
Daniel Golliher: Thank you very much for having me.
The reality of political science education
Patrick McKenzie: A thing that I have encountered unexpectedly a few times in my adult life is being thrust into a position where I have to concern myself with how power actually works in the world. You've mentioned in writing previously that the sort of pat answer as to how the American government works that we teach ourselves in high school in civics courses, and that is taught at the university level in political science courses, does not actually match the complex systems that we find ourselves encountering. Can you spin out that thought a little bit more?
Daniel Golliher: Yes. And for anyone who's listening, if you haven't read Patrick's very thorough overview of his work with VaccinateCA in Works in Progress, it's a very good overview of encountering how the processes of power and government work.
But my general thing: I founded a civic school called Maximum New York, and the idea is our existing systems for training people on how the government works either don't exist or they get you further away from the truth than closer to it. So for example, if you get a political science degree, almost everyone who graduates with that degree cannot tell you what a statute is. They can’t define it with confidence. That's terrifying.
[Patrick notes: Yep! And I’d say borderline unbelievable, but then again, we administer a game playable by 6 year olds to professional software engineers to cheaply test whether they are minimally capable of following instructions.]
And so the existing channels just don't work.
There are of course some exceptions, but the rule is the rule, and there are a lot of reasons why this is the case. An essay I wrote, Political Science Degrees Must End", went decently viral. It argued not against the existence of the degree, but against the paradigm the degree currently represents.
Of course you will have some people pop up and say, "Well, that's not what college is for. It's for learning how to socialize and et cetera." And I think, fine, sure, but what I said is still true. These students are not learning how the government works, and that includes whatever level of university you want to talk about. I have a government degree from Harvard and can speak from personal experience at that department, although it may inculcate several things in its students, high technical rigor and understanding of the government generally is not one of them.
[Patrick notes: It has been observed in many places, such as Recoding America, that technical rigor is actively selected against in government, both in politics and civil service, because it is considered a low-status implementation detail rather than high-status policy work. After you’ve done the important policy work, you pass it off to staffers who will pass it off to a bureaucracy which will pass it off to a contractor. We continue to labor under the misapprehension that this has a chance of producing positive outcomes.]
The apprenticeship model in government
Patrick McKenzie: So you've identified these ways in which political science fails to inculcate an understanding in how government actually works. My general impression is that there is something of an apprenticeship model in some institutions where you build up this understanding over the course of years and they're deployed against actual problems, which I suppose is running into in government the same problem we see in government and civil society, which is one distinction that often gets elided. But we are seeing the same crisis of the institutions that we're seeing everywhere, where the old apprenticeship based model is becoming less functional because the people who are supposed to be teaching the apprentices no longer do so. And in some cases, the institutions no longer have sufficient slack to bring someone up for decades.
And then interesting decisions get made by people in their twenties. [Patrick notes: People in their twenties are capable of producing excellent work! But probably not without a system around them that supports great work.]
Daniel Golliher: Yeah. This is something that I point out to people because I do walk around saying sentences like "no one knows how it works." People are not trained in government, but clearly the system continues to spin in some fashion. There are people who know quite a lot. Well, where did they learn it? And it is usually just a very large amount, a lifetime's worth of on the job experience. There's a couple of problems with that model though. Number one, the intake valve is not very large, so the apprenticeship model is much more bespoke and it just can't handle a larger amount of people. And if it breaks down, there's not really a backup.
Patrick McKenzie: I think there's also multiple knock-on consequences to this model. One that would excite many people, for example, is that to the extent that there is no legible process for getting into these things, the distributional consequences of who gets into the doors of the corridors of power, and then is able to exercise power, are perhaps not what we would like them to be if there were a more designed pathway in.
Challenges in government training
Daniel Golliher: Yes, and that is in general what I'm working on doing. So the general idea is if you remove the autodidactic blob of friction between all the kind, smart, ambitious people that you know and the government so that they don't have to teach themselves or they don't have to go find one of these very illegible apprenticeships, which there are not that many of, then more and more of these people will go into the political system.
Because for the most part, people feel like they're on the outside looking in, especially if they're coming from great success in private industry where maybe they're able to—that system's legible to them. They may be able to work faster and then all of a sudden they're presented with government. This whole other world, they feel like they have no way in. And so typically they just hit their head against the wall a few times, give up, and write off government. And when this happens at a large enough scale, if the government was breaking, it continues to break because it doesn't have a way to take in new talent.
So I would say my mission is don't break the apprenticeship model that we have, those pipelines still exist for the most part. They're going to continue to exist. That is part of the overall process of learning how government works is on the job experience. So I don't think you should get rid of it. I do think though, you could have a broader base of training that brings in far more people. And so when they do get on the job, they're far more prepared and can go further faster.
A story that I often tell, because of my blog and because of the materials on it and my Twitter presence, I often get a lot of calls from people who have graduated college with a poli-sci degree, and then maybe they land on Capitol Hill or somewhere and I get an email from them asking just to talk.
One particular call I had that I've had many versions of is a call from someone and they're maybe—they've gotten out of work. It's like 8:00 PM. They're—you can hear the pacing in their tone, and they are explaining to me that they feel overwhelmed on the job. They don't know how anything works. And they feel betrayed by their university and by their degree program because they think "I have a government or a poli-sci degree from this prestigious university. And yet I don't know anything. I don't feel like I recognize anything that I've walked into. What was I doing for all four years?" And this is not how universities should treat their young people. This is not how alumni of a government program should feel.
[Patrick notes: I have to say: this doesn’t sound like an unusual experience of one’s twenties in the tech industry? Of course you don’t know how anything works; you just got here. The way you skill up is to pick things that are impossible and do them anyway, with occasional feedback from more senior engineers. But perhaps government studies selects for people who want to have a kinder, gentler experience, with clear expectations and a glide path to success in their role. And indeed that is feedback the tech industry has adjusted to over the years, more in some places and less in others.]
And of course, if you're in a high pressure job like that, if you're a legislative aide or staffer, those are high pressure, intense jobs. You should probably, especially if you're new, feel some form of overwhelm and that's normal. That's the kind of job you're signing up for. But instead of feeling like you know zero, you should feel like you understand the contours of the system you have stepped into. You shouldn't feel like you're day one at the gym where everything hurts and feels broken, you should step onto the job feeling like you're, I don't know, day 20 or 30. Yeah, your muscles are gonna be sore, but you know what you're doing. It feels relatively smooth.
The role of legislative aides
Patrick McKenzie: For the benefit of the people who haven't worked in government before, legislative aides are a phenomenon that I find endlessly fascinating and important to understand about the government, and I mean the following descriptively. And this is in no way a criticism of the people who have kept the nation running for several hundred years.
Most legislation is written by 20-somethings who have no particular expertise in the field that they are legislating. That's just a neutral statement of the world in the American political system. And we expect them to become experts in everything from heart surgery to how airplanes work to the proper care and feeding of worldwide financial markets. And to become experts in the compressed timeframes at which legislation typically happens where there is a very long tail of thinking about something then a burst of energy that happens on Capitol Hill or similar, where something goes from "well, maybe there's a deal possible here" to legislation actually being voted on in a matter of, let's say, weeks.
[Patrick notes: I don’t have a citation ready off the top of my head for e.g. Obamacare’s legislative history, but I seem to recall that being both a rehydration of draft proposals that had been circulating for decades plus an extended-all-nighter for a group of young policy wonks.]
And in that period of six weeks, you have to become going from literally "what is a stablecoin" to "okay, we are going to put a hundred requirements into the current stablecoin legislation. And those will be the requirements. We don't get a do over. We don't get to say it was actually should have been 102 or these five, no one actually cares about them. We can ignore them later." [Patrick notes: Example chosen because of the recent episode with Hasseeb Qureshi and the ongoing march towards stablecoin legislation.]
And then you're stuck with those decisions, both the people who wrote them and the rest of us who are bound by duly passed laws, for the next couple years, which increases the level of urgency on making sure that for those that the American political system trusts with this level of power and authority, we should equip them adequately for their first day on the job. And indeed their 300th day on the job.
My sort of cached impression is that legislative aide serves in that capacity for a few years and then goes on to do other things. And thus there are very few 40-year-old legislative aides. But is that impression naturally accurate?
Daniel Golliher: I think it depends on which level of government you're talking about. And this is one of my general disclaimers when making statements about "the government." There is the federal one and there are 50 states and there are many localities. And although you can find commonalities between them, a lot of statements about the government are highly dependent on the context of that government.
So certainly at the federal level, people age out of these positions. And if you look at the state or the local level, I think it changes a little bit. For example, if you're looking at the New York City Council, they do have staff who are attached to each city council member who assist with the bills. They also have, just like Congress does, an internal legislative division staff—lawyers who help write these things.
So I think the New York City Council, the average age of the person who might be writing this legislation is actually a little bit higher than it would be in Congress. And I think that has something to do just with the velocity of people coming in and out and running through the pipelines. The pipelines are smaller. There's less competition for the New York City Council, that sort of thing.
Patrick McKenzie: I think we observe this fractally in state level politics and city level politics, or at least have been told this by people who have attempted to do political entrepreneurship at various levels and then realize the vast differences in the culture locally versus the culture in say, Washington, DC or for national level politics, which are not necessarily conducted in Washington DC specifically.
But different talent pool, different sort of context.
Local government and housing policies
Daniel Golliher: And this is also why I focus sub-federally. So just like San Francisco pulls the lion's share of a certain kind of tech talent, and that is where the prestige gradient also ends. Similar with Washington DC and talent that would like to flow into politics and law. But everywhere needs tech and everywhere needs competent civil servants and competent elected officials.
So generally the focus of my work is to show people there's quite a lot you can do in New York City and state, and in fact, we need very talented people to go into this domain because you would not believe the things that are subject to legislation at the city and state level that you really can't do much about if you choose to climb the ladder in Washington DC.
Patrick McKenzie: I would probably believe it, but for the benefit of the audience, what are some things that they might not appreciate are local concerns?
Daniel Golliher: I think the easiest answer, because it's getting headlines the most right now, is housing. Where do you build housing? How much housing do you build? And again, this varies from state to state, and it varies between the city, the system that any individual locality has with the state.
But in New York, for example, the state legislature has immense power to dictate what can be built even in New York City. For example, up until about a year ago, there was a thing called an FAR—Floor Area Ratio cap on New York City residential buildings. So it says they have to be small, relatively small. And that was imposed on New York City by the state capitol. It was not imposed on any other city.
So you can think for yourself, how much sense does it make for a state capital to impose such a strict growth control on the largest city in the nation that is smaller than the buildings that were built historically. So if you think New York City should grow and should have more housing, in this respect and many others, it's purely an act of the state legislature. It's purely within their power to change. And it must be changed there. The city can't do anything about that.
Other things though are up to the city and the city can do quite a lot. Would you like to build a 1000 unit apartment building where there is currently a parking lot? That is a decision that the city can help you make. Or it can also keep that parking lot because it can block the person who would like to build it.
These kinds of decisions are immensely consequential because just imagine, if New York City were allowed to grow to a city of 12 million, say from its current eight, eight and a half ish million, the agglomeration effects would shift massively. The political economy of the nation would shift massively. All of that is in the hands of local elected officials nobody has ever heard of. So it's actually very consequential which people choose to go into city and state government law. So housing's the clear, easy one, but there are so many others.
Patrick McKenzie: I've previously had my father Jim McKenzie on this podcast. Dad was in commercial real estate for his entire career. And housing, commercial real estate generally, and other things are not merely heavily influenced by local elected officials, but hyper-local down to block-by-block activism and similar. It also legendarily comes down to the one cranky neighbor with enough time on their hands to make it out to the meetings. [Patrick notes: Professional real estate developers deal with this reality in ways which sometimes sound corrupt, sometimes sound abusive, and sometimes sound clever-with-a-hint-of-naughty.]
And just getting across to people that you can choose to be the cranky neighbor who makes it out to the meetings is net beneficial for at least some folks. At least while we have the system configured in the way which promiscuously distributes vetoes to whomever shows up to the meetings.
But there's also sort of a metis layer for how one chooses to interact with the government, both in one's capacity as say a constituent showing up to a meeting and attempting to be persuasive to individuals that are at that meeting, inclusive of other constituents, and then the elected officials or appointed officials. There are better and worse ways to go about engagement. And then there are better and worse ways to attempt to run for office or similar.
Effective political engagement
Do you teach people across the entire stack? Do you focus your efforts on particular areas?
Daniel Golliher: My marquee class is called the Foundations of New York City, and it's an overview of city government and law and political history. We do discuss the state and the federal government too, because they matter immensely for what happens in New York City.
But the goal of the class is when you finish it you can do things like draw a diagram of the entire city government with its dependencies at the state and federal level. You understand what the pieces are and you understand the principle processes that knit them together. So that is how does a bill become a law? How does a rule become a regulation?
But also we have a thing called ULURP, Uniform Land Use Review Procedure. Very scary sounding. But that is if you would like to do a discretionary land action of some kind, you would like to build an apartment that was previously a parking lot. You have to go through ULURP. So people know all of that and what I tell them, assuming they passed the final exam, is "now you're in the real world. Now you're doing politics for real, using what you've learned. You have to pick a channel."
And through class we'll have discussed case studies of what does it look like successfully to write a blog that changes policy, the posting to policy pipeline so-called. And so we will have looked at examples of someone wrote this Substack post or Tweeted X, Y, or Z in the city and then it made its way into a law or it changed an official's decision on something. So for some people, that's the best fit.
Running for office is also something that I encourage people to do and I do help them if they do it. And for the most part they're successful. You can also try to get appointed to certain positions, which is a different kind of race that you have to run if you would like to be appointed. But fully a third of my alumni of over 200 have some sort of local position, elected, appointed, et cetera. So if you are skilled, it's relatively straightforward to get these.
The power of communication in policy
Patrick McKenzie: The posting to policy pipeline is endlessly fascinating for me. I have sometimes terrifyingly wandered backwards into doing some of it myself, and also seeing other people with very varying levels of understanding achieve some level of influence through that. [Patrick notes: We are currently living in the Essay Meta.] What distinguishes an essay on Substack or similar that successfully changes the law locally, versus simply kvetching on the Internet about how everything's broken?
Daniel Golliher: I think the first thing someone has to do is see if they can model who their audience is. So, are you a lawyer in the mayor's office? Are you a city council member? Whoever you are, you have a particular set of constraints imposed on you by the system within which you operate.
And so a very good Substack post will somehow communicate that the author understands the constraints that the actor inside the system is operating within and will address those constraints and point a path forward. So I think that's the first and most important characteristic.
The second one is that it has to be technically specific enough to be useful. So anyone can say the sentence, "the government should do X, the government should do Y." Okay. That doesn't tell me which line in the administrative code has to be changed to do that and often—to a degree people wouldn't expect—lawmakers themselves don't know which lines in the law have to be changed.
So if you can point that out and demonstrate that you appreciate the constraints they're operating within, give them some draft language for the legislation and explain why this is a good idea, that's the general nature of a good post. It appreciates the constraints of the reader and it's technically specific enough to be useful and actually get the job done.
Patrick McKenzie: I also find that there are certain, let's say marketing micro tactics that make things more likely to be successful. Just in terms of optimizing for the form factor, naming things, both in the sense of titles on posts but but also naming general concepts is anomalously highly leveraged.
I think some of the best known examples of the success are where the title of the essay has forever become how people will refer to that problem in the future.
And go figure. If you are totally in control of the language and about the language that people discuss your pet issue in the future, you'll tend to get all that you want out of discussions that are conducted in the arena that you have constructed. [Patrick notes: Some people read Orwell as a warning, others as an instruction manual, film at 11.]
But appreciating how the target audience consumes things and how information, and thus power, flows around the organization local to them is very useful.
The thing that has recently become popular is people have gotten down to—certain varieties of government officials work much more with respect to PDFs than they do with respect to "random Substack posts." And so if you have something which is morally speaking a blog post, but you call it an essay and you make a downloadable PDF version available, and you just do the tiniest amount of print design on that downloadable artifact, that makes it much more likely that it will be formally citable in the various epiphenomena of the government, than if you had left it in the classic blog post stylings.
[Patrick notes: Situational Awareness is a good example of this, and just in the AI policy space, AI 2027 and several other initiatives ran with the theme.]
Daniel Golliher: Yes. And I give this advice to my students and there's so many of these little things, like, for example, Maximum New York is a small-b blog. So the posts will range from three to 400 word things that I dash off in a half hour to 6,000 word policy overviews that are tightly edited and deeply researched, but when I do have something like that, you'll see at the top link to download PDF.
And so there's a lot of these little things that are obvious in hindsight, they make sense when you explain them, but I don't know if people would predict them beforehand from first principles in any way that makes sense.
Patrick McKenzie: Another one is caring anomalously much about what the experience of a person interacting with your artifact on paper is. Like, make sure control-P does something sensible. There are a lot of decision makers of a particular generation. There is also a common information processing style at some offices where the comments will be circulated on a hard copy, period. If your thing is unreadable on a hard copy, well then comments will be made on something that is not unreadable as a hard copy.
Witnessing government in action
Daniel Golliher: Well, and part of this is all, what we're discussing now for me is just people need to appreciate how the system actually works and they need to understand how it actually works. And the amount of daylight between that and what people are taught formally, if they're taught anything at all, is just—there's quite a lot of daylight. It's quite bright.
So for my classes, in addition to a lot of reading or guest speakers, people actually have to attend, I call it "witnessing government homework." You have to go to a city council hearing, you have to go to an administrative rule hearing, you have to go with me to night court, or we watch arraignments.
And usually when you have people actually sit in the room and they observe—people sort of underestimate how much value they can get out of those because, for the most part, if they have ever gone in the past, they sit there and they're bored. And usually what I tell them is, you're not looking at all the little things that are happening here. And usually they don't have someone with them who can point them out and explain.
But within the context of the class, it takes maybe one city council hearing for someone's model of how the entire government works to entirely flip as long as you have someone there to point out what's actually happening. There's so many little micro interactions.
Patrick McKenzie: You presumably wanna go there with sort of the eye of an anthropologist or ethnographer, who is attempting to figure out, okay. Beyond the formal description of, you know, here are the rules for votes. Here's the org chart, et cetera. Let's start drawing up teams, lines of influence, and where does this person get their information that they're citing? How would you get upstream of that information?
That by the way, is one of the most under-exploited things for Internet natives. Everyone has an information diet, if they make decisions that matter. And sometimes one level upstream of the information diet is "too difficult to influence." [Patrick notes: I think this is a self-limiting belief, even though it has frequently been one I’ve held, even after e.g. the New York Times asked me to write about national scale disasters because someone there apparently reads HN and my post on a topic made a good impression.]
Two levels upstream of the information diet is the same blogs everyone reads.
I think there's a real way that the New York Times editorial board is effectively part of the United States government. They can influence a policy agenda by what they now choose to talk about or not talk about. To the extent that that is a worthwhile model of the world, then Slate Star Codex is also somewhere on the org chart that is the United States. [Patrick notes: There’s some bad blood there, and I think part of the reason for it is the NYT saw somebody new on the org chart and was surprised.]
Daniel Golliher: Statecraft by Santi Ruiz is increasingly becoming something like this, I think for people in the federal government.
Patrick McKenzie: Can you hum a few bars for someone who might not have read Statecraft on what's the influence of that happening?
Daniel Golliher: Yes. So the general idea is it's series of interviews that are done by Santi Ruiz at the Institute for Progress and is interviewing the people who were there and did the thing and they're now telling you exactly what happened on the ground with enough detail so that maybe you can take that information and operationalize it wherever you are.
So I think the very first one was about, relevantly to the current situation, PEPFAR, like how exactly was that program structured? Why does it work? What were the issues with it? Another one is about how DARPA innovates and came to be innovative. Because you know, the sentence, "DARPA is innovative" is very abstract. You don't actually know what was going on there, but who were the people and what did they do? What risks did they take? The interviews there I think are great. And I see them cited more and more. So I think that's an example of a very good ascendant newsletter.
Patrick McKenzie: Internet native distribution mechanisms are shaving decades off the kind of learning feedback cycle for these sort of things. There's always been sort of media epiphenomena about government work, but the deep research, not just the writeup in the newspaper the next day about what was said at the council meetings, but the "here are the players, here's the scorecard" often takes decades to enter into the public eye.
Caro's very well cited works [Patrick notes: most relevantly to Daniel, his biography of Robert Moses, who had about as much impact on the built environment of American cities as any other individual], which are often described by people as "this is the best encapsulation of how power works in the American political system," took decades after the events of them to distill and recount those events and then informed many people who then—it will take someone a couple years from the point where they read Caro to the point where they are making decisions based on that, et cetera, et cetera. We're looking at a multi-generational cycle time.
And if things are going—not to toot my own horn too much, but VaccinateCA was 2021. We were very public about what we did there. And people were copy and pasting some of the lessons into the Discord (literally) for the Save PEPFAR effort earlier this year in 2025. And so going from a 40 year cycle time to a four year cycle time is an improvement, and there's no reason in principle why it can't be a six week or six month cycle time.
Daniel Golliher: And if you are someone who's listening, who's wondering, "How do I get upstream of some of the information that is in, I don't know, a city council or a state legislature?"—also, this is also applicable to Congress or something. Too often if people feel like there's too much, they feel overwhelmed and they don't know how—these legislators know everything. They seem to know everything. They're in a hearing, they will be citing a ton of laws and referencing a legislative history that goes back decades. And people think, "how can I possibly get up to speed with that?"
And the short answer is, you don't. Because the legislators aren't up to speed on it either. What they're doing is reading from a briefing report that was prepared for them by staff, and usually you can have access to that briefing report. Sometimes it's literally publicly available. Sometimes you maybe have to take another extra step or two.
But for example, with the New York City Council, these things are called committee reports and they are downloadable from the city council's website and you can get them before you walk into the committee hearing and sound just as prepared as anybody.
Patrick McKenzie: And this is another opportunity for engagement. If you know what a committee report is, it is probably not the case that you are orders of magnitude less competent than the person that drafted the committee report. So you are almost as capable of sketching out the table of contents as they are, and then figuring out what are the sets of artifacts, statistics, citable authorities, et cetera, that this report is going to require.
And then, simply getting on top of Google for the heading of any of those things that will appear in the table of contents is just absolutely enormously valuable. If you predigest doing the work so that report authors have to do less while they are under time pressure, et cetera, et cetera, et cetera, then… who decides in the world what set of authorities is the authoritative one? It's the first person that writes the paragraph that cites them where that paragraph looks facially plausible to the right kind of person in the right kind of place. And you can choose to speak paragraphs into the world, then they can speak their own truths after that.
[Patrick notes: A related, and somewhat terrifying, realization is that LLMs can spit out facially plausible paragraphs about almost any topic. We do not yet have a society-wide immune response to this.]
Daniel Golliher: And if someone thinks that they could not match the quality or the thoroughness of a committee report, I think the easiest way to disabuse yourself of that notion is to read a committee report. Because, and again, this is true across government, but certainly of the New York City Council, they contain quite a lot of basic errors as well, up to just misciting the law.
Like, that's not the right sentence, you know, when I click on the hyperlink or I go look that up in the statute—this is not the right law. So I have no idea what you're basing this comment on. And there's frequently quite a lot of stuff like that. So if you write well and you simply cite yourself well, it's pretty easy to get a reputation that's useful.
Patrick McKenzie: And if you do the effort of trying to look at a few of these reports and calibrate yourself better by finding errors. in many cases, the level of error will not be someone failed in the middle of their deep dive into 70 years of legislative history. It can be in literally middle school math. “You inverted the numerator and denominator.”
But if you find errors like that, you can frequently find the authors of the report will have their email addresses publicly available, often published prominently in the report itself. And say, "Dear colleague, I was enthusiastically reviewing this report. And if I can point you to page 64, I just want to let you know that there seems to be a bit of a math error. I might have come to this conclusion instead. And here are some supporting citations."
You are reasonably unlikely to get that report corrected, although stranger things have been known to happen in the wild American experiment. But you are quite likely to end up on their informal list of people to check with in future iterations of the game.
Daniel Golliher: Yes. And you might also be asked to come and testify should that issue come up again. So you do get on a list of people who are well regarded and know about the topic, which can be useful, and does wind up shifting the text of legislation.
Patrick McKenzie: A thing that repeatedly blows my mind is that I worked in communications for a while. [Patrick notes: Comments made are, of course, not necessarily endorsed by any particular Communications department.]
If you are ever in the newspaper about a particular topic, regardless of how you get there, but if you show up once and don't sound like a serial killer while you do it, you go on a list of people who are authoritative resources with regards to talking about X ever again in life.
One highlight of the VaccinateCA experience, which I would commend to anyone, is that if you care just a little bit more about your communication strategy than most people in policy circles do. Just to get someone to see, "Oh, these are good do-gooders who likely know what they're talking about” magnifies the impact of everything that you do later in a legislative or other government process. [Patrick notes: I realize that “politicians and activists care too little about PR” is not a widely held point of view but my POV is a bit more nuanced. They overvalue press hits and undervalue having any sort of strategy which would be advanced by particular words in particular places.]
Daniel Golliher: Yeah. I've had conversations like this with a lot of students or with people who are interested in taking my class. And usually if we've gotten this deep into the weeds or we get to the point where we're talking about committee reports, usually I tell them to take a deep breath because I think that they are a little lost in the sauce.
Learning government and politics
So if anyone is listening to this and you think to yourself, I already feel like this is a vast, complex system that I don't feel like I can apprehend—I would like to remind people that this is very learnable. And it is in fact, what I spend most of my time doing is teaching people how it works.
And you should think about it like you think of any STEM field, for example, if you're thinking about calculus or physics or something, you see that there are very advanced practitioners of that field. A certain kind of person would not say, "Oh, I could never learn that. That's just magic beyond me. I have no way in there." You would think there's tons of YouTube videos, there are tutors. There's a lot of ways to learn it. So I would say adopt that attitude with government and politics if you're listening to this and you start to feel like you're on the outside looking in with no way in.
LLMs and Policy
Patrick McKenzie: I think the existence of LLMs is all but magical here. Less because of the ability to quickly come up with the deep knowledge of people that have been in the field for 45 years, although they can do that too some of the time. But more because of things like there are shibboleths and how one engages with the government. There are ways to write committee reports. There are ways to write "white papers" which will get one prioritized towards the top of the pile or towards the bottom of the pile when, for example, federal staff read public comments on pending regulations or legislation.
And if you present as playing the game better, you tend to get treated better by other participants in the game. And the LLMs are just scary good at style transfer. So if you are capable of writing up your desires in any format whatsoever, and then saying like, "Can you just government this up for me?" Again, scary good at that.
Daniel Golliher: I don't know if anyone has—I mean, we'll see when this comes out based on—we're recording shortly after GPT-4 has dropped and ChatGPT was already pretty good at even writing model legislation. GPT-4, as far as I can tell, is much better. And it will get you—it still makes small errors sometimes, but it gets you 95% of the way there.
In one of my more recent classes, I just had my—I was projecting my computer up onto the wall and asking some LLMs to write model legislation and it can do it. The one caveat I will note is I did ask it to write a specific statute that caps the amount of buildings that can be designated as historically significant in New York City, just for the sake of the class. And it cited my blog quite a lot and pulled from it.
So the other thing I'll note here, which I think people are talking about with regard to AI tools is it's pulling from the indexed corpus of what has been written. And so another weird benefit of writing about especially state and local law is there is not as much competition for people who have written about it. And so you might start showing up in LLMs sooner than you think.
Patrick McKenzie: I think this is underappreciated by almost all writers. You should be doing something very differently with your life if you assume that as opposed to a generation earlier or even five years ago, most of the direct effects of writing will be by people who actually read what you wrote. And you have the opportunity, a near certainty that most "people" who read what you write in the future are not going to be humans. But humans will interact with what you write with an indirection layer in the middle.
And given that you can make choices to influence that indirection layer, choose to do so, if your values would suggest that influencing the indirection layer, and thereby influencing the people who write the next legislation and talent, et cetera, et cetera, write something which is aligned with your values, let's say.
Engaging with local politics
You mentioned that there is just less competition. There are quite a lot of people who are interested in local housing policy, but the number who go from simple interest or discussing it with the other parents at school to actually taking action on it is far lower. And so when you get to like hyperlocal politics, there might be 10 people that matter. And in any group of 10 people there are one or two good writers maximum, basically. And so you can be that person for basically the price of showing up.
I think I have a good buddy Alicia, who decided she wanted more courtyard housing to be built in Chicago and has been doing some combination of policy advocacy slash just creative use of LLMs to get herself into that. And she has gotten an op-ed in the Chicago Tribune and an op-ed in Crain's Chicago Business, which is a locally influential publication.
This is the former comms professional in me sticking a hand up. Every producer of, for lack of a better word, content, needs to feed the monster every day.
[Patrick notes: I hate the word “content” because it autocommoditizes the value created by professionals, but nobody has yet come up with a way to say “writing” in a way that encompasses podcasts and video. Shakespeare and Taylor Swift aren’t remembered for their content and you won’t be, either.]
Given that there are relatively few people in the world that will crank out 800 word newspaper editorials and make them interesting on any given day, and every editor needs to find them for every day that ends in Y to do a daily news publication, just sticking your hand up and saying, "I can talk intelligently about a problem of relevance to your readers. And I can listen to editorial feedback and I'll make this easy for you," sifts you to the top of the pile.
And then after you've been cited by the Chicago Tribune as knowing what you're talking about with regards to local housing politics, there is no one in the city of Chicago who will say that person actually doesn't understand local housing.
Daniel Golliher: I think part of this is a larger way of recognizing the system, especially sub-federally, is just much smaller than you imagine, because for most people, they go off the emotion of how it feels to try to enter the system. And it feels very big, like a giant black box. But it's usually a very small number of people who you can talk to or DM on Twitter. And you can DM state legislators and they often DM you back. Most people just don't try.
And this in many ways, it's just much easier—it's much easier than people think. They simply do not try even a little bit. And I think a good example of this is the race for the New York City Council speaker, sort of like the second most powerful office in the city behind the mayor. There was a recent headline from the publication City and State, and I thought it was a great headline. It said something to the effect of, "How to win the race and influence the 125 people you need to become the speaker of the city council."
And that's about right. It's about 125 people scattered around the five boroughs to become the second most powerful person in the city. But everything in politics tends to be shaped like that. And usually the numbers much, much, much lower.
Influencing policy
Patrick McKenzie: Yep. My brother used to work in constituent services for a U.S. Member of Congress in Illinois. Most of the time for constituent services, you're actually helping people get through the various processes in the government.
[Patrick notes: Pro-tip if you’re ever stuck with any federal agency: call your Congressman. No, you don’t have to know them or be a donor. This is how we ration access to smart polymaths in retail customer service in the U.S. federal government. Your agency of choice very likely has a dedicated phone number for Constituent Services to call, and has an inaccessible-to-the-public staff of institutionally trusted troubleshooters which pick up that phone very reliably by the standards of the U.S. government.]
But if someone calls up and says, "I would like to let my representative know how I feel about tariffs," someone is going to write down that piece of information and put a tally mark on a sheet and the number of people who call up and sound like they know what they're talking about, that would rise to the attention of a federal legislator is literally single digit.
And people just do not appreciate that the bar is that low. And I've only had call to speak to the staff of a representative once in my life. [Patrick notes: … On reflection, not strictly accurate, but I didn’t perceive myself as initiating the other one.] And it took three minutes and 28 seconds. And I got a letter back from the representative, which of course, okay, the production function for this is, there's a 20-something at their office who drew the short straw and had to write the letters back that week.
But they hit, they robo-sign it Member of Congress. And that is signed like that for a reason. And it was very clearly like there were five incoming calls in the hopper that needed to be specifically referenced in this week's letter. And I was one of the five for three minutes and 28 seconds of time, which was not exactly scripted like the invasion of Normandy. It was just a tax issue relevant to the software industry. [Patrick notes: Shoutout to Michele Hansen, who did excellent work in making many in the small software community aware of the necessity to capitalize salaries of engineers, which was a mostly-unintended consequence of a compromise on a tax bill a few years ago, and which has ruinous consequences for small software companies.] And I spoke as a software business owner who sounded intelligent for three minutes and 28 seconds, and boom, changed what a member of Congress believed about it.
Daniel Golliher: And to me, I think this is a great example of the system—if you choose to interact with it, and you're reasonably circumspect, you're just not yelling at them—the system really does listen to you quite a lot of the time, much more, far more than people would expect.
I call members of Congress quite a lot, even the ones who are not my representatives. For example, it's often the New York delegation, but often a member of Congress will write an open letter to say the governor or someone asking for a policy change at the state level that's consequential for the federal government for some reason.
And for example, that letter might not be very well written, might miscite some law, et cetera. And there's one example that I've written about on my blog just as an example of how I do these things. And I agreed with the federal member of Congress on the issue, but I did not like that the letter so poorly represented that point of view and got some things factually incorrect.
And so I just called their office. The phone was picked up by a young 20-something. They were very kind. I explained the issue and then it wound up being corrected. So normally members of Congress want to hear from their constituents and if you want to call someone who is not your senator and ask them to vote a certain way, I don't think they're gonna listen to you as much as you would think. But you can inject knowledge all over the place into the system in a way that is helpful, even if you are not their constituent. It's just a matter of the politesse of the call.
Patrick McKenzie: And a bit of the habits of Danielgerous Professionals here that might be useful for people. If you call up, they will pick up in two to three rings. It's an amazing feature of the American political system, which people largely don't get that experience when they call into agencies directly.
But you get picked up in two to three rings by a young person in typically constituent services. They are not the person that has decision making authority with regards to whatever it is you want to do. But if you can sound like an intelligent person that is not a serial killer for about 30 seconds, you can often get information on how to route something within the office.
And so if there's a particular bit of legislation that you are interested in and the elected representative made good comments at a hearing, you could say, "Hello. I just watched the blah, blah, blah hearing on C-SPAN. I liked the Member's comments. I had some comments relevant to the thrust of their argument. Who would be the best person to address them to?" The next words outta their mouth will often be, "Oh, Daniel, the legislative aide, is—if you mail us the comments or Daniel at blah, blah, blah, he'll get them. Thanks very much." "Can do that."
Also another micro tip. There are a lot of ways to use money poorly in politics and use it in ways that are either illegal or which fail to accomplish goals. One way, which is certainly legal and absurdly leveraged for using money to accomplish goals is if you ever need to send a letter to a member of Congress or similar, FedEx, $15, they get it tomorrow and it will say to everyone who like passes the chain of custody of that letter, someone felt sufficiently interested in whatever the contents of this envelope is, that they paid $15 to get it here, which automatically prioritizes you above 98% of incoming correspondence, including correspondence from the government itself. So absurdly useful.
Daniel Golliher: Yeah. And then all of this is part of the broader tool set of things people can use to influence policy, move policy. There is a very broad and wide tool set, and with some deliberate training, you can become quite effective with it. And again, I would emphasize it is in many ways easier to be far more effective sub-federally, again, because there's much less competition in a variety of respects.
Running for office
It's also much easier to be the person in the elected seat generally. So people are often intimidated by running for election or seeking appointment. In New York City specifically, although these kinds of opportunities exist differently everywhere, we have a lot of offices, it is possible to find something to run for. You can run for an election that nobody will ever hear about. And if you lose, nobody will ever know.
And so if the idea of running for office feels scary to you, but you do think better people need to go into the system, it's possible to get your reps in, to sort of get the muscle memory for how to run a campaign, how to get signatures to get on the ballot, how to knock on doors, et cetera. You can do this in ways that have very little downside for your reputation, so you can get used to it. It is not as scary as people think.
Patrick McKenzie: I've met a surprising number of people over the years who decided to run for either state level legislative office or literally run for Congress, where random person in the tech industry, not who I would expected to try to run for Congress, but apparently it's like achievable if, you know, some people have a hobby that is woodworking. I do small plastic models. Some people run for Congress.
And at least one or two of those people accidentally won the election. And so, what can you do? It might be useful state knowledge for people, for many local elected officials, the office that they're elected to, it's full time in that they are 100% of the people elected to that office, but it is not 100% of the professional output that they're going to do in a year. And so particularly common in small towns, but I would assume even in New York, many members of the school board, for example, may be full-time members of the school board and similar.
Daniel Golliher: Yes. Although, again, you find weird—government is very strange and there's a lot of strange context. So for example, if you compare the legislature in Albany to the legislature in city hall, City Council—the city council is really, it actually is a full-time job, for real. Like you get paid close to $150,000 a year and it very much is a full-time job.
Strangely, being a state legislator in Albany is not considered a full-time job. And you get paid less than you do if you're in the New York City Council. And the New York City Council sort of sits perpetually. They do take some breaks, but they're generally always in session year round. The New York State legislature is really only in session for maybe four or five months at the top of the year.
And so you would sort of think, "Oh, it's the state legislature. It would be more active than the city." And just that's not the case. And for people who might not know, state legislatures vary quite a lot. Some of them meet only like four or six months out of every two years, for example. So the amount that that is a full or a part-time job can vary quite a lot. It can mean you only meet every other year, not just you only do 20 hours of work a week or something like that.
Patrick McKenzie: I think the historical explanation was that way back in the day, very many legislators were large landowners or farmers, and literally tied to the planting cycle. But these days there are certain jobs that have embedded flexibility in them that tends to act as a great filter on who joins legislature. And so people will frequently show up from entrepreneurial backgrounds, lawyers, et cetera, who have some ability to tell the firm, "Okay, I will have less billings for the next six months because I am down in the state capital getting stuff accomplished. And then I will be back to normal for 18 months and then repeat the cycle."
And we put up with it, one, because I'm a partner at this firm, and two, because broadly speaking, there is no law firm that does too poorly because too many of its people get elected.
Blue tape
Patrick McKenzie: Okay. So you've developed on your blog this concept, which you call Blue Tape. Can you tell people a little bit about it and how it contrasts with the sort of more typical model as to government inefficiency?
Daniel Golliher: Yes. So I use Blue Tape because the color is partisan-coded, for the Democrats.
[Patrick notes: As a service for non-American readers: the parties are coded blue for Democrats and red for Republicans, by ancient practice. It was my perception as a young adult that the code words metastized to usage outside that after the Bush/Gore electoral map, colored in the traditional fashion, caused repeated rounds of commentary on “blue states” and “red states.” This overly reductive explanation of the geography of American politics has led to more sophisticated commenters talking about “purple states.” ]
A common line of commentary now is red states can build things, blue states cannot, housing or green infrastructure or whatever. Like for example, Texas has pulled well ahead of California and installed utility solar capacity, even though Texas is not particularly known for banging the green drum. And maybe they don't have as many subsidies, and yet they still seem to install more than the state that on paper says they care about it much more. So why is that happening?
Patrick McKenzie: This paragraph was published virtually verbatim in Complex Systems earlier today. [Patrick notes: This episode was recorded on April 17th, a few hours after the episode with Tim Fist had made substantively the same point at substantial length.]
That aside out of the way: Unfortunately that that has indeed been how the world has shaken out.
Daniel Golliher: And you can point to many examples of this across many domains. And so I call it blue tape because, well, if it's—I kind of want red tape. If red tape is what enables you to install all these things and build all these things, maybe we need more of it. It's really blue tape where a lot of the problems come in.
And why are some states seemingly so unable to build the basic necessities of basic material comfort for their citizens, even if they're the ones proclaiming from the mountaintops that they very much value this and want to do it? And so this can lead a lot of people to be frustrated, to burn out, to be pessimists, to have a bad time. They just say, "Woe is me. These states are broken forever. I'm gonna give up on them."
And I usually say two things to them. Number one, how high a resolution picture into how the state government works are you operating on? Clearly not everything is broken. Many things still work, so don't be so pessimistic about it. I bet you can change the law so that things are not entangled in blue tape, number one, although it'll be a lot of work.
Number two, the government is huge, especially in New York. It does a lot of things. I bet there's a lot of stuff that's being improved that has never been in a headline and you have never heard about. So if someone is interested in building a model that has reasonable fidelity to reality, so if you think red states build, blue states don't, I think broadly that's true with regard to a lot of important things. But if you want a better model, you would think, "Well, is there anything that blue states are doing?"
For example, New York City almost went bankrupt in 1975. As a result, the state, New York State imposed a lot of controls, accounting controls on the city. One of those is imposed by a thing called the Financial Control Board. And since 1975, they review many capital contracts for capital projects in New York City.
And if it's above, up until recently, if a lot of those projects were worth $10 million or more, then they had to be reviewed basically twice by different sets of people according to different sets of standards. And this adds weeks and weeks and weeks onto the time for these contracts, which makes them more expensive, et cetera.
That was changed in 2023 in a way that was great, that sped things up, that made government better. And a lot of stuff like that's happening all the time under the hood. But people don't know about it. So one of the drums I bang quite frequently is, it's worth knowing about these things, number one, because it can pull you back from the edge of destructive pessimism. But number two, if you want a model that's oriented towards reality, you need to see where the system works and ask why did it work there and how do you extend that elsewhere?
Patrick McKenzie: I think it's also useful to underline, as I said earlier, that even the government is not always legible to itself. And so there's an exercise in software engineering, "put a profiler on it," where if you really want to know why something is not being delivered, whether that's housing or utility scale solar projects or similar, a very useful exercise is to diagram out, "Okay, what is the actual series of gates that this needs to go through? And then how long is each of those parts of the cycle time?" And just write that down.
Put a profiler on it. Where in this complex system are we actually spending the time?
And what we'll frequently discover as a result of putting a profiler around something is that a lot of the calendar time is being taken up by a process which takes calendar times for reasons. That process has a reason to exist most of the time. But there is simply no back pressure against that process for this case that will cause it to take any less than the amount of time currently allocated for it. And the amount of time currently allocated for it is absurdly long.
And so in a process that might take six months end to end, there might be 90 days in the middle where no one is doing anything simply because a particular department is allowed 90 days to do a review. And they've routinely turned that review around onto day 87, where the review takes them six hours to complete.
And so if you have that artifact, you now have that artifact being a description of the process. You can now go to many people and say, "Well, this is a political priority for us, right? I've identified exactly the problem and just this one 90 day schedule in the middle, and I have a candidate solution. Let's move that 90 day timeframe down to two weeks. It seems like that department would be able to do it in two weeks. In fact, I called that department and here are three senior decision makers who say, 'Yeah, we could probably do it in two weeks with appropriate resourcing, et cetera, et cetera.' So let's get that done." And now we've increased throughput through your system by, finger to the wind, failing to do fraction math, about 40% increase.
[Patrick notes: I apologize: my brain works in mysterious ways, and one of them is that I can produce intelligible speech or I can do mental math but I cannot do both at the same time. That’s a 70% increase in throughput if the step under consideration was 90 days at a 35% increase if it was 60 days. I used both those in the hypothetical then rounded to 40% because I didn’t catch the error in the scenario.]
Daniel Golliher: And if you want to become the kind of person who does these kinds of things, maps these kinds of things out, it does require technical knowledge. You have to be able to walk up to the system, look at it, understand what you're looking at, and then map it out, and then suggest an alternative.
Most competent people you would find walking around industry, you can become this kind of person well in under a year. If you want. You could do it maybe even in a couple of days, just really depending on what it is. But this is not beyond reach.
And usually what I—you know, I get a lot of inbound from people who want to become this kind of person in one way or another. They just want to be able to put their thumb on the scale. You maybe have to invest 10, 20 hours in learning, maybe not even that much. And you can become the kind of person who can do this.
Patrick McKenzie: And there's existence proof walking around in your government office of choice. Again, the legislation is written by people who have thought seriously about the issue for six weeks or less. And so there's a lower bound for how much you can learn about something in six weeks. Like you can only write the legislation that governs it for the entire nation.
But yeah, I would generally encourage people—you have more agency than you think. You are allowed to do more things than anyone has come up to you and told you you were allowed to do. And the subjective level of—a phrase in English—the subjective level of the artifacts that you have to deliver to meaningfully influence the political process is far lower than most people's resting state impression of how important, how well-crafted, how thoroughly researched, how institutionally blessed an artifact would have to be to influence the political process.
And so take some whacks at the apple. One of the nice things here is due to a variety of processes, success begets success. Even if you just figure it out how to do it in microscale for small issues locally, you'll get better at it for the next time, both because you understand the process more, and because you become increasingly legible to the system as a result of racking up successes.
Daniel Golliher: And there is a certain kind of person who gets, when I tell this story, or if they listen to this story, they get a little bit frustrated because they think something like, "Well, I'm very successful in industry. I'm delivering valuable products at great prices that are improving people's lives. Why should I have to pay attention to the government? And it will take me away from doing all this. It feels like a dead weight loss for talent to have to pay attention to the government." And I disagree with this sentiment.
In a self-governing society, you will need talent to pay attention to governing. But usually what I tell them is if the government doesn't work well, it's not gonna be good news for private industry. It's the fundamental substrate. It is property rights. It's contract. It's how fast you can get certified to do something. It's a good use of time for talent to pay attention to the government. And it doesn't have to be hard to get into.
Again, this is why I set up a school and I see the transformation happen all the time. My basic class is 10 weeks and the transformation that happens from people knowing nothing to 10 weeks of instruction with a couple of hours each week—it's a fundamentally different way of looking at the system.
Patrick McKenzie: I also think the skills are quite transferrable. I have this coinage Danielgerous Professional, where there is essentially a class that sits astride society. The same skills that make you good at getting banks to do what you need them to do when you write in a letter also makes you good at influencing the government. Also makes you good at getting better outcomes for yourself and your family when dealing with healthcare administrators and similar.
And so to the extent that you can cross-train in getting good at government influence at whatever your desired level is, you'll be better at understanding the bureaucracy you find yourself in during the day job. But you'll get better at a variety of sort of government adjacent things that work, dealing with regulatory affairs, dealing with government relations, communications, et cetera. As a knock-on effect of understanding how the broader system that you're meshed in operates, even if influencing that broader system directly is not the thing you spend 90% of your cycle on.
Daniel Golliher: A fun analogy, that's not even quite an analogy. The city of New York at its base is a corporation. It's a municipal corporation. So it is literally a corporation under the law. And so as a result, it shares many characteristics with say, a C corp in ways that people wouldn't expect.
And so if you have spent your life working in these kinds of corporations, you will find far more parallels to the system that is the city of New York, that is a municipal corporation than you would expect. So who are the decision makers? Who can you email cold? Who should you go through an intermediary for?
I mean, as a citizen, you can sort of think of your city council person as the head of a different department. You are allowed to email the head of a different department if it's relevant business that concerns you both. And as it just so happens, as a constituent of that legislator, that is business that concerns you both. And so thinking about the government as if you're all just part of—I've heard this analogy come from a lot of different people, but thinking of it as if you're just in one large corporation together is a way it helps snap into focus affordances, what you are allowed to do.
Patrick McKenzie: I think it also gives people an issue to snapping in the affordances, might cause them to recollect that, "Oh, I've been successful at things that did not involve just my direct team before. What were the things that were effective? Well, understanding who was having the meetings was effective, writing documents internally was effective, et cetera, et cetera. Oh, that's all things that are within my locus of control and have been this entire time."
So mutual exhortation that people just try it, see how you like it. The worst that—there's a great phrase about exercising. If you don't like it, you can have your old life back. And if you don't like your stint in political engagement, well you can have the same country you had yesterday.
Patrick McKenzie: So Daniel, where can people follow you on the internet?
Daniel Golliher: Thankfully my handle everywhere on the Internet is @danielgolliher. It's just that's my handle everywhere. Or see Maximum New York and you’ll find information about the school, the classes, and my general Internet presence.
Patrick McKenzie: Thanks very much for taking the time today. It was a great conversation. And for the rest of you folks, we'll see you next week on Complex Systems. Thank you very much.
