Adam:Today I’m speaking once again with Patrick McKenzie. This time I’ll be interviewing him rather than the other way around. Patrick’s also known as patio11 on the internet, a prolific writer about institutions and finance, an angel investor, senior advisor at Stripe, and recently (or several years ago), the CEO of the extraordinary VaccinateCA effort, which provided information about vaccine availability to millions of Americans, a task which, as we’ll discuss, the US Government proved itself incapable of. Patrick, lovely to be talking to you again.Patrick:Thanks very much, and I’ll make the obligatory disclaimer that your audience is probably very familiar with. All of my views are my own and not necessarily the views of other institutions I’ve been affiliated with, such as Stripe or Call the Shots, which ran the Vaccinate CA initiative. And they might also not be representative of other board members of the organization, employees, funders, and similar, or our government or tech industry partners. That disclaimer out of the way, let’s go to the juicy stuff. Legibility and Legitimacy Adam:Great. So just before we started recording, we were talking about the phenomenon that is LinkedIn and copy-pasting knowledge, primarily from Twitter, about issues like AI into LinkedIn. These influencers establish credibility amongst an institutional audience that maybe isn’t plugged into AI research and then go on to speak at these institutions, in my experience with negative consequences. You were just actually speaking at the Bank of England. I understand as well that your newsletter Bits About Money is doing the rounds in the likes of Washington, DC. Can you tell us about how that’s been going?Patrick:Yeah, so nobody is more surprised about this than me, although I’m less surprised it as my career advances. I still have a self-conception as this. I got my start working for the internet writing bingo card creation software next to a rice paddy in central Japan. now descriptively speaking, it seems like the Senate Banking Committee and staff members thereof read Bits About Money very attentively and ask excellent questions based on it, including of other people—a little worrisomely. Well, for them, not for me. Similarly, I received an email a few weeks ago from the Bank of England. Not that I’m so unaware about how the world operates that I’m surprised that the Bank of England has email accounts, but I was surprised to get that email. And they said, in essence, ‘We became aware of you a few years ago when you were on Tyler Cowen’s podcast, you seem to not be an axe murderer. We’ve followed your stuff since then and we’re central bankers here. We deal with systemic risks of the financial system. We feel pretty good about some of those systemic risks. We feel less good about having an understanding of the technical realities of software in payment systems and in the financial system generally. And you seem to be credible on that subject. Can you come in and talk to us about it?’ So I flew out to London to walk down to Threadneedle Street and then chatted with them for an hour about three topics. They gave me permission to publish a derivative of that, and so I’ll publish it in my own spaces in the next couple of weeks. We talked about stable coins, about AI and whether it represents additional systemic risks to banking. Then I spent a lot of time walking through one particular incident, the CrowdStrike Falcon incident of July 2024, which extremely negatively affected the U.S. banking sector in a way that is less than institutionally legible. I wanted to bring them in. So here’s the after-action report that I was able to find by the Fed. And it seems like the Fed is kind of busy right now and has not written that after-action report. But I gave them the gloss of Microsoft’s after-action report and CrowdStrike’s after-action report and said, ‘I don’t want to brag, but I am an American, and I did successfully anticipate many of the things that were in these reports several weeks earlier in my newsletter. You should probably read that if you don’t already.’ So, yeah, an interesting time was had by all.Adam:Yeah, and I guess to the Bank’s credit, they made sure to give you a couple of years before bringing you on as well, rather than searching around for somebody that seemed vaguely credible from social media.Patrick:Legibility, man, I don’t know how many times I’ve given a shout out to Seeing Like a State in my various works and it’s quite popular in our community, but it is just downright difficult for institutions to legibly decide who has credibility on various topics and social media gives them a wider reach than they previously had. They don’t just have to like sort by institutional prestige and say, ‘Okay, you were previously at Harvard, that sounds good. You have a PhD, that sounds good. All right, we can bring you in to talk about macroeconomics’. There’s, you know, nobody knows if you’re on the, if you’re a dog on, if you’re on the internet and nobody knows if you’re getting a thousand likes per tweet or LinkedIn post, whether those likes are coming from, you know, relatively well calibrated people evaluating you, or from a botnet that you’ve hired, or just from folks who maybe not should be as weighted in the decision-making process as they are by