Patrick McKenzie — Money laundering, big tech censorship, SBF & Japan

Dwarkesh Patel: Today, I'm chatting with Patrick McKenzie. He is known for many things on the internet. He's known as Patio11. Most recently, he ran VaccinateCA, which probably saved on the order of high four-figure number of lives during COVID. He also writes an excellent newsletter called Bits About Money. Patrick, welcome to the podcast. Thanks very much for having me.

Patrick McKenzie: So what was VaccinateCA? In early 2021, we were quite concerned that people were making 20, 40, 60 phone calls to try to find a pharmacy that actually had a dose of the COVID vaccine in stock and could successfully deliver it to them. I tweeted out randomly, you know, it's insane that every person or every caregiver is attempting to contact every medical provider in the state of California to find doses of the vaccine. California clearly has at least one person capable of building a website where we can centralize that information and send everybody to the website. If you build that website, I'll pay for the server bill or whatever. And Carl Yang took up the gauntlet and invited 10 of his best friends and said, basically, all right, get in, guys, we're going to open source the availability of the vaccine in California by tomorrow morning. This is at like 10 p.m. at night, California time. And so I looked down into the discord where, of course, all medical infrastructure is built and gave a few pointers on, you know, making scaled calling operations. And then one thing led to another and ended up becoming the CEO of this initiative. At the start, it was just like this hackathon project of a bunch of random tech people who thought, hey, we can build a website, make some phone calls, maybe help some people find the vaccine at the margin. And it grew a little bit from there. We ended up becoming essentially the public-private partnership, which was the clearinghouse for vaccine location information for the United States of America. That felt a little weird at the time and continues to. Yeah.

Dwarkesh Patel: Okay, so the obvious question is, why was this something that people randomly picked up on a Discord server? Why wasn't this an initiative either by an entity delegated by the government or by, you know, the White House has, or the pharmacies have a website where you can just sign up for an appointment?

Patrick McKenzie: Oh, there are so many reasons and a whole lot of finger pointing going on. One of the things was that there were almost no actors anywhere in the system who said, yes, this is definitely my responsibility. Various parts of our nation's institutions, county level, public health departments, governor's offices, the presidency, two presidencies over this interval, which will become relevant. You know, they all said, well, I have a narrow part to play in this, but someone else has to do the hard yards of actually like putting shots in people's arms. And someone else is clearly like dealing with the logistics problem, right? And the ball was just dropped comprehensively. And no one at the time really had a plan for picking it up or didn't feel like it was incentive compatible for them specifically to pick it up right now. It would be great if someone could do this, but just not me.

Dwarkesh Patel: Okay, so to explain the context and how important it was that people get vaccines at the time and how much these delays mattered, you can account for it, obviously, in the amount of lives saved. You can even look at it in terms of when vaccine news was announced, how much the stock market moved. It was clear it was worth trillions of dollars to the economy that the vaccine be delivered on time. And so it should be priority number one that people know where the vaccine is. A sort of meta question is why? Why are I kind of heard about this problem for the first time when I read your article about this? Why is this? There's there's a bunch of controversy after COVID about people pointing fingers about masks or different kind of protocols. Why was this not people are getting called in front of Congress? Why were we not able to deliver the one thing that was needed to arrest the pandemic as fast as possible?

Patrick McKenzie: Oh, well, folks, why don't we 27,000 words on this? My article in Works in Progress called The Story of X and ACA goes into some of the nitty gritty. Broadly, I think a matter of incentives more than matter of people choosing to do evil things. Although I will say we did choose to do evil things and we can probe on that if you want to. But if you look at the federal government specifically, the federal government institutionally learned one, I believe, wrong lesson, which has terrible consequences from the healthcare.gov rollout a number of years ago when Obamacare was first debuting. And the thing which many actors in the federal government and the political parties came away from is that a president can doom their legacy of their signature initiative if those bleeping tech folks don't get their bleeping act together. And so the United States has decided like there is virtually nothing. up to and including like the potential of national annihilation that will cause us to actually like put our chips behind making a software problem that is somebody else who does not have to deal with an electoral mandate or getting called in front of congress or etc etc somebody else's problem unfortunately software is eating the world and delivering competence in the modern world requires being competent at software and the united states it will tell you differently and there are wonderful people in the in the government who are attempting to change this but broad strokes on an institutional level, the United States federal government has abdicated software as a core responsibility of the government.

Dwarkesh Patel: I understand why they didn't initially want to pursue this project. I still don't understand the answer to the question of after everything went down now, why is it not more of a news item that this was a problem that was not solved?

Patrick McKenzie: So I think we're memory holding a lot of things that happened in the pandemic. And I wish we wouldn't partly because, you know, of political incentives and we're approaching an election year and because of the uh quirky way that the american parties and the candidates bounce off each other is no one's and it's no one's real incentive to say okay like i would like to re-litigate the mask issue for a moment we told people that masks don't block airborne viruses uh and we were quite confident of that and the entire news media backed us up on it and then we were like 180 a month later um No one wants to relitigate that. No one wants to relitigate California imposed redlining in the provision of medical care. And that was wrong and evil. But the party that was in the case pro-redlining does not normally like saying that it is pro-redlining. And the other party does not really consider that a hugely salient issue. And so, you know, there are no debates and no one is asking, you know, Governor Newsom. So when you got on TV and said that you were doing geofencing for the provision of medical care, geofencing in that context was the same as redlining. Can you explain your support for redlining? And no one has asked Newsom that question. Maybe someone should. But we're surrounded by... the effects of incentives and the effects of iterated games, and sometimes they don't play out the way we would ideally like them to play out.

Dwarkesh Patel: Yeah, we can come back to this. I still don't feel like I really understand. So maybe it was that everybody has the blood on their hands, and so... That's so confusing. I think if you have other kinds of emergencies, like if you lost a war, I don't think you just brush it aside. The generals would have to come up in front of Congress and be like, what happened? Why didn't we get that battlefield? Actually, we just lost a war. Maybe that didn't happen.

Patrick McKenzie: If one goes over the history of military conflicts, I don't know how many people, losers on either side of the conflict ever actually did that reckoning of like, hey, could we attempt to win in the future? I think there was a broad lack of seriousness across many trusted institutions in American society, in the government, in civil society, in the tech industry about like really approaching this like a problem we want to win. And I think a wonderful thing about our country and our institutions is like on things that are truly important to us, we win outlandishly because we are a rich and powerful nation. And yet like this was obviously a thing where we should have decided to win and we fundamentally did not approach it as a problem that we needed to win on.

Dwarkesh Patel: Okay, so going back to the object level here, one would think that instead of different people calling different pharmacies and asking whether they have the vaccine, the obvious thing that people who have not read your article would assume is that either there were just some company would build a platform like this, the government would build a platform like this. I guess you explained that the government didn't do it. The pharmacies might build a platform like this. And I want to meditate on the incentives that prevented random big tech company or Walgreens from building this themselves. Can you explain that? Sure.

Patrick McKenzie: So the federal government and the state government, the American governmental system is quite complex. And there were multiple distinct supply chains with multiple distinct technological systems tracking where these vials were headed all over the country. And there were many attempts at various levels of the government to say, hey, can we commission a consultancy to build a magical IT solution which will get these databases to talk to each other? And those largely failed for the usual reasons that government software procurement projects fail. Why didn't tech build it? I'm constrained on what I can say and cannot say. So I know a little more than this answer, but I will give you part of the answer. The tech industry, both at the level of like Appamagoo Facesoft, which is my funny sardonic way to refer to like some of the most powerful institutions in the world, and many other places that hire, you know, many number of smart engineers who can build the world's least impressive inventory tracking system. felt political pressure in the wake of the January 6th events in the United States. This is another thing that's gone down the rabbit hole, but in the immediate wake of the January 6th events, people in positions of authority very clearly tried to lay that at the feet of the tech companies. And internally in the tech companies, their policy teams, the teams that are supposed to make the company legible to government and avoid government yanking its permission to do business, And their communications teams, PR departments, told everyone in the company, like, mission number one right now, do not get in the newspaper for any reason. We are putting our heads down. And when people in those companies who work on public health, and a thing that might not be obvious to most people in the world is that, like App Emigu Booksoft, they're literally like teams of people who their job is public health because they are the operating system of the world right now. And the operating system of the world needs public health care. Those teams said, hey, we've got this thing. And other people in the company might have overruled them and said, it would be really, really bad right now to have the tech industry saying we're better at the government's job than the government is. So shut that down. okay so that's that's so insane so it is absolutely the local incentives like it makes sense in the meeting when you're saying it and you are not in that meeting projecting i'm going to cause tens of thousands of people to die at the margin by making this call and yet that call was made right okay so the um

Dwarkesh Patel: There's two culpable actors here. One, you could say, well, the big tech companies for not taking the political risk. I think what is even more reprehensible is the fact that they probably correctly thought that appearing more competent than the government and saving tens of thousands of lives as a result will be held against them in a way that significantly matters for their ability to continue their other businesses. um now okay so suppose that they had built the software um let's play with the scenario and then what would happen they would get hauled in front of congress and explain why they weren't delivered even faster because of the ultimate bottlenecks in the supply chain because you can only manage what would happen like they build it and it's like better than the government so many things

Patrick McKenzie: could happen. One, if you build the thing, this has sometimes been called the Copenhagen principle of culpability. If you build the thing, various actors in our system will assume, okay, now your responsibility for not just the consequences of the thing you built, but for the totality of consequences of everything associated with the American vaccination effort. So you built the thing. Oh, you big tech geniuses. Well, what did you do about localization? You didn't do enough about localization yet. You hate name group of people here. Like, Don't you see the disparity in death rates between demographic A and demographic B? Why haven't you fixed that yet? You have killed so many people, et cetera, et cetera. And no one in government, no one who is making that moral calculation says, I have responsibility for killing people by doing nothing. The person who is doing anything has responsibility for killing people by taking up the burden of doing something. And it is a absolutely morally defensible thing, which you will see over and over and over again in our discourses.

Dwarkesh Patel: They get hauled in front of Congress, but it's not just because they made a sin of commission. It's also because if you're right, then after January 6th, they would be held in. there's one answer where it's like they touched the problem and there's another where it's they did it they did it better than the government could have and those

Patrick McKenzie: seem like two different right uh you touched the problem and so you've uh immediately taken liability for any number of sins of a mission because uh even at the scale of the largest companies in the world you have not allocated infinite resources to this problem and also uh the stealing a march on the government and embarrassing us will be held against you and so like you can point back to the cambridge analytica thing where Cambridge Analytica is like shorthand for, there was this one time back in the day where the news media in New York and the government in DC and places like it talked to each other a bit and convinced themselves that a small team of people with budget of approximately $200,000 have rooted the United States presidential election. Now, rooting the United States presidential election, perhaps on behalf of a foreign power, would be an enormously consequential thing. Good thing that did not happen in the world that we live in. However, people believe very passionately in that narrative. And as a result of that narrative, they did very aggressively attempt to clip the wings of tech and tech's core businesses. Like say advertising, right?

Dwarkesh Patel: There's so much that's crazier, okay one question you might have is we figured out that our we couldn't delegate to big tech or any of the competent actors and that the Native infrastructure that we had that was specifically earmarked for dealing with public health emergencies was extremely incompetent to the extent that discord servers vastly outperformed them and Supposing that public health is not uniquely incompetent among the different functions that the government is supposed to perform that don't get tested until the actual emergency is upon hand. How would we go about, if the president cures this and is, let's say, concerned that the people running that nuclear bomb reaction aren't up to snuff, or the earthquake reaction, Is there some stress that is tests that you could perform on these institutions before the thing goes down that you could learn beforehand, whether they're competent or not?

Patrick McKenzie: If I look at the experience of the last hundred years and a little more back to the flu pandemic in 1918, if you read histories of the flu pandemic, a vastly less wealthy, vastly less technologically sophisticated nation with many less people involved in the actual fixing of this problem, competently executed on nationwide vaccine campaigns and other various measures. And so in some fashion, we should be urgently concerned with what decayed institutionally in the interim. I think one of the things that health departments specifically faced is that if you had just given them these kind of vanilla vaccination campaign, maybe they would have done better than they actually did. I'm not positive of that. I do think there are actually, and here I have to stop for a disclaimer, I think that people in county health departments did real important work. I think they probably did work that saved lives on the margin. I do not think that the United States should be satisfied with our performance in 2020 and 2021. We should be very dissatisfied and we should get better for the future. And that requires like recognizing that we underperformed by a lot. There was a political decision made that the successful administration of vaccination was not going to be measured solely by saving lives. The prioritization schedule that came up, which was Byzantine and complicated and routinely befuddled professional software engineers and health administrators, and which I could not diagram out on a whiteboard, even if you paid me a million dollars to get it right on the first time. was downstream of the United States as political preferences. And so schedule 1A versus 1B versus 1C was like in the first five seconds of the discussion dictated by medical necessity. But immediately after that was about rewarding plums to politically favored groups. And so one of the complexities of this is that the pharmacies and healthcare departments are not set up to discriminate along the axes of whether one is politically powerful or not, because that is not a thing that they have to do in most vaccination campaigns and not a thing that they have to do on like, the typical Tuesday of providing medical services. We asked them to do this like radically new thing, which is in part responsible for the failures that we had. If we had had a much simpler tiering system, for example, we would have had more than 25% of the shots successfully being delivered in the state of California in January of 2021.

Dwarkesh Patel: Can you, for people who are not aware, what was the political tiering system that you're referring to?

Patrick McKenzie: Oh goodness. So this was different in different places and confusingly different places in the United States use the same names for these tiers for different people. But in the state of California, Tier 1A comes before 1B, comes before 1C, comes before the tier 2, et cetera, et cetera. So 1A was, descriptively speaking, at the start, and this changed over time on like a day-to-day, week-to-week basis, sometimes in mutually incompatible ways at the same time. It was an entire mess, but... Descriptively, tier 1A was, okay, healthcare professionals and a few others and people above the age of 75. No, wait, we'll change that to 65. Tier 1B was, we're going to put a few favored occupational groups here and some other folks. And tier 1C was people who doctors think will probably die if they don't get the vaccine, if they contract COVID, but who have not appeared in group 1A or 1B yet. And so who got 1A? Well, healthcare professionals. So like doctors administering the vaccine. That sounds like pretty reasonable. Also veterinarians, because veterinarians are like urgently required by society at the time. Not so much because the California Veterinarians Association is good at lobbying. And that isn't just me alleging that. They sent a letter out to their membership saying, we are so good at lobbying. We got you guys into 1A. Congrats and go get your vaccine now. I have that on my website. Okay, so like tier 1B school teachers were classified as a tier 1B. Why? Because go figure teacher students have political power in the state of California. And they said, well, we'll accept not being in 1A, but we are going no lower than 1B. And probably no one in that meeting ever said like, I definitely think that 25 year old teachers who are currently under stay at home orders should be in front of the line of people who will die if they get COVID. We made that choice.

Dwarkesh Patel: In your article, you discussed that the consequence of that was not only the misprioritization of the vaccine, but the bureaucracy around allocating it according to these tiers, resulting in 75-year-olds not having the capacity to fill out the pages of paperwork that were

Patrick McKenzie: You're required to decide what tier you're in. The state of New York commissioned a consultancy to administer to 75-year-olds a 57-page web application, which required uploading multiple attachments to check for their eligibility. And, like, talk with a technologist if you don't believe me. We try to remove everything from a web page that people successfully get through it. Like, if you can make it, you know, two to four fields, that's already taxing people's patience. Yeah. And you are asking people who might be suffering from cognitive decline or be less comfortable in using computers to do something which would literally tax the patience and cognitive abilities of a professional software engineer. That wasn't an accident. We wanted to do that. Why did we want to do that? Because it was extremely important to successfully implement the tiering system that we had agreed upon. Why was it extremely important to implement the tiering system? Because that was society's prioritization. Was that the correct prioritization? Hell no.

Dwarkesh Patel: Can we just count off everything it just so it's it's enraging not only because obviously people died but because like nobody talks about it it's it was um there's all kinds of controversies about covid about whether i don't know a vaccination and um side effect kind of things and whether the masking orders were too late, too early, whatever. And then the main thing about whether we got vaccines in people's arms on time because of these political considerations. So you're not allowed, yeah, go ahead.

Patrick McKenzie: Can I jump in with one bit of optimism?

Dwarkesh Patel: Yeah.

Patrick McKenzie: We achieved something incredible, which was getting like the first cut of the vaccine done in two days as a result of, you know, many decades of science done by very incredible people. And we successfully got that vaccine productionized in a year. We should have got it productionized in far less than a year. Right, right. The fact that we were able to do it in one year and not three was enormously consequential. And so we should feel happy about that. A little annoyed that we didn't have better protocols at the FDA in places to get that vaccine prioritized for testing much faster than was. Quite annoyed at the fact that that was a political football and people probably made decisions that pessimized for human lives and optimized for defeating a non-preferred political candidate.

Dwarkesh Patel: Are you talking about the fact that the vaccine was announced the day after the election results or something, right? Yes, I'm basically subtweeting that.

Patrick McKenzie: Yeah. And I strongly believe that was a political decision, but I don't, I'm just a software guy.

Dwarkesh Patel: So, okay. The particular kind of craziness that we had during this, during the two... at 2020 and 2021 about equity and wokeness, how much was that uniquely responsible for the dysfunctions of this tiering system and geolocation slash redlining would basically if it was if that happened in another year where there wasn't a bunch of cultural craziness, would it have gone significantly better?

Patrick McKenzie: It's difficult to ask that question because like we were clearly in a unique time in 2020 and 2021 and yet like Point to me in the year in American history in which American society was truly united and had no social issues going on. And if people like counterfactually like point to say World War II, I will say like read more history. But be that as it may. Was it the case that like strong societal feelings in the wake of, you know, George Lloyd's death in 2020 and the racial reckoning, like strongly dictated policy? Yes. That's a positive statement rather than a normative statement. That is absolutely the case. There's this thing we often say in the tech industry called bike shedding, which is if you're building a nuclear power plant And many people cannot sensibly comment on what is the flow rate through the pipes to cool a nuclear reactor. But if you build a bike shed next to the nuclear power plant, it's very easy to have opinions on the color of the bike shed. And so in the meetings about the nuclear power plant, you will have... a truly stupid amount of human effort devoted into what colors you should paint the bike shed. So it is very difficult for most people in civil society to successfully inject a vaccine into someone's arms, to successfully manage a logistics network, to successfully build a nationwide information gathering system to centralize this information and pass it out to everyone. And we aggressively trained the entire American professional managerial class starting at seventh grade or earlier in decrying systemic racism, which to be clear is a problem. And so any discussion about what should we do with regards to information distribution, which goes out to a broad audience in the American professional managerial class, who essentially call all the shots in the US system, will almost invariably get bent to, I have no particular opinions on server architecture here and nothing useful to comment, but what's our equity strategy? And the equity strategy dominated discussions of the, the correct way to run the rollout to the exclusion of, you know, operationalizing it via medical society, medical necessity. People brag about that fact. That fact is enormously frustrating to me. And if you say it with exactly those words and emotional valence, people will say, no, no, that's not exactly what we meant. But when they're talking to other audiences, they'll say, no, this is absolutely what we mean.

Dwarkesh Patel: Yeah. Okay. So, I mean, even on that point, maybe the culprit here is a scarcity mindset involved here with caring more about the proportion rather than just solving the problem. This is where...

Patrick McKenzie: This is one of those few times where we were genuinely up against a scarcity constraint that like, you know, physical reality was there were a scarce number of vials and we needed to have a prioritization system. And some people who urgently needed the vials were not going to get them first. Everyone was going to get them eventually. But the mad rush in our political system to dole out favors around the prioritization for those first vials. exceeded the actual distribution and successful injection of the vials as a goal. Again, California reported to the federal government that it was only successfully injecting 25% of its allocation. It had the most desirable object in the history of the world. And rather than adopting any sensible strategy for getting it into people's arms was bickering over who should get it first. We should be outraged about this, and we're mostly not.

Dwarkesh Patel: I don't even know what to ask next because it's so obviously outrageous. And there's no clear answer, at least to me, of why there isn't more outrage about it. Also, what the solution to it is... I mean, literally in the exact context of what would we do the next time there's a pandemic, it's not clear to me that we've learned the lesson, let alone the broader lesson of if there's a different kind of emergency, if there's some isomorphic emergency, would our state capacity be better? And you mentioned the point about 100 years ago, we maybe would be able to deal with this problem better. I don't know, what changed?

Patrick McKenzie: One of the things that America used to do is when the federal government lacked state capacity for something, it would say, who in civil society or private industry has capacity for this? And then say, congratulations, by order of the president, you are now a colonel in the United States Army. Like, what do you need to get it done, sir? And that was an option. That was an option that was not taken, but, you know. I will play no fights in either of the two administrations that both individually made terrible decisions, but plausibly some more enlightened counterfactual administration could have gone to Google and said, who is literally your best person for solving the data problem that we currently find ourselves in? Great. Will they accept a commission as Colonel? Great. Here's an order from the president. You have a swearing-in ceremony starting in 30 seconds. And you will present your project plan tomorrow. And again, the successful project plan was made by a bunch of like rank amateurs at this topic on Discord in the course of a couple of hours. And similarly, this is like one part of the huge overall vaccination effort. But you could imagine going to Amazon and saying like, hey, Amazon, we hear you're pretty good about getting packages between A and B. This package has a really hard thing about it. It has to be cool as it's delivered. That's like a totally unsolved problem in material science, right? And Amazon would say, we literally do that every day. Yeah. Back in December, people were getting on the night in the news and saying this is going to be an unprecedented logistics challenge because the vaccine has to be kept at ultra low temperatures, which are the same temperatures at which milk is transported. We understand how to do cold chain logistics. So Amazon would correct that misperception. They say, oh, you guys seem to know what you're doing. We have an absence of that here. Congratulations. Here's your colonel uniform in the United States military. And now your job is we are going to give you a CSV file every day, interface with this other colonel, please, from Google on like where this thing needs to go. And you get it there on time every time. And if you can't get it there on time every time, like call the White House and we will like find you a political cover is what a functioning system would have done. Yeah. Granted, the American system is dysfunctional in its own way. I think another thing that's been underdone in the course of the last couple of years is looking internationally. I don't know if any country anywhere with vastly different political systems is happy with the outcomes that it got. Some were obviously vastly better than others, but there are journals of comparative international politics, and why are those journals wrong? writing anything but like who succeeded at what margins and who did not and what do we learn about like the proper functioning of political systems civil society and like the united states consider it as one like hugely complex uh machine no that

Dwarkesh Patel: okay that's a really interesting point i actually asked tony blair tony blair institute they were recommending to the british government different ways of distributing the vaccine and they they made the obvious recommendation that you should give everybody one dose now and then do the second dose later. You know, obvious things like this, which would save lives. British government didn't do a good job there. I think it's actually a very interesting question. There's governments all across the world which have very different political systems. They have, hopefully, I don't know, different infrastructure would deal next to this. Why did nobody get this right?

Patrick McKenzie: So on the, like, give people the catchphrase for this was first doses first. Yeah. That was not the procedure in many nations, which have many smart people in them. And it was not the procedure in the United States until, I think, you know, first doses first. You can, like, sometimes trace policy back to individual blog posts. And so to the extent that one can be traced, I think it was Alex Tabarrok on Marginal Revolution who was right. This is very obvious and overdetermined. If we want to win at this, first dose is first is objectively the correct policy. And after this ping-ponged around the political system for a while and they talked to medical experts and et cetera, they were like, yeah. Yeah, it turns out that, you know, this is the equivalent of like saying you should probably consume calories at some point in the typical week. That is better than not consuming calories. Well, we checked with the medical experts and it took six weeks of meeting, but they definitely agree. Beating. Beats not eating for living. So we're going to do that now. And on the one hand, like it is a genuine strength of the United States that like, you know, he's not just actually some rando, but relative to like the topic in question, some rando on the internet, like wrote up a 2000 word blog post and we stopped doing stupid things. We could have stopped doing the stupid thing sooner. Yeah.

Dwarkesh Patel: That doesn't answer the question of why nobody got it right. If you think there's something particular to the late stage bureaucracy that we have or something, maybe another country is fresher. But even the countries that have more authoritarian models who can just crack down or something, they did abysmally as well. They made errors often in many cases that were worse in America. There's so many countries, Patrick, why did none of them get it right?

Patrick McKenzie: I'm under-informed on much of the international comparison, partly because in 2021, I was sort of busy. But I remember Israel, for a variety of institutional reasons, having a broadly functional response on this, in particular around end-of-the-day shots, which end-of-the-day shots are, in the grand scheme of things, a minor issue. But they're a... A good like quick heuristic for do you have good epistemics on this at all? Okay. You know, physical reality of the COVID shots is there is five, eight or 10 shots in a single vial. That single vial goes bad after 12 hours. That is a bit of an oversimplification. It doesn't actually go bad, but for essentially regulatory reasons, we have to pretend it goes bad after 12 hours. It can't be resealed. Okay. So if you vaccinate two people, then the other shots are on a timer and those shots will decrease in value to zero after however many hours are remaining on the timer and then get thrown in the trash can. So quick question to test if you are a rational human being. At the margin, would you prefer giving a shot to the most preferred patient in your queue of patients who needs it for medical needs or like the trash can? You prefer the most preferred patient. Now, follow-up question. Would you prefer giving it to the least preferred patient or the trash can? You still give it to the human rather than the trash can. And Israel adopted the policy of like, if the shots are expiring, forget the tiering system, forget anything else. Literally walk out into the street and say, I've got the COVID shot, I need to administer it in the next 15 minutes, who wants it? And in the United States, we had a policy ban on doing that. We said, no, to protect the integrity of the tiering system, to embrace our glorious cause of health equity, you should throw that shot out. And that policy was stupid. And it was announced by governors proudly in December in front of news cameras. And then a couple of weeks later, reality set in and they were like, people told them, sir, it turns out that throwing out the vaccine is stupid. And the governor did not go go on the nightly news again and say, I gave a, you know, a very confident policy speech a month ago in front of this news camera where I said that I would prosecute anyone who gives out end of day shots. But he literally said, He literally said that. This is almost a direct quote, and you can see the actual direct quote in my previous writing on this, but I will not just prosecute people. I will go aggressively to try to maximize the reputational impact to your firms and your licenses. We were pointing metaphorical and, when it came down to it, literal guns at physicians in the middle of a pandemic for doing unauthorized medical care. crazy. But when the system corrected, it did not correct all the way. The governor did not go out and say, hey, that thing I said a month ago was effing insane. I take that back and apologize. No, it was like, okay, we're going to quietly pass out the word. That's no longer the policy, but we don't want to own up to the mistake. And people in, say, the regulatory departments of pharmacies make rational decisions based on the signals that you're giving them. And the rational decision a pharmacy makes is not, okay, we've been quietly passed the word that the old policy is persona non grata. But can we really trust the quiet word here? One, because like, do we trust that this actor is not gonna change their mind in two weeks and consequence us for something we authorized today? Just throw out the shots. And pharmacies did not cover themselves in glory. A lot of pharmacists did. Some pharmacists did. But pharmacies, like institutionally, we deliver almost all the medications for almost all the diseases routinely in America. We cannot blow up either that position of societal trust or our business results over one drug for one disease. And so throw out the shots and make sure we can still delivering medical care in California. tomorrow. I understand how that decision was made. We should not endorse that decision. There were individual acts of heroism by particular pharmacists who said essentially in as many words to us when we called them and said, hey, what's the procedure for getting the shot? Okay. An individual like the one you just described cannot formally get the shot right now. So I would tell that individual to go to the county website, tell whatever lies are necessary to get an appointment with me. They come in for an appointment and I will inject them rather than verifying the lies that were on the appointment card. Because basically F the rules, I swore an oath.

Dwarkesh Patel: Honestly, I don't know where to begin with some of these things. I want to understand a bunch of the... First of all, can we just go back to 25% of the vaccines that were allocated to California were actually delivered in people's arms? Literally the entire world economy was bottlenecked.

Patrick McKenzie: on this, right? So if you want another funny anecdote, I then asked some people in positions that might know, so how real do you think that 25% number is? And they said, well, the good news is in addition to being incompetent at delivering the vaccine, we're also incompetent in counting. So it was probably a bit of an undercount. I'm like, oh, so the good news is like the true count was like 100% or 95% or something. Well, no, not nearly close to that. But we got better at counting after the governor yelled at us because he was embarrassed. We were the 48th state.

Dwarkesh Patel: Like counting the item, which is the bottle. Like where is the thing that is going to like rescue us from the thing that is destroying the world? We had,

Patrick McKenzie: so like pharmacies, generally speaking, could today, if you ask, you know, someone deep in the balls of pharmacies accounting department, could you, by the end of the business today, give us a count of like how many bottles of aspirin the pharmacy has? physically in the world. By the end of the day, they would have a shockingly accurate number for that. It wouldn't be exact, but it would be shockingly accurate relative to that number is truly millions. If you could say, break it down by address, please. Where are they physically present in the world? Yeah, easy problem. Managing inventories of drugs, that's what we do. The United States could not do that and did not perceive that to be an urgent problem to be solved.

Dwarkesh Patel: I do want to ask you about the actual finance and software stuff at some point, but I think this is such an important... I mean, the world is watcher standstill. We still haven't learned a lesson. So I'm just going to keep going on this topic because I still don't understand the. OK, so here's here's another question that's sort of related to this. You have many rich tech industry friends. And I read your article and you're saying, we're trying, I'm filling out these grants for 50K and that's like taking up all my time. And I'm trying to raise, you know, a couple hundred grand here, a couple tens here. And I'm thinking to myself, how is this not as trivial a problem as, hey, XYZ, if you give me a money that you can find between your couch cushions, we will save thousands of lives and get the world economy back on track. How is raising money for this hard? Or why was it hard, you know?

Patrick McKenzie: So again, like trillions of dollars are on the line. The United States is spending tens of billions of dollars or more on its COVID response strategy. Like the true biggest issue is like, why is it coming down to like Patrick McKenzie's ability to fundraise in the tech industry for us to like have a system here? Okay, bracketing that, the tech industry underperformed my expectations for what the tech industry should have accomplished here. There were some bright spots and less bright spots with regards to the fundraising project. For those of you who don't know, the total budget of this project was $1.2 million, which is not quite couch cushion money, but is not large relative to the total amount of resources that the tech industry can deploy on problems. And in some cases, I looked at my email this morning to refresh my memory. I sent a CEO at a particular company, and I'm not going to name people, but they're welcome to claim credit if they want to claim credit. I emailed the CEO at a particular company. Hey, I saw you like to tweet about this on Twitter. I'm essentially raising a seed round except for a 501c3 charity. And, you know, we urgently need money for this. Here's a two-page memo. And I sent that email at 4.30 p.m. And 4.30 p.m. California time. And he got back to me. There was some internal emails of like routers to this person, routers to this person. He run that by blah, blah, blah. 9.30 the next morning, he said, I'm personally in for $100,000 out of my own pocket. My banker is going to contact you as a wire clear the same day. So, yay for that. On the, like, yes, less yay side, like, tech is not exactly a stranger to having bureaucracies. And in some cases, it was a matter of, like, oh, indicatively, we want to support that, but we have a process. And that process went on for six weeks. And by the time six weeks was over, it was May and not to the credit of funders by May. Most people in the professional managerial class who had prioritized getting a vaccine for themselves and their loved ones had succeeded at that. And they said, okay, so the vaccination supply problem pretty much solved, right? I'm like, no, it is not solved right now. It is solved for the people who are smartest about working the system in a way it was not solved for even them back in January. But there are many people who are not yet vaccinated and they say that's a vaccine hesitancy issue. No, it is not merely a vaccine hesitancy issue. It is still the case that there are logistical problems. It is still the case that people don't know that you can just Google the vaccine now. It is still the case that around the edges of the American medical system in places that are like underserved, etc., people don't have it or they can't get transportation, etc., etc., You should continue funding this team for the next couple of months so that we can do what we can around the edges here. And I was told, again, you know, people can do what they want with their own money. And I understand that charitable funders have many things. It's just like, okay, relative to the other places we can put money to work in the world. Further investment in the American vaccine supply situation as of May and looking forward doesn't make sense for us. Could you do it in another nation? And we said like, okay, we're the American effort. We have some advantages here. We would not have them in the other nation. We did talk to people there. We tried to see if we could help a team there or go there, etc. But we don't see that there's a path to positively impacting the problem there in a way that there's manifestly a path to positive impact here. And we lost that argument. We didn't get the money. The last $100,000 in was my daughter's college education fund.

Dwarkesh Patel: Oh, my God. Okay. Look, I agree that it shouldn't be up to tech to solve this huge society-wide problem. But given that nobody else is solving it, I still don't understand Um, have, have you gone back to any of them or have any of them reflected on, yeah, maybe I should have just, you know, wrote you a million dollar check and saved you all this hassle. So you could have gotten back to business.

Patrick McKenzie: So, uh, you know, ultimately I'm the CEO, like responsibility for fundraising, uh, lies with me. And so, uh, like I thought any number of things about how could I have done that better? How could I have strategized that, you know, I, I did not stop fundraising efforts, but I stopped lighting up new conversations for a number of weeks because I thought, okay, we've got the $2 million that we need to run this till the end of August. And that's my sort of internal target for the point at which it doesn't quite stop being useful, but it starts like actually being, you know, a question on the margins where it's not a question until the end of August. So could I have done better? Probably. Some of the folks in the broader effective altruist community, I'm not a member, but I've read a lot of stuff that they have written over the years, and I broadly consider them positive. They are the but-for cause of VaccinateCA, but ask me about that in a moment. Some EA funders talked to me after my piece about it had come out and said, this is physically painful to read. We wrote bigger checks with less consideration to projects that had far less intensity of success. Why didn't you just ask us for money? And the answer there was twofold. One, I thought I had high quality introductions and a high quality personal network to people who were likely already going to fund it. And so I didn't light up additional funding sources. And two, this is a true answer. I'm a flawed human who has a limited number of cycles in their day and was running a very, very complex operation. And it literally didn't occur to me, hey, maybe those people that have been making a lot of noise about writing a lot of money for pandemic checks would be willing to write a pandemic check, which like that was not entirely an irrational belief for me because I had like reach out to people who are making a lot of noise about writing money for pandemic checks. And they said, not in the United States, not in May. And I thought, oh, well, it's, you know, if I light up a conversation totally cold with someone now, it's likely to just get a no again. I should try to like scrimp and save and break the piggy bank for my daughter's college education fund. Which by the way, she'll go to college folks, no worries. But it's a, like how far down the list of like plan A, plan B, plan C, we were down to like plan C at that point.

Dwarkesh Patel: Just to be clear, I'm definitely not blaming you. It goes back to the Copenhagen interpretation. No, but I should blame me a little bit because I should be rigorous about my performance. I think, you know, you go back to the commission versus omission. It's like the exact same reason that we shouldn't have blamed Google if they got involved. I did that to themselves and maybe made a mistake. It's like, come on, to remove the bottleneck that was... basically stopping all global economic activity and causing millions of deaths for the action you were taking about it. Take money out of your daughter's college fund. It's so insane. So can I say this?

Patrick McKenzie: There's a positive takeaway here. So there actually is a positive takeaway in that there is one tiny actor who understands that he has unitary control over some decisions, who is capable of betting boldly on those without a huge amount of process when it is important to bet boldly on things. Yeah. Not to toot my own horn here, this is literally what happened. On the first day, we're getting in Discord together and there's a bunch of infrastructure where we have to sign up. We have to get hosting, yada, yada, yada. And there's an annoying mechanical step at this point where it's like you have to put down a credit card for a potentially unbonded expense. And people were like, okay, There's a list of things that we want to do, but since there is no money here, I'll take this one and you take this one. And after I heard this conversation go on for two minutes, I said, this is not a conversation we should be having. Here is a debit card for my business, which have just spun up on the backend because this is literally my job, which has $10,000 on it. Spend the $10,000 on anything that accelerates this project. There is no approval process. Don't get a receipt. Don't worry about the paperwork right now. and why did I do that because we were like doing things like well you know okay there's like a scrape up like the information about where hospitals exist and what their phone numbers are is probably like scrapable from the internet for free or we could buy a commercial database but that's like a stupid amount of money it's like two thousand dollars I'm like Relative to the importance of this project, $2,000 is a trivial amount of money. Just spend the $2,000 immediately rather than spending like four hours writing a scraper. And we don't think about that in government procurement and in charities. We have some sacred virtues about like you must minimize waste. You must minimize opportunities for corruption. You must maximize for like the funders of the charities for their money. you know, line item, support of individual things that charities buys. And those sacred virtues conflict with winning. And at the margins where they conflict, we should choose winning. We should choose human lives over reducing corruption. Like, And one of the few things we were reflecting on is the tremendous amount of waste and fraud that happened in the PPP loans and other corona stimulus things. And I'm like, okay, I'm not just saying this to be contrarian, folks. we should be glad there was waste in like COVID stimulus. If there was no waste, we were clearly not choosing the right margin to focus our efforts on.

Dwarkesh Patel: So by the way, for the people who don't have context on how much money typically goes around in Silicon Valley, they think, oh, 1.4 million, how hard should that be to raise? If you right now, given your reputation, like literally tweeted out, I'm not gonna tell you my idea, but I'm raising $50 million seed around or something, I would be like, that's gonna get filled. I think people don't understand. I have friends who are 16 years old who have some GPT wrapper and they don't have to worry twice about raising $1.4 million.

Patrick McKenzie: Not trying to brag, folks. Just telling you the reality of Silicon Valley and also the reality I put in the document. I have some knowledge of how seed funding would work if I attempted to raise seed funding for a for-profit company. And I thought originally... We're probably going to be charitable, but I'm going to pitch this to people as essentially like a seed investment, which they expect to like spend all the money as quickly as possible and go to zero while driving the total addressable market of the company to zero. But I'm bummed. This is what passes for humor with me. And so I told folks pretty confidently in the first couple of days, I'm pretty sure I can get us $8 million. And then I was actually able to deliver on 1.2 after far more tooth pulling. But like, yeah, descriptively, if I was asking for a seed stage investment, if I wanted to get $8 million wired by tomorrow, I think I could probably do that. And that is a civilizational inadequacy because you can literally get $8 million for a blank check for something that has a profit motive behind it. But if I write on the check, hey, we want to fix the manifest inability of the United States to figure out where the COVID files are, that blank paper becomes less valuable by the fact of writing that. So maybe on reflection, I just shouldn't have told people and said, oh, the blank check company was this thing and we're making it a 501c3, which some ethics issues in that, but the ethics issues are less bad than allowing people to die.

Dwarkesh Patel: Yeah. Okay. So the last episode I released, at least while we're recording this, was about within AI... a former AI researcher who thinks that the field is progressing in such a way that you will need to nationalize the research in order to protect American national security. And when I hear this about the inability of the government to keep track of vials of COVID vaccines or to get them in people's arms, should for any other emergency that we might be worried about, whether AI, I don't know, the fallout of a nuclear war or something, Should we just discount any government response to zero and then just... If your plan requires some sort of competent administration by the government, discount to zero. It has to be something on the side.

Patrick McKenzie: So discounting to zero is like the opposite of wisdom here because we didn't accomplish zero. We accomplished a... extremely impressive thing in aggregate, which vastly underperformed like the true thing that we were capable of. And so have to keep both of those parts of the equation in our minds at the same time. I think that people in tech need to become radically more skilled at interfacing with government to the extent that it is, you know, we have some manifest competency issues in government right now. We can't simply sit out here and gripe about this on podcasts and et cetera, et cetera. We've got to go out and do something about it. I think it's been reported that there was a meeting among tech leaders early in the vaccination effort where a bunch of people got in a room and were like, this is going terribly. I hope someone fixes it. And I hope someone fixes it is no longer a realistic alternative. I think we have to be part of that solution there. Partly it's like having higher fidelity models for how Washington works than simply, oh, they're bad at everything. It is important to understand that the government has some manifest competence issues. It's also important when working with the government to understand that telling the government to its face you have manifest competence issues is not the maximally effective way to keep getting invited to the meetings. I was very religious about not criticizing anything about this Californian response effort in 2021 because we needed to be in the room where it happens. And, you know, that was a choice made. And am I 100% happy with that choice? No, but we kept some relationships that we really needed. And I'm not saying don't criticize the government, obviously, but be strategic about the sort of things, like play like you were attempting to win the game. And, you know, on the government side, one, like dispelling the massive UGG field that surrounds software, this is going to be a part of the future, whether you like it or not, we need to get good at it. We can no longer accept incompetence at this as the routine standard of practice in Washington. Two, you know, it is enormously to the United States credit that we have a extremely functional, extremely capable tech industry. maybe we shouldn't treat it like the enemy. Just putting that out there. Again, this is a thing the United States has done before. There are laws in place. There are decades of practice. We could put a colonel's uniform on somebody. Think seriously about doing that next time. Do I think we have institutionally absorbed all the correct lessons from this? know when I see like after action reports the after action reports praise a lot of the things that people think are like very important for maintaining their political coalition which were either not productive or anti-productive and they fail to identify things that were the you know the true issues and to the extent that they identify things that were the true issues the sort of recommended action is well I hope someone fixes this next time and It's no longer sufficient. The default case is that the ball will be dropped. And goodness, those of us who were involved in an accident CA kind of dread what we call the bat signal, where, God willing, there will not be another worldwide pandemic killing millions of people as long as we live. If there is one, we know what numbers to text to get the band back together. society should not rely on us as plan a how did this happen i i mean the the the

Dwarkesh Patel: point about uh gripping a podcast that's that's definitely what i'm doing but i just i maybe you're um humble to say this yourself but i i do want to commend you for there are very few people i think who kind of You tweeted it out. There are probably other other projects that other people could have taken up that are not taken up. In this case, you tweeted it out. You saw that there was a thing that could be done and you did it. You like quit your job and you did this full time. You even what this and the reason you had to dive into your a kid's college fund was because somebody who had promised a donation didn't follow

Patrick McKenzie: up on it right so effectively every time that the uh uh we had you say a verbal green light with regards to money i would uh advance the uh the company the charity charities or companies by the way folks i don't know if that has that is obvious uh was called the shots incorporated so I would in advance call the shots the money that was sort of soft committed before the money would actually arrive in the bank and the theory that like this accelerates our impact we should always choose acceleration over other things like say minimizing credit risk and then you know some of the people who had like soft committed did not actually end up wiring money at the end of the day. And I was like, oh, shoot. Well, you know, choices now are either don't run the last payroll or do run the last payroll and, you know, do not recover the money I've advanced the company. Well, okay. You know, do run the last payroll. Did you end up recovering it in the end? No. Ended up being a donation from me personally to the effort.

Dwarkesh Patel: what the fuck that is that is the least important part of the story folks uh sure but overall i'm just like like kudos isn't obviously enough to commend uh to convey what i mean to say here but like i mean yeah i'm glad you did that um and i'm grateful like four figure amount of lives you just like you it's hard to sort of plot that on a graph and make sense of what that means but

Patrick McKenzie: If I can be pretty explicit about it, to the extent kudos are deserved to anyone, Carl Yang for taking up the torch and finding 10 people in tech industry who would jump into something at 9 o'clock. Those 10 people, the other board members, the hundreds of volunteers, the team of about 12 people who worked on it full-time for very full definitions of full-time, virtually ceaseless for five, six months. And there were other projects in civil society. There were many people doing this as their day jobs. Like the American response effort is not one like small group of people anywhere. It's the collection of all these things bouncing off of each other. And I think like, you know, I'm happy about our individual impact. I'm happy that if you like, descriptively speaking, if you Googled for the vaccine at any point, like vaccine near me after a certain day, you know, before a certain day, there was no answer. After that day, there was an answer. That answer came from us. a little dissatisfied that that didn't come from people with vastly more ability to have caused that to happen much earlier. But the ultimate takeaway is not about this little tiny piece of the puzzle. It's how can we make the total puzzle better next time.

Dwarkesh Patel: So as you can tell by now, Patio 11 is somebody who loves digging into the weeds of how the financial system actually works. And so it goes to figure that he used to work for this episode's sponsor, Stripe. Stripe is how millions of businesses around the world move money and accept payments. And one of the advantages of Stripe's scale is they can invest in fractional optimizations, which are not viable for individual businesses. but add up to massive revenue and conversion increases. For example, last year, Stripe shaved 300 milliseconds off the render time for its payment links. You can think of Stripe as an entire company full of patio 11s dedicated to making sure they can figure out how to make the payment rails work smoothly for you. And that's why technology companies like Amazon and OpenAI and even traditional companies like Ford and Hertz choose to partner with Stripe. Now, back to my conversation with former Stripe, Patrick McKenzie. Okay, let's talk about some finance. So you write, in addition to saving thousands of lives to vaccinate CA, what you've been doing over the last year or two is writing this finance newsletter, which is very excellent, called Bits About Money. And you explore the plumbing in the financial system. My first question about this, crypto at its peak was worth $3 trillion or something like that. If you buy the crypto skeptic perspective that you have, how do we think about this number? What does it represent? Was it just the redistribution of wealth from savvy people or from dupes to savvy people or to the extent that usual applications didn't come out as three trillion? What does it represent?

Patrick McKenzie: So I think I have two broad perspectives on this. One, people often treat the market cap of something as like implicitly that is some sort of cost on society. And I think the true cost to society of crypto has been that anytime one engages in attempting to do productive enterprise, some actor in society has said, okay, I will stake you with some of society's resources, which these resources are rivalrous. They cannot be applied to any other things society needs in the hope that you will produce something that is worthy of being staked with this. How much have we spent on crypto, not on trading tokens around, but on building infrastructure and spending rival resources that we can't get back, whether that's, you know, GPUs or ASICs or electricity that could have gone to other things in China but went into mining or the time of talented and intelligent people that could have been building other software products but was instead building crypto. That number is in the tens of billions or hundreds of billions of dollars. What do we have to show for that tens of billions or hundreds of billions of dollars? I am very crypto skeptical, and I could give you an answer to that question. I think crypto-like fans would not like to hear it from me, so I prefer Vitalik Buterin's articulation of this question from 2017. He asked, at the time it was $0.5 trillion, a trivial number, only $500 billion in market cap. He said, and I'm paraphrasing a tweet thread, have we truly earned this number? How many of the unbanked have we actually banked? How many distributed applications have a meaningful amount of value doing something which is meaningful? And he has about six other meditations on this. And I think, you know, like, Crypto folks certainly aren't accountable to me. In some manner, you're not even accountable to Buterin, even though he's clearly a leading intellectual in the community. You're accountable to producing positive value in the world. But what is the answer to Buterin in 2024? How many of the unbanked have we truly banked? What is the best use case for crypto right now? Once crypto has a responsive answer for that, that is sized anything like proportionate to the hundreds of billions of dollars, tens of billions or hundreds of billions of dollars of resources that we've staked crypto with, I think crypto people should feel enormously proud of that accomplishment in some future where it hypothetically arrives. And in some future where that hypothetically arrives, you have my sword. I will love your initiative. However, For the last many years, we have been saying, you can still get in early. You can still get in early. You can still get in early because the value has not arrived yet. And so that is my capsule summary on crypto 14 years in. We've staked a group of talented people who are very good at giving a sales pitch with tens of billions or hundreds of billions of dollars. And look at what we have built. this would be a failure in any other tech company. So capital F failure. So either like radically pivot and unfail it, or maybe we should stop continuing to stake you with money.

Dwarkesh Patel: So two potential responses from the crypto optimistic perspective. One, I have people who help me with a podcast who are around the world. Not that many, but the couple I have around the world. I have a clip editor in Argentina. I have a shorts editor in Sri Lanka. And all of these people... have asked me, I haven't prompted them. I asked them, how should I pay you? And they say USDC. And so maybe there it wouldn't be that much harder for them to set up a wise account. But I think it's notable that all of them prefer the just how should I how should I get paid? The first answer is a stable coin. Absolutely.

Patrick McKenzie: That is evidence. And, you know, like some tech savvy people have a good payment rail and they have a payment rail that they did not have access to 15 years ago. but at the cost of tens or hundreds of billions of dollars. Counterfactually, if one had thought, okay, we really want to work on debt payment rails specifically, another way one could hypothetically have deployed $10 billion is on the best-funded lobbying campaign in history in the United States to work on AML and KYC regulation to allow more easy transfers of money worldwide.

Dwarkesh Patel: But why does it have to be compared against the best possible counterfactual use case? It's the sins of commission versus omission again, where it's like, On the margin, it made things better.

Patrick McKenzie: Don't judge it by, like, hypothetical worlds. Just, like, keep in mind that hypothetical worlds might exist. Judge it by, like, the actual realized utility at the moment relative to the amount of resources consumed.

Dwarkesh Patel: The second point is, if you look at, for example, the dot-com bubble, literally close to a trillion dollars were invested in laying out the fiber and the cable for this artifact that now you consider is the most valuable thing that humanity has built. And at the time... Reasonably, a lot of the companies that built this went bust. There was a bubble-like dynamic where many of the investors who spent the capital to build out this infrastructure weren't paid back, didn't see immediate use cases from what they'd built. And the bubble had sort of served as a shelling point to then in the future get things rolling. And that was hundreds of billions of dollars, a trillion dollars, tens of billions of dollars. If in the future something cool comes out of it and it's a useful use case, that's probably worth it, right? Cool. Yeah. At what point do we get to say that didn't happen?

Patrick McKenzie: A trillion dollars? Like, at what date in the future do we judge, like, someone has been right or someone has been not with respect to, we have created, like, you know, people in crypto have very confidently stated in various places that this is the next iteration of the internet. This is, you know, will revolutionize the world, not just how payments are conducted, but it will be, like, a fundamentally new computing architecture. Okay, like... At what day do we compare notes on whether that claim was accurate or not? Does 2030 seem like a reasonable year or too far? It seems like reasonable to me. Can I make a prediction of what is said in 2030? You can still be early. Crypto has created huge amounts of things, but it's not achieved anywhere near its true potential. Please invest in our new crypto startup. And, like, let's check back in 2030, folks. Like, please, you know, tweet me if I'm wrong in 2030. I will happily eat crow. I want to eat crow. Like, you know, crypto people are like, how couldn't you be interested in programmable money? I'm like, I'm interested in programmable money. Obviously. Money is programmable money. But, like, my friends who are trying to sell me on this since, like, 2010 weren't wrong. This should, like, totally smash my interest based on what I usually find, like, intellectually edifying. And I don't not find crypto intellectually edifying. I actually think there are some interesting things that have come out of the movement. But I find a computer built in Minecraft out of redstone to be intellectually edifying. It's a wonderful educational device for people who don't understand how a CPU works. And I'm not proposing to use the Redstone emulated computer in Minecraft to be the next computational infrastructure for the world, because that fairly obviously will not work very well.

Dwarkesh Patel: So another answer of what the value here is, listen, we want some sort of hedge against, and I think this is actually kind of a reasonable argument. I actually don't buy the capabilities that have been unlocked by crypto, but I do buy the argument that we want some kind of hedge against the government going crazy and KYC, AML leading to state surveillance, all the compliance departments in the banks just like start seeing if you've been to political protest and debanking you. And it may seem unlikely, but it's good to have this alternative rail, keeps the system honest, given that there's an alternative. And if things go off the rails, this is a worthwhile investment. I mean, society as a whole can't even count that low in terms of the other resources that it spends. So it's a good hedge against that kind of outcome.

Patrick McKenzie: So I'm actually much more sympathetic to crypto people in this folks than they expect most people who have a traditional financial background to be. I think it is descriptively accurate that the banking system and all companies which are necessarily tightly tied to the banking system which might be all companies well let's come with a central example so that second set first uh are a policy arm of the government and i think uh uh you know whether people articulate that in exactly those words or or not uh varies a little bit but when you have your mandatory compliance training you'll be told in no uncertain terms that you are a policy arm of the government And I feel for crypto folks that say that things like this feels like, you know, warrantless search and seizures of people's information and like very undirected dragnet fashion. And some somewhat complicated thoughts about this. So the modern edifice of know your customer KYC and AML anti-money laundering dates back to the BSA Bank Secrecy Act in the United States, which was late 1970s, early 1980s. And at the time, the Federal government in the United States was like strictly rate limited in how much attention it could give to KYC and AML. And so maybe because we thought we had very limited state capacity at the time, like the government would make rational decisions and maybe it will go after like 10 or 100 enormous white collar crooks and drug trafficking cartels a year and not surveil down to literally everyone in society. But the regulations we wrote and have continued to write and have continued to tighten on over the years do effectively ask for like transaction level surveillance of every transaction that goes through a bank. And so it is the actual practice, and this is not a conspiracy theory. I'm making nothing up, folks. This is like... acknowledged that the actual practice in banks is that they have descriptively about as many intelligence analysts as the American intelligence community has, who get this scrolling feed of alerts that are generated by automated systems. And for each alert, they either go, don't worry, don't worry, don't worry, don't worry, don't worry, don't worry, don't worry, millions of these alerts every day. And then for some tiny percentage, they say, oh, dear. this one might actually have been a problem i'm going to write a two page two to four page memo and file it with the financial crimes enforcement network and in all probability no one is ever actually going to read that memo but we have like an intelligence community sized operation running in banks to write memos that no one ever reads right because some tiny portion of those memos will be useful to law enforcement in the future And if you had explained that trade in like a presidential debate in the 1980s, I find it extremely unlikely that any part of the American polity would say, yeah, yeah, I want to buy that. Could we perhaps spend tens of billions of dollars on it? But we did that. And so to the extent I'm like extremely copacetic with crypto folks on this point, A, like this thing actually exists in the world. I agree with you that it does. B, in an ideal world, I don't think this thing would exist. I think I agree with you. There are very real like, you know, privacy fears. And then, however, crypto has this habit and, you know, people who are good at sales have various sales pitches that they give to various people and crypto will, actors within the crypto ecosystem will talk an excellent game about privacy as long as number goes up. And when you say, ah, but you can choose between like being tied into the banking system, which is necessary for number go up, or you can choose privacy. They will say, excellent. I choose number go up.

Dwarkesh Patel: And so. But there's different protocols. You can use the ones that allow privacy if you care more about privacy. Right. So.

Patrick McKenzie: Yeah. Descriptively, that is a very tiny portion of crypto.

Dwarkesh Patel: Also, to riff on what you were saying about privacy. the analysts they had that number as many as would be an intelligence agency these apparatchiks of were connected to the government's policy just analyzing each transaction as soon as the government gets the competence to run an llm across each of these millions of queries right this is like a legitimate worry because as we

Patrick McKenzie: You know, this is funny in the echoes of, like, we have extremely low state capacity for this thing that we didn't think was important, which was, you know, successfully administering vaccines. But we do have extremely high state capacity with regards to running the security state. And if they successfully manage to, like, get their technological ducks in order, which... There are pluses and minuses there, but they have built some things that are extremely impressive technologically. And then just run it on this data set that we've passively been producing. The sort of implicit ongoing invasion of privacy is much worse than we kind of baked into the system in 1980 when it would have been people going down to archives to look at things in microfiche to try to do this.

Dwarkesh Patel: And I mean, I'm not even necessarily making a point over crypto here, but I think it's worth... meditating on the fact that the default path for this technology is that a very smart llm is going to be looking at every single transaction that is electronically aka every single transaction ever um and it's it's intelligent enough to sort of understand the implications how it connects to your other transactions and what's the broader activity you're doing here yep and maybe this is just a broader point

Patrick McKenzie: about step back from crypto and finance for a moment sure uh i think this is this is one of the least understood things about the tech industry where we have this uh uh this like societal level question that is not being addressed directly but it's being addressed by misunderstood proxy questions on Taking as written that the finance industry is a branch of government in a meaningful sense. Should the tech industry also be a branch of government? And we don't ask that question directly. We have asked instead things like, should the tech industry be responsible for minimizing the spread of misinformation, et cetera, et cetera. And there was an injunction issued in a court case last year on the 4th of July, which I find oddly aesthetically motivating. The court case is Missouri et al versus Biden et al. But the argument made in the court case, which the judge accepted, which is extremely well supported by the record in front of him. is that various actors within the United States government puppeted the tech companies and used them as cat's paws to do, frankly, shocking violations of constitutionally protected freedoms like the freedom of speech. And that there were like, not on the level of we've built this, you know, unaccountable, hard-to-inspect system of LLMs and heuristics. And we started turning off a lot of people's feeds on Facebook. But no, there was an individual person in the White House who was sending out emails like, when are you going to address me on this tweet, guys? We can't have things like this anymore, et cetera, et cetera. Again, a feature of the United States. We are very good about keeping records and transparency and having a functioning legal system. And the record before the court is like... I was following along as this was happening and what was happening was much worse than I understood to be happening. An example of something like, as we were growing up as children, would you ever think that the United States federal government would tell, I believe it was the state of Missouri, Hey, you have town halls where like citizens can come in and speak their mind and advocate for their policy preferences. And you probably have a civics class and talk about like the first amendment and things. Yeah. Someone said something we don't like in a particular town hall. Take down the recording from YouTube. That happened. That is a violation of the constitution of the United States. That is against everything in the traditions and laws and culture of the United States. That is outrageous. And yet it happened. Yeah. And we have not repudiated the notion of using tech as cat's paws or using, in some cases, this is like literally written in the decision, by the way, which I would urge everyone to read. There was an individual in a non-governmental organization which was collaborating with the governmental organizations in doing this, which said, to get around, I'm not quoting exactly, to To get around legal uncertainty, comma, including very real First Amendment concerns, comma, in our ability to do this, rather than doing it in the government directly, we are outsourcing it to a bunch of college students who we have hired under the auspices of this program. What? Like, one, just as a dangerous professional here, you've violated the character from The Wire, Stringer Bell. Stringer Bell's dictum on the wisdom of taking notes on a criminal conspiracy. But you literally wrote that down in an email. The outrageous part is not that you wrote that down in an email. The outrageous part is you, with full knowledge of it, engaged in something that is outrageously unconstitutional, immoral, illegal, and evil to the applause of people in your social circle. And everyone involved in the story thinks that they're the good guy in it. If you write that email, you are not the good guy in the story.

Dwarkesh Patel: Okay. So on that, by the way, so what is your sense of what the judicial end result of deliberations on this will be?

Patrick McKenzie: I think there will be a limited hangout and walk back of some particular things. And do I predict that? And probably, you know, there does exist an injunction. I predict that that continues to happen in the future. Yeah. What do I know? I'm just a software guy, but do people want to achieve power? The tech industry descriptively has power because it is good at achieving results in the physical world. This is certainly not going to be the last time that someone desiring power thinks like, okay, can I force you to give me the power that you have accumulated? And like that is, you know, ultimately like this is fundamentally political decision about how we construct our democracy and we should make good decisions about that.

Dwarkesh Patel: Well, I mean, maybe that's the crux here, which is that through the story you illustrated of the Vaccinate CA and the lack of accountability that showed in our institutions, The idea that you're going to go back to Congress and get them to pass some law that says, oh, we're KYC AML. We realize that without alums, it's going to be a bad deal with regards to privacy. We're going to roll that back. We don't like that the collusion between tech and whoever's in power and being able to dictate what can get taken off the platform. And we think about that as free speech. We're going to pass laws and take that back. There's no sense in which society comes to a consensus and here's the privacy concerns we have. Here's the free speech concerns we have. So at least one argument goes, the solution has to be that you just go off a rail, you start a new rails for these kinds of things that cannot be constrained in this way. And it's not a matter of just changing the KYC law or that's just not that's implausible given the manifest decline in state capacity we've talked about.

Patrick McKenzie: Yeah, I don't accept that that is the only thing possible for us. I don't accept that the United States is incapable of doing nice things. We can't accept that. We have to be optimistic about the future. Otherwise, what are we doing here? And in the tech industry, we know it is not like a physical law of reality or of large institutions that one cannot make systems that work. Making systems that work is the job. We have a few existence proofs. We should increase our engagement with government on, you know, like, hey, state capacity, we can help you build some of that stuff. Also, you know, Constitution of the United States is a document. We feel like kind of attached to that document. We understand, like... Again, incentives rule everything around us. Tech industry was in early 2021 very concerned about being told in no uncertain way by people in power, if you embarrass us, we'll end you. And the one thing in the judicial record of this case is that the White House routinely, White House and other government actors, routinely overreached and asked the tech companies, we would like you to censor us. this and this and this and this. And the tech company said no in a bunch of cases. And so like continuing to negotiate for the right outcome rather than the one that people in power merely want is important. there are some things that will feel unfortunate and maybe a little bit outside of the, like our true sweet spot of what we would want to be doing on Tuesday in the tech industry, where like, maybe we have to ratchet up the amount of public policy advocacy that we do. Lobbying is a dirty word in the tech industry. It probably shouldn't be. Maybe we, you know, not just lobbying, but like, when the do not embarrass us order came down and people were getting very quiet about the fact of feeling constrained by this. Maybe we should have spoken out more and spoken more boldly about it. Maybe when it was like the routine case that the White House was telling Facebook, Twitter, et cetera, et cetera, everyone knows the names. companies like on a tweet by tweet communication by communication basis and also with regards to broad rules that affected the entire of the citizenry of the United States and residents of the United States and also everyone else in the world because these are the operating systems of the world like giving direct orders on there's a certain kind of free speech rather there's a certain kind of speech act which we find vexatious and we would like you to stop that everywhere like very plausibly should get on the nightly news and say i received the following email from the white house that says uh we should uh uh stop this everywhere uh like if you point a gun at me i will comply with this at the point of a gun that is

Dwarkesh Patel: what it will take but this is it almost requires civil disobedience in the sense of if you're right about the earlier statement that in 2021 there are going to be serious political repercussions on the tech companies so Okay, so say that's right. And then they published this email, like, here's what the White House just sent me to take on this tweet. And now Twitter's market cap has just collapsed because people realize the political implications of why Jack Dorsey just got up and said at the time. So then the solution is that you need tech companies to basically sacrifice their capacity to do business in order to... I mean, maybe that is a solution, but that's not a story about optimism about the ability of the US government to solve problems this way.

Patrick McKenzie: You need to have a risk tolerance, right? Every business everywhere, including the financial industry, including the tech industry, is balancing various risks. And the risk tolerance was poorly calibrated. One can achieve results in the world by doing things like embarrassing government officials. And embarrassing a person in a position of authority is not a zero risk behavior. It is relatively low risk in the United States relative to other places. That was extremely important for VaccinateCA. People thought at the beginning, are we going to get in trouble for publishing true information about vaccine availability that will embarrass the state of California? And I said, I have a very high confidence that no matter what we do vis-a-vis the state of California, that you cannot get in serious trouble in the United States for saying true things. you know, the First Amendment exists. We have like backstopping infrastructure here. And if push comes to shove, we will shove back and we will win. And like, this is just me as a guy who like took the same civics course that everyone took and does not have like a huge amount of resources relative to say like the entire tech industry. Like maybe we, you know, need to have a certain amount of intestinal fortitude, like a, okay, You know, you've asked us to do something. You've threatened us with taking away all the wonderful toys and the great business models that make this an extremely lucrative area to work in and to sacrifice a value that is very important for us to continue to do that. No, we're going to fight you on this one. And, you know, the comms trained part of me is like, don't use the word fight. We are going to collaborate with stakeholders across civil society to achieve an optimal outcome balancing the multiple disparate and legitimate interests of various arms of the government and civil society and blah, blah, blah, blah, blah. Sometimes that requires fighting.

Dwarkesh Patel: We should fight when it does. On Tyler's podcast, you said something like, America doesn't have the will to have nice things, and Japan does. But if you think about... In some ways, yes. If you think about your own essay about working as a salaryman, you're working 70-hour weeks, and you're killing yourself to get that marginal adornment on the products you're making... isn't it more efficient isn't it better that we have a system where we put in 20 of the work to get 80 of the results we spend the rest of the effort on i don't know expanding the production possibilities frontier and it's it's good that we don't have the will to have nice things we just get it done i don't think these trade off

Patrick McKenzie: against each other at the relevant margins and i you know nothing about culture is monocle sensational um also i don't think culture is a uh sufficient explanation for some of the differences that are achieved in the United States versus Japan, for example. There's a great book, Making Common Sense of Japan, by a person whose name I'm blanking on at the moment. An argument he makes in it persuasively and at length, which I don't think is 100% true, but is more true than most people in well-informed on either side of the Pacific belief, is that when people say they do this because it is Japanese culture, what they're often saying is, you know, I usually have a model for why people do things driven by incentives. And I understand this incentive, this incentive, this incentive. And then there's some error term in this equation that I don't understand. I'm going to call that error term culture. I think culture is a real thing in the world, to be clear. But I often think that we reach for that error term far faster than we should. So as my minor observation about culture, with respect to... You know, there are places and pockets of the United States that have the will to have nice things. And often they discover, sometimes surprisingly, that the only thing you need to do nice things at the relevant margin without spending more money, without having people like kill themselves over 90 hour weeks for the entirety of the career, you can just choose to have nice things. Let's choose to have nice things. Let's not be embarrassed about choosing to have nice things.

Dwarkesh Patel: So you understand all this financial plumbing. If you were an investigative reporter, what is the thing you're looking at that the average reporter at some newspaper wouldn't know to look at to investigate a person or a company or a government institution?

Patrick McKenzie: I have an enthusiasm for the minutiae of banking procedure in a way that few people have enthusiasm for the minutiae. Sometimes banking procedure causes like physically observable facts to emanate into the world. And if you know that those facts are going to emanate, then like you can have a claim made about a past state of the world. Like I did this thing or I did not do this thing. And that claim will, if it is true, cause metadata in other places. And you can look for the metadata. This is actually how a lot of frauds are discovered because the fraud is like basically the definition of fraud is you're telling someone a story. The story alleges the fact about the world. The story is not true. And you're using the story to extract value from them. Most frauds will allege facts about the physical world. As the physical world gets more and more mediated by computers, as it gets increasingly sharded between different institutions, there will often be institutions who are not under control of the fraudster, who have information available to them, which will very dispositively answer the question of whether the alleged fact happened or not. And as a reporter, understanding how institutions and society interact with each other and the physical reality of, okay, if this thing happens as alleged, then these papers will be filed, then these API calls will be made, et cetera, et cetera. And then doing the core job of reporting and finding people at the institutions who will tell you the truth. As an example of this, Mt. Gox many years ago was insolvent, and that fact was widely rumored but not reported, presumably because the global financial news industry didn't find it convenient to have someone call into the Japanese banking system and ask the right questions in the right way. The CEO of Mt. Gox alleged on Bitcoin Talk that the reason that they were not able to make outgoing wires was because they had caused a distributed denial of service attack on their bank's ability to send foreign currency wires. That bank was Mizuho. Mizuho was the second largest bank in Japan. Many people at, say, well-regarded financial reporting institutions in New York City find it incredibly exotic and difficult and maybe in some ways kind of unknowable to extract fax from Misehole, which there are addresses. FedEx will deliver letters to them. They have phone lines. We also have fax machines. We love our fax machines. Like, could you send a fax to anybody at Misehole and say, hey, quick question. Are you sending, like, wires today? And Mizuho would, like, receive the fax, look at it kind of quizzically, and say, in response to your fax earlier, yes, we are still sending wires because we are the second largest bank in Japan. Do you have any other easy to answer questions for us? You know, like financial reporting, drop the ball on just asking like individuals, let me use a whole simple questions about reality. Maybe you should do that next time. And like, to the extent that you understand, like, okay, like one, understand that the CEO is giving out gold on Bitcoin talk under his own name, where these are like obviously reportable statements, the statements and alleged facts about material reality, and maybe like chase down the truth value of that. That's hard. It's so much easier to just repeat what he says on Twitter and say, as said by this person on Twitter, and then quote the Bitcoin price feed. But reporting is hard. Be good at it.

Dwarkesh Patel: Why aren't short sellers doing this? Because they should have an economic incentive to dig to the bottom of this, right? And so we should have a deluge of financial information from short sellers who call the banks and trace through the API calls.

Patrick McKenzie: Yeah, that is an ongoing interesting question. I think short sellers provide an enormous service for the world in like being essentially society's best solution to financial fraud. And yet they fail to detect lots of them. And not just throwing short sellers or reporters or anybody else under the bus. I failed to detect SBF's various craziness despite having sufficient information available to me as a well-read person on the internet to have detected that. Where were the freaking wallets? Everybody assumed someone else was looking at it, essentially. So that's one reason. Short sellers often assume, okay, I need to find a like first get put on the path of something and have a differentiated point of view. And then like another issue for short sellers is you have to find an instrument and you have to find another side of the trade to successfully do that. And there was, without being expert on Bitcoin micro mechanics, it was like difficult in size to make the trade. Mt. Gox is insolvent right now, other than like pull money out of Mt. Gox, which people were definitely trying to do. I got a number of interesting business proposals. In the 2012 ish timeframe from people who said, Hey, you're an American and you clearly understand like international banking and you live in Japan. Could I like cause you to get some yen and have you wire that to me in America and you can take a percentage. And I said, I really don't like where this is going. And they said, well, there's this company and I've got some money over there and they can send yen, but they can't send dollars. And I'm like, is that because they don't actually have the money? And they're like, no, no, no. It's a Japanese banking thing. I'm like, No, it's not. Japanese banks are very good at sending wires. And they said, no, no, no. It's really this thing. This is totally clean. I'm like, you would not be having this conversation with me if it was totally clean. You need a money launcher.

Dwarkesh Patel: I will not be your money launcher. How hard is money laundering? If on one hand, you mentioned earlier the state capacity that banks have where every transaction is analyzed and flagged. And if it's notable enough, they write a report about it. How sophisticated does the cartel need to be in order to move around seven figure amounts of money, let's say?

Patrick McKenzie: So the definition of money laundering is like extremely stretchy and like taffy. And there's a spectrum of people, much like there's a spectrum of sophistication and financial fraud. There's a spectrum of sophistication and money laundering. So if you want to look at probably the most sophisticated money laundered in history, he's currently a guest of the U.S. government, but wherever SBF is staying. He was sophisticated? Oh, this is a disagreement I have with a lot of people. SBF was extremely sophisticated. Like a person, not just SBF. I think people identify him uniquely, right? They identify the inner circle uniquely as being at fault here. There was an entire power structure there, which was extremely adept at figuring out how power worked in the United States and exercising it towards their own ends. And then it blew up. But until then, goodness, like, you know... they decided we need regulatory licenses. They're called money transmission licenses in the United States. And those are done on a state-by-state basis. They got 50 regulators to sign off on it, et cetera, et cetera. Like there were many objective indicia of them being very good at their jobs until

Dwarkesh Patel: they lost all the money. But it was more about politically getting people to look the other way rather than we figured out how to structure the wire in a way that won't get flagged.

Patrick McKenzie: It's not merely a matter of getting them to look the other way. But if you go back to the original SBF interviews where he's telling the founding myth of Alameda, he says like very loudly, you know, the reason why I got this opportunity to do Bitcoin arbitrage between Japan and the United States is because I was able to do something that the rest of the world wasn't. I was able to, he doesn't say this in this many words, I will say it, suborn a Japanese bank. because you need that as one of the pieces to run this arb. And then they pulled tens of millions of dollars out of this. I don't think people really listened to what he was saying there. And he literally says in the interview on Bloomberg, if I was a compliance person, this would look like the sketchiest thing in the world. This looks like it's obviously money laundering, because it is money laundering. money laundering. And then, interestingly, Michael Lewis retells the story, and he locates the story in South Korea rather than in Japan. And some people who were involved say, we tried it in South Korea and Japan, which people would like pull on more threads there. Like, there's still lots of that story that we don't know. Anyhow, how sophisticated do you have to be to launder tens of billions of dollars around? SBF did that. So that is Abar for sophistication. He was eventually caught. He was not caught for laundering tens of billions of dollars around. He wasn't even under suspicion for laundering tens of billions of dollars around. SBF was Tether's banker. Alameda Research, one of the parts of the corporate shell game that they were playing, moved tens of billions of dollars of cash around the financial system, largely under full color of law. on behalf of Tether to move it from wherever Tether had it, goodness knows, or wherever their buying customers had it to, I think at the moment it's mostly at Cantor Fitzgerald. Some shoe has to eventually drop there. I will eat a lot of popcorn when it does, but be that as it may. There's many other ways to launder money. You can do things like, let's see, I... I establish a shell corporation and I buy a piece of real estate in New York City. And then I rent that real estate out to people and I collect a stream of rents from that. And that money looks clean because there is an executable business. It's my shell corporation that is renting this real estate that really exists to a totally legitimate person. This money is clean. And the money that I put into the system to buy this on behalf of the Shell Corporation, I'm just going to wire it to a lawyer. And the lawyer is going to answer any question from the bank with, I don't know where it came from. I don't have to tell you. I'm a lawyer. It's a real estate transaction. What do you want from me? In one sense, that's money laundering. If the original money was the proceeds of crime. In another sense, that's how every real estate transaction goes down at those scales. And so often a facility at money laundering is one facility at operating the economy plus money. you know, willingness to do that to hide the proceeds of some other crime. I think I would be really good at money laundering. I'm glad I haven't done it professionally, but it's fascinating intellectually. Previous communications departments I've worked at probably would explicitly anti-endorse that sentence. What can one do?

Dwarkesh Patel: We are reasonably confident that you're not laundering money. I would be much wealthier if I was. Well, a separate topic, but when you look, you've emphasized that people tend to undercharge for the products they serve. If you have identified somebody who actually does charge for products where they can get away with, what psychologically do they have that the rest of us don't?

Patrick McKenzie: So I think interestingly, here is one of those places where culture is not necessarily merely the R term, but is actually descriptive in some ways. Without pointing fingers at particular examples, because it gets contentious, there are some cultures in the world that institutionally have adopted more of a how do you put it, pro-capitalist, pro-mercantilist, et cetera, like less of an ingrained skepticism regarding like earning money and accumulating resources as a goal. And there's other cultures which have an extreme ingrained skepticism about earning money and accumulating resources as a plausible goal. And those cultures generate people who have very different negotiation strategies. And when you, like, impact people with different negotiation strategies against, like, the reality of a well-operated, I don't know, Google, for example, they arrive at very different numbers. What's the, is it Amy Chua who wrote Market Dominant Minorities, a piece about, a book about that subject, etc.? ? One of the things in that is that, you know, all people are equal in the eyes of God and hopefully in the eyes of the law, but, uh, like, uh, not all cultures like physically make the same decisions with regards to the same, same facts on the ground. And that causes some disparity outcomes. Uh, and so that, that is one tiny part of that thesis, which I don't, uh, I've read a lot of books. I don't necessarily endorse every word in every book that I read, but, uh, I think there is something to that. I think another thing is that, uh, uh, you know, there's a certain personality type cluster, I guess you would say, of people that got into tech. And many of us, it is over-advanced that the tech industry and the pathologies of the tech industry are caused by the nerd versus jock distinction in American high schools. Heavily over-advanced. Amount of truth to that? Not zero. And like many of us were, you know, We came up, we feel we were largely getting beaten down by the system around us. We were not worthy, yada, yada, yada. And then we have carry those issues into our professional lives. And some people work their way out of it quickly. Some do not. Some people work and, you know, they, for class and etc reasons go to institutions like stanford and hear from i don't know who you hear things from if you go to stanford because i certainly didn't go there but i don't know an elder fraternity brother that says yo bro this is the way the world works you really got to negotiate when you get in the offer with when you get in your discussion with google i've talked to so many brothers and they don't negotiate and so the ones that do make a whole lot more money you're like wow good for that most people don't like get that talk from their elder fraternity brother because they do not go to stanford or don't have an elder fraternity brother and so like Until VaccinateCA, the most important thing I'd probably done professionally was writing a piece on the internet about salary negotiation. I think it's subtitled, Make More Money, Be More Valued, which is just an exhortation for descriptively mostly young people who had some of the issues I had when I was young and growing up. Like, hey, you're allowed to negotiate. That's not a moral failing if you have no less right to the marginal dollar than a company has to the marginal dollar. Like, Go get it. And then you can put it towards all sorts of interesting ends. 500,000 people a year read that piece, and it's now 12 plus years old. I keep a folder in Gmail about who has written and said, I got $25,000 to $100,000 per year more as a result of reading the piece. I used to keep a spreadsheet. I stopped keeping the spreadsheet after it ticked into the eight figures. I'm like, keeping the spreadsheet was just an ongoing source of stress for me. Eight figures? Yeah, per year. Like you would assume that 500,000 people read it per year. Some take the advice. Most who take the advice probably don't write me to say, hey, I took the advice. Thank you. And maybe I missed some emails, yada, yada. So like the true economic impact is probably larger than that. There are probably people who have like the inverse of that spreadsheet where it's like, darn it. We got quoted a petty 11 against us again. but like because you know we got these numbers and there's only like a few firms in the tech industry that do hire scaling it you know scale hiring anyhow okay um

Dwarkesh Patel: There's less people today who are in their twenties who had prominent software businesses than maybe 10 or 20 years ago. Having been in the software industry last 10, I don't know how, when you started exactly. Is your sense that this is because the nature of software businesses has changed or is it because the 20 year olds today are just less good?

Patrick McKenzie: I've met many young and talented people over the course of 20 years in the software industry and young and talented people continue being young and talented. I think one thing that is partially explanatory is that when there's a new frontier that opens up, the existing incumbents, both in terms of institutions and in terms of people with deep professional networks and personal resources and similar, do not immediately grab all the value in that new thing. It's Terra Nova. And so to the extent that tech is no longer Terra Nova, I think you would expect less people who are less resourced, who are younger, et cetera, to rise to the heights of prominence in tech. To be clear, I'm not at heights of prominence in tech. And when I ran companies, I was not running companies like some other guest's podcast run companies. It was bingo card creator. I was making bingo cards for elementary school teachers while living next to a rice paddy in central Japan. But that's kind of like my dominant hypothesis. There are, you know, some things that are affecting the youth that I think are negative. I think some products that the tech industry has created do not maximize for the happiness or productivity of people that consume those products, TikTok, et cetera. But, you know, I continue to be bullish about the youth and I have two children who knock on wood will accomplish things in their lives. And so I'm intrinsically skeptical about, oh, the kids these days, they're just bad kids.

Dwarkesh Patel: How much do you worry about video games as a sort of wireheading that somebody like you, but 20, 30 years ago, or 20, sorry, now has access to Factorio and will just wirehead themselves to that instead of making a really cool software product. How much should we see this in the productivity numbers? Oh, goodness.

Patrick McKenzie: I don't know about the productivity numbers generally. I do know that Steam keeps a counter of how much I am playing video games in a given time. And the knock-on would have accomplished a few things in my career. And also, against that, what was my Steam counter up to? So Steam didn't include World of Warcraft. World of Warcraft was at least a thousand hours. For me, Factorio recently was 750. And then if you like some over 20 years, I've probably played video games for four thousand six thousand hours and so that that's you know two to three years of professional effort if one thinks that it trades off directly with professional

Dwarkesh Patel: effort uh but do you because if you then include every single young guy who's um a nerd and often they you know how much we worry that they're a bunch of their productive time is going to video games instead of making the next software business I worry at least a little bit about it for myself.

Patrick McKenzie: So I recently started working for an executive assistant, and one of the first suggestions that he gave was, hey, Patrick, will you friend me on Steam so that I can see how much you're playing in a given week? And so if you're not making your priorities happen, we can have an honest discussion about priorities. That's really good advice, given that I spent far too much time rat-holing on Factorio relative to my true preferences. You've got a really confident DA. Go, Sammy, go.

Dwarkesh Patel: Hey, boss, can we have an honest conversation tonight?

Patrick McKenzie: I don't know if I'm inserting those words into his mouth, but the suggestion was genuinely his. And he heard, you know, on the one hand, Factorio, wonderful game. I actually think Factorio matters far more to the world than most video games do. That's an entirely different piece that I'm trying to write at the moment. Be that as it may. Have you read Byrne's piece on this? The Factorio mindset? Yes. I love many things Byrne writes. I think I luckily have a differentiated point of view on this one, so I will hope to get it out to the internet someday. But on the one hand, I loved it. I think Factorio Space Exploration mod specifically was the best video game I ever played. And yet I spent 750 hours on that over the course of a year. And I was, you know, on sabbatical and recharging from six very hard charging years at Stripe and also running the United States vaccination location information effort. But, you know, There's a question of like at relevant margins, are you maximizing for your true values? And I was like, oh, a little worried about that. And so now my EA checks on me. Do I worry about it for other people? I will say that when I was young and a World of Warcraft raid guild leader who spent a thousand hours on that, that it was a substitute advancement ladder for me. The actual job was working sailor man in central Japan. It gave me no scope of control over things. And I thought if I was a startup CEO, I could make decisions right now. I could build something awesome, but I don't have that ability. And I don't see myself as being the kind of person who could become a startup CEO. So if I can't be satisfied in my nine to five in terms of making things happen in the world. At the very least, I can make sure we kill the dragon in two hours. Don't stand in the fire. Come on. Team leaders, you need to make sure that people are equipped with the resist gear before they get there. Oh, we're having internal spats about allocation of resources. We need to have a better DKP system. By the way, World of Warcraft raid guilds and all other places where intellectual effort comes together in the video game community are much more sophisticated than people give them credit for. When I started VaccinateCA, I told people my sole prior leadership experience is having 60 direct reports in a raid guild, and that's true. And also, like, we don't rat hole on the subject but like there are parts of vaccinate ca that are very very definitely downstream of the uh like intellectual efforts about managing raid guilds specifically there are multiple people internally who are like yep like we are running the raid guild playbook right now um but um uh be that as it may uh like you can do something with your life like choose to do something something with your life and then if you want to play you know a reasonable amount amount of video games then play a reasonable amount of video games And then, like, with respect to individual people, I sometimes worry that you kind of get into... I think I've struggled with depression at some points in my life. Many people struggle with underdiagnosed, under-treated depression. I think sometimes you get into a self-destructed spiral measured against, like, your true values and similar values and preferences where... Due to depression and other factors, you aren't making as much progress on the true goals you do. You use video games as an escape from that, and not just video games, books, television, etc., etc. There are many poisons available in life. You pick your poison, use that to escape. The amount of effort you put into the poison causes you to have less effort available to do the thing. So you get less good results than thing. So helping people out of those self-destructive spirals is something that we as a society could stand to get much better at.

Dwarkesh Patel: Speaking of Bird, by the way, I noticed that a bunch of my favorite writers are finance writers. There's you, Byrne, Matt Levine. Is there some reason why finance has hogged all the writing talent? I think there are many good writers in the world.

Patrick McKenzie: Derek Thompson, for example, has written some... He's a chemical engineer and he's written some things which I barely understand. I have enough of an engineering degree to appreciate half of the chemistry and I can't appreciate the full totality of why... uranium hexafluoride or whatever is a terrible substance to work with and he has

Dwarkesh Patel: some excellent writing on why that is a terrible substance but the broader question being why does finance have a greater concentration of writing talent and not just about current bloggers but you know finance histories are some of the best history books out there the bethany mclean's and so forth and i'm curious why this is is it just an intrinsically more interesting subject or

Patrick McKenzie: There's some path dependence. If I had ended up working in a water treatment plant, I'd be writing about water treatment plants because I like writing. And I am positive. I know enough about myself to know that a discussion about how alum works in water treatment plants, which is something I read when I was like six, that could totally captivate me for multiple years on end. And I would write about that if I was captivated by it. I agree with the point Matt Levine made once, which is that finance and the tech industry have for a while been a relatively reliable way to turn intelligence into money. Many good writers are very intelligent. Not all people who are very intelligent are good writers. I think it is a skill that more intelligent people could learn. But simply if we create an incentive system, which will tend to allocate not yours truly, but descriptively, like a lot of the country's top brains into like particular fields where they will become experts at those particular fields. I would expect also a lot of the writing talent to be there because good writing is good thinking. That's a Paul Graham quote, I think, but that is broadly true.

Dwarkesh Patel: Speaking of which, so you went to Japan because, if I remember this correctly, you thought that after the dot-com boom that the programmer alone demand will diminish and it will require some combination of skills. Separate from, and you weren't the only one, there were many other people who thought this way. The Wall Street Journal said it and the Wall Street Journal had never been wrong in

Patrick McKenzie: my experience as a 19-year-old who didn't know anything about anything.

Dwarkesh Patel: But separate from the object level predictions that you might have gotten wrong about the software industry, at a meta level, what was a mistake you made? How would we characterize that? Sure.

Patrick McKenzie: So I had a whole lot of rigor chasing a decision that had no basis in fact. So believing in the Wall Street Journal that all future engineers would be hired in places like India and China and not the United States. And so there would be no future engineering employment in the United States. I made a spreadsheet of like, okay, here are the languages my university teaches. Here's my best estimate of the number of people in that country who speak it, Americans who speak it, the amount of their software that gets sold here, amount of US software that gets sold there, blah, blah, blah, blah, multiply these together, sort by column H descending. And this is like LARPing and having rigor here, but it felt like a good decision-making process to me in the time. Now, the meta discussion is, like, don't LARP it having rigor. Actually have rigor. But what would that look like in this context? What would you have done differently? Ultimately happy with the decision I made, although, like, as presented, it's the wrong decision. I'm happy about it for other factors about life. Working in Japan as a young engineer, there are some very rough parts about that. But, like, on a meta level... Maybe this is an early opportunity to trust institutions less and trust systemically viable reasoning patterns more. The Wall Street Journal has asserted that no future Americans will get hired at software companies. Assume this is true. What happens in the American software industry? Maybe I did not have enough knowledge to confidently predict that at the moment, but I can confidently predict some things now. Software companies are going to start to break as older engineers age out of the engineering population and they have no one coming up to replace them, etc. The industry institutionally must hire people every year, and hiring freezes are necessarily temporary as a result of that. That's as close to the law of physics as one can have. So if you are telling me something which says the law of physics has been suspended and will be in the future, I don't agree to that. How would I learn the law of physics? I did not know anyone in software engineering at the time, which is partly why I was getting my advice from this from the Wall Street Journal. But if I had been slightly more agentic about it, I was at a research university in the United States.

Dwarkesh Patel: I could have found someone who knew someone who was in the software industry to explain this to me. Couldn't you use that same logic to say that the journalism jobs won't go down because senior journalists will have to be replaced by new journalists? And that's true, but that doesn't take that many people to study journalism. And it actually would have been a bad call to major in journalism and pursue that as a career.

Patrick McKenzie: But the fundamental thesis for journalism is that the total size of the pie is decreasing in journalism due to structural factors. And the Wall Street Journal's thesis was not simply that the size of the pie would decrease in the wake of the dot-com bust, which I don't think they even got as far as articulating that, although it's been 25 years, but was more like, oh, you know, companies will maximize for cost of labor, et cetera, and therefore ship out all the jobs. I see.

Dwarkesh Patel: I see. Yeah. You've emphasized that founders should do more A-B testing. That's like one of the main themes of your blog, if you go back through it. So they're under-optimizing on this. What do they tend to over-optimize on?

Patrick McKenzie: So interestingly, I was the marketing engineer earlier in my career and really thought that that was important. And in terms of like high level advice I would give a founder, I would probably tell them a little bit about marketing engineering, but wouldn't spend 95% of my time talking about that unless that was explicitly what they brought me in on. Founders spend too much time on playing house and chasing status are both two sort of like well-known pitfalls where you can get addicted, addicted is not quite the right word, Your incentives will draw you into playing the role of the CEO of a successful company before your actions have earned the company its level of success. The fundamental nature of early stage businesses is that investment in you is not a vindication of what you have achieved. It's an advance on your future accomplishments. And you need to rigorously pursue actually making those future accomplishments. And there's many ways to rigorously pursue it. Talk to more users, write more software, make something excellent, get more people to use it, get better at selling it, etc, etc. That's an important strain of Silicon Valley culture. I'm glad we have it. I'm glad we are popularizing it to as many places in the world, including to me in central Japan. And yet, you know, there are always other games that are going on, and those other games are less important, but they are very, very attractive games. Not video games here, but why don't you go to a conference? Why don't you meet more interesting people? Why don't you show up at the best parties, etc., etc. ? Showing up at the best parties does not, at most margins, increase the number of users you talk to. It doesn't write functioning code. Do those things that actually matter. And some distractions proudly wave, I am a distraction from everything that matters. And some don't. Some definitely say, I'm real work. I definitely feel like real work. And they're just not. And so don't do the things that don't matter. Which sounds vacuous, and yet...

Dwarkesh Patel: Okay, I want to go back to VaccinateCAA and I want to close the loop on one question that I had that I don't know if we got a good answer to that I think is important, which was, suppose you're the president of the United States, maybe you're a new one. So this one's been replaced and you're looking back on what happened. If you personally were president, what is it that you're doing to bring accountability to what happened and making sure that in a future crisis, which might look different than COVID, things are ready to snuff? Maybe you like fire the right people, but beyond that, there's a lot of different things that could go wrong. How do you make sure that we're ready for them?

Patrick McKenzie: One cultural practice of the tech industry that I think it would be more salubrious for the broader civil society to adopt is the concept of blameless postmortems. We talked earlier about who to blame for various failures, etc. I broadly believe that there is some amount of blame which performs useful societal purposes. Uh, and then beyond that, like the diminishing returns set in pretty quickly. And we are, um, you know, the magic word in Washington is accountability. We want accountability for failures, et cetera, et cetera. People are terrified of accountability. Uh, and, uh, that, uh, causes there to be fields of distortion around things that actually happened, mistakes that were actually made, opportunities that were not pursued, um, and similar. And so, uh, uh, changing our general practice about how we accomplished accountability towards the direction of, okay, first, let's get a dispassionate record of what actually happened. It's less important that it was a, you know, this named official. It's less important that it was under this legislative authority, et cetera, et cetera. What did we do? Okay. Now, how did what we do lead to the outcome that we got? Given that what we did led to this outcome, what could we have done better? Okay. Given that there are these things that we could do better, how do we inject that back into the currently running system such that the next time this happens, we don't do the mistake again? Sometimes that will involve someone losing their employment, although hopefully not that frequently. a non-zero amount that is important to say but there's many things that we would post-mortem of that of the experience I don't know if we have like several thousand hours of time to go over all of them you know there should be an inquiry it's an easy thing to say like Let's at a minimum like ask all involved parties, hey, can you write down the history of the COVID experience? Just passionately like record dates, times, actions taken. We want it to be truthful and comprehensive and highlight what you think is the most important part about this.

Speaker 3: Mm-hmm.

Patrick McKenzie: Like that is step one. Like maybe we ask for step one and then try to get to step two. Yeah. Final question. What are you going to work on next? I don't exactly know what will be my next big professional splash in. I've been on semi-sabbatical this year. I've been writing bits about money, but bouncing between like 20 and optimistically 80% productive relative to what like 100% productive looks like to me. I might do a software company next. I might raise a small VC fund, not entirely decided. I might do something entirely different at the moment, kind of focusing on family-oriented things and our family immigrated from Japan to the United States and are going through uh you know all the fun adjustment issues and i'm partly i focused very much on career and other things over the course of the last eight years and i'm you know rebalancing that to help them on this adjustment and then figure out whatever happens next for the next chapter of life excellent patrick thanks so much for

Dwarkesh Patel: coming on the podcast thanks very much for having me Hey, everybody. I hope you enjoyed that episode with Patrick McKenzie. As always, if you did, the most helpful thing you can do is just share it, send it to your friends, group chats, Twitter, wherever else. It's also helpful if you can leave reviews and ratings on Apple Podcasts, Spotify, YouTube, wherever else you listen. And of course, as you can tell, I'm doing ads on the podcast now. So if you're interested in advertising, go to dwarkeshpatel.com slash advertise. Other than that, I'll see you on the next one. Cheers.