default, if one is just sorting by likes descending. The Story of VaccinateCA Adam:Sure. So I’m keen to talk about the implications of VaccinateCA. You and I, when we were in San Francisco together after our [earlier] interview, we talked at length around some of the implications. It’s a story that I think hasn’t received anywhere near the level of scrutiny and attention it deserves. I’ve recommended it to my audience previously, and written about it online. Go read Patrick’s breakdown on Works in Progress. It’s long, but absolutely just a microcosm of modern institutional failures. You’ve also talked about the saga, particularly I’m thinking of the Dwarkesh Patel interview. Before we take the next step. Would you be able to just give us a five minute refresher of how it all came about?Patrick:Absolutely. So I’d like to ask the audience to cast yourself back to where we were in early 2021. There were widespread media reports in the States that the vaccine had been substantially delivered in November slash December of 2020. The physical artifact was available earlier, but the FDA approval only came down in November. Distribution began in many places in December and it was originally prioritized based on guidelines which were created by various central government actors in the United States and then promulgated to the several states and local authorities with some amount of variation based on where exactly you were, what exact authority you were listening to, and day by day, week by week, sometimes multiple times a day changes in policy. We can get into all that. But there were widespread media reports in January of 2021 that said patients and caregivers of patients, most typically their adult children, are calling dozens of medical providers across the state of California to find who physically has the vaccine available for patients that the tiering system says are immediately eligible to receive it. And on reading these media reports, I got frustrated on behalf of my friends and family in the state of California and thought, clearly the California tech industry is capable of centralizing this information. Just publish one website somewhere that has the information of who physically has these vaccines in inventory.And then rather than every patient calling every doctor’s office and every pharmacy, just look at the website, find the people that will say yes if you call, and then engage them in whatever process is required to get an appointment, whether that’s calling or going to their website or going to the county health department or similar. And I tweeted out on, I think about 14th, you know, if someone wanted to do an excellent Civic Technology project, you could create a webpage.do the work of calling around to get the information to put on that web page. Put it in one place and then allow people to access your web page and use that centralized information. This will decrease the load on the system in terms of the number of calls required to find each dose and it will get this information out in a much more scalable fashion. And I thought,’ well, it’s easy to criticize the rest of the world for not doing things well and just tweeting, I’ve got a brilliant idea. It doesn’t really produce any value.’ I had in my head an image of, like the right person to do this is a 24 year old with some tech industry experience and some time on their hands. I was a 24 year old once. I know what I would have said. I would have said, ‘well, what if that webpage blows up in a good way and tens of thousands of people are accessing it? My server bill might be more money than I’ve ever seen in my life’. I’m no longer a 24 year old. I’ve worked in the tech industry for number of years and I said, ‘if you get a $10,000 server bill as a result of being outrageously successful and helping the people of California Let me know about it. I will make it go away’. Then I thought well, ‘okay that’s done with’. Then a gentleman named Carl Yang took up the torch and he invited ten of his friends from the tech industry into discord, a program which is better known for arranging slaying dragons than arranging life-saving medical care, and they worked overnight to start putting together the groundwork for this website. I think it was originally a spreadsheet and then launched the next morning just in time to start calling pharmacies and hospitals across the state of California to ask them, ‘do you have the shot and could a 75 year old get it today?’ Carl mentioned on Twitter that he had created this Discord server, and I popped my head in with the idea of just saying, ‘I’m very glad you people are helping with this. Let me know if I can do anything to be helpful.’ One thing led to another. I got a little bit more involved than that. I think that was a Thursday. By the time the weekend was rolled around, we saw the graph of how many people were accessing it. And we had anecdotal reports of people successfully using the information to actually get an appointment and then close the loop by actually having the patient receive their vaccine. I had a heart to heart with the other ten-ish organizers and said, ‘look, I think we all just join this as sort of like a hackathon project, but this is very clearly, we have the chance of creating critical public health infrastructure, in which case we should not go back to our day jobs on Monday someone needs to lead this thing. I wish there was an obvious executive around this table to pass that off to. There isn’t. I recommend you make that me, and I recommend you tell me to immediately go to the tech industry and start raising millions of dollars to pay people reasonable salaries and to pay for reasonable expenses like the call centers were certainly going to need, and then we do this until we no longer need to do it.’ After a bunch of pairwise discussions, the team said, ‘all right, let’s do it’, and so we were off to the races. We ran that initiative for about six months until… we well we can talk about exactly the reasons why I closed down in six months, as opposed to seven or eight, but comes down to partly accomplishing the mission or being perceived as partly accomplishing the mission and partly due to fundraising failures, but we can talk about that later. We achieved substantial success. The reason anyone in the United States of America was able to Google the vaccine was Google was able to rely on us for the information. We partnered directly with the government of the state of California, many of the county governments in California, partnered on some level of direct or indirect with the official CDC blessed initiative for the United States of America. By the end of it, we’re tracking 75,000 plus locations across the country, with pins on the map and detailed information about vaccine eligibility, and how you would get an appointment at every one of them, which again, we were redistributing to the largest publishers in the world.Adam:Yeah, incredible achievement really for a bunch of, dare I say, largely amateurs in the realm of public health infrastructure.Patrick:Yeah, from just thinking about the people that we had attached as employees and core volunteers on the project, I think there was one person with material policy experience who had worked in the Obama White House on an entirely different thing, and perhaps two people who had some experience with doing public health infrastructure at scale, but from a very different view of public health. Nobody had epidemiology experience. Nobody had direct provision of healthcare experience. Nobody had a PhD on this topic. I joke sometimes that we were rank amateurs. Well, we’re tech professionals. Scaling systems is kind of what we do. But from the perspective of this exact thing, I think many important societal institutions would look at exactly the roster that solved this problem and say, wait, no, that is not who we would have allocated to the problem. Titanic Gaps in State Capacity Adam:So there were many, and again, go check out Patrick’s account, but many, let’s say charitably, insane policy decisions made that ultimately resulted in many people dying as a result of limiting vaccine availability and various other restrictions that were counterproductive or unnecessary that perhaps, as far as I’m aware, nobody’s ever really been held accountable for, either formally or informally via press and media discussion, broader societal, ‘hey, what happened there?’ That’s a whole topic in of itself, but I’m keen to just focus on the state capacity side. So I’ve got a few quotes from you that I’d like to read out. So the first one: We have a titanic gap in state capacity. The largest and most well-resourced organizations in the world (and thinking here of the US government, the state of California and so on), did not conceive of, approve and immediately execute an obvious and largely successful operational plan that non-specialists were able to draw up on Discord in a matter of hours. Implicit here is, and discussed elsewhere in the piece, that they actually spent hundreds of millions of dollars doing ineffective substitutes instead. What happened there? You know, we reach out to some sort of provider that’s already been pre-approved to try and do something quickly, spend tens of millions of dollars putting something together that is ultimately completely unsuccessful compared to what you guys managed to do overnight. How could this be so?Patrick:So. It’s worth saying that the obvious plan here is very not magic. There is no high technology involved. We are building the world’s least impressive inventory tracking system. And our proof of concept for that tracking system worked was I literally got on the phone to the, there exists a Walgreens at the corner of 4th Street and King Street in San Francisco. I’m aware that Walgreens exist because I had previously been to San Francisco on business and you know, their number is in the phone book or on Google.I called them up and I can recite almost verbatim exactly the way that call goes. ‘Blah, blah, blah, blah, press four to speak to the pharmacist, hit four. Hello, this is Walgreens. Hello, could a 75 year old individual get the COVID vaccine from you today? No, we don’t have it. Check back in two weeks. Thank you very much. Click.’ Now, like the fact of that phone call has proven that all you need to do is make tens of thousands of phone calls. It is not beyond the capability of society to make tens of thousands of phone calls. We have call centers available that can scale that straight to the moon. It’s strictly a math problem after that point. Then, given the capability of making that phone call tens of thousands of times, just write down who says yes and then publish it in any fashion to the internet. Honestly, I would have been very supportive of not merely publishing it to the internet, but there was a substantial amount of societal goodwill that government had in the moment for breaking the back of the COVID epidemic. If the government had just used suasion with places like the nightly news and said, we would really appreciate if for three minutes on the nightly news every day until we don’t need this anymore, you can just read out the places that have the COVID vaccine available. That would have been hugely impactful for exactly the demographic older folks that might not have internet access, might feel a bit of a digital divide, etc., that most needed the vaccine as of January, February, March of 2021. But what do we do instead? There’s often a narrative of tech solutionism where, ‘you engineers think everything can be solved by software, but the real world is complicated, etc. etc.’ That narrative mispredicts what actually happened here. The government policymakers were saying, we think that this is just a matter of, ‘There are multiple parallel distribution paths for this vaccine. Each of them is recorded in a separate database somewhere. We need a vendor to go out and make a system that will make all the databases talk to each other, use an API or something, maybe put it on a blockchain, I don’t know. Then after all the databases talk to each other, all of this mess is solved.’ The technologists said, ‘We don’t want to negotiate an API with all of the various supply chains that might have access to the vaccine right now. We can’t even get a list of all those supply chains. We just know that this must be administered in the physical universe at addresses that exist in the phone book. Each of them has a phone number attached to it. The simplest thing that works is call that phone number and ask the highly paid, responsible medical professional who sits next to that phone all day because that is their job, whether they have the vaccine or not.’ That’s another critical thing to take official notice of. If a database says that Walgreens at 4th and King doesn’t have the vaccine, and the pharmacist at Walgreens at 4th and King says, ‘Yes, I have the vaccine, I have my physical hand on the vial’. The pharmacist is always right, and the database is always wrong. For ‘Seeing Like a State’ reasons, the government often wanted to track towards databases or track towards spreadsheets versus tracking towards material reality. A spreadsheet can’t inject anyone with life-saving medication. The pharmacist can. We should prefer what the pharmacist says.One explicit example of where the diverges from reality is: the state of California in its infinite wisdom was sending allocations of the vaccine to any number of medical providers, including the various University of California healthcare systems. For people who are not familiar with the state of California, the University of California-at-{name of city here} is a large healthcare system which might have dozens of physical locations. It might have only one row in the spreadsheet though for a facility that healthcare is probably not actually conducted at. It’s just their central warehouse. They receive many things at the central warehouse, saline water, scalpels, radiological supplies, and the vaccine. Then they put it on trucks or in vans to all of the, you know, 45 places where they actually conduct the delivery of healthcare. Those might be quite far from the central warehouse, could be in a different county even. So the state of California institutionally believed that we can only expose availability of the vaccine in this location, the central warehouse, where the vaccine is not actually available to any healthcare seeker. We said, ‘no, no, the actual place that healthcare is delivered is these 45 locations that your spreadsheet does not know exist, but that we do. And we have called them, and here are the subset of that 45 that actually have it available’. I felt a little bit of frustration as a software engineer who has no reason to understand, you know, how the University of California healthcare system at City X conducts healthcare delivery. Why am I explaining this to public health officials? Why don’t they already know this? But we often found ourselves in that sort of situation. Government Software Development Adam:So one of the things that’s often criticized about government procurement, and delivery of software in particular, is this waterfall process. I’m gonna put something out there. So it sounds to me as if your process was something like, ‘Here’s a reasonable idea, doing this simply, the capacity of call centers is massive compared to the amount of calling we’re gonna need to do. I think we could put something up reasonably quickly and iterate from there’. The alternative approach is somebody had an idea that there was some solution out there associated with databases and APIs. Then wrote up an enormous amount of procurement documentation and so on about that, then everything that all of the failures subsequently are sort of downstream of that decision because there’s no capacity to revisit that because perhaps it was made at the federal level, President Biden’s busy and he’s got a million things to do, and there’s no capacity anywhere further on down the chain to go, ‘hold on, this isn’t working. We need to be rethinking our approach here’. Would that be a of a fair summary, do think?Patrick:I think that’s broadly fair. I will give the government officials some grace here. They were busy, they had many things on their mind, etc., but I can give some explicit examples of shortcuts that we made. We as a society decided, ‘OK, we are constrained by the RFP process. We need to get vendors who know what they’re doing into this.’ and so what often happened was, ‘well, we don’t want to spin up an entirely new vendor relationship because that will take 18 months to deliver the one software solution that will solve all these database problems. So instead, we’re going to vendor relationships and repurpose software that already exists’. We can talk about specific negative consequences of that, but one thing that happened was Castlight Health had a contract for having the United States vaccine availability database for routine, largely influenza, vaccinations for 10+ years. They said, ‘Great, we already have a vendor, this will cut out so much paperwork, we’re gonna use them for the COVID thing as well at the federal level’, but there were some technical consequences of that decision that reflected into the future in ways that were acutely negative for Americans’ ability to access the COVID vaccine, where the trade-off made sense in 2012 when it was just the annual flu shot, which obviously there are lives riding on the annual flu shot, it wasn’t stopping the entire economy on a dime issue that COVID was. There would have been political will to change those decisions had anyone ever explicitly brought up that decision, as like ‘this is a thing that senior decision need to make a call on on behalf of the American people’. But some combination of this being covered by multiple layers of bureaucracy, and some combination of just—‘We have such bad experiences in Washington with doing software projects that we only do them through the approved vendors and then we throw them under the bus if anything goes wrong’—resulted in no one making what I feel was a very obviously correct call with respect to particular subparts of that. Adam:So on that note, it actually brings us nicely to the second quote that I have here. The United States of America can be decisive and effective. It knows this. It teaches case studies about its ability to be decisive and effective in its history classes. The existence of history classes is itself evidence of decisive, effective action. But [the US Government] chose not to be decisive and effective because it believes itself incapable of simple things. This is one such example you were talking about earlier, websites that believes itself incapable of doing it. So it chooses not to be. It’s quite an interesting formulation you have there as well, that it chooses not to be decisive and effective despite knowing that it can be. Can you unpack that a little bit for us?Patrick:Yeah, the single issue that I think most causes the American government’s distrust of software as a way to cause its goals to happen in the world was the rollout of the ‘healthcare.gov’ nationwide marketplace for, well, not nationwide, very broad marketplace through to the delivery of insurance services under the Affordable Care Act, which was during the Obama administration.Adam:This is ‘Obamacare’?Patrick:Obamacare, but the relevant people in DC hate when you call it that. Well, on one side of aisle, hate when you call it that. The other side of aisle… Politics. I’ll stick to what I know.The received wisdom in DC is that you could be a well-loved president and have your signature initiative and legacy tarnished by the engineer’s failure to launch a good project and therefore, never be in the position where you are taking technical risk as the government. This dominates thinking about a lot of software including ones which were directly implicated here. I have lot of respect for the team that ended up running the website, vaccines.gov. Vaccines.gov received staff augmentation from the Digital Services Agency, and the Digital Services Agency was very clearly given the brief, and you can see the consequences of this in their formal write-up of it, ‘Okay, President Biden is gonna get on the nightly news and he is going to announce the existence of a website. We know what happens when presidents read a URL on the nightly news, it crashes, and then they look terrible. You will make sure that President Biden does not look terrible when the website crashes. That is like the number one operating principle. Staff all of your engineers on scaling this website to the absolute moon’. And the engineers did what they were ordered to do, which is...Again, I have immense respect for this team. I have immense respect for the people at the DSA. That is a misread of the likely impact of the president reading a URL on the nightly news. There was, in fact, no reasonable expectation that there would be a standing wave of traffic against this, when there was a standing wave of traffic against the healthcare.gov website back in the day. The misallocation of resources there caused cognizable harm to the people that the policy was intended to benefit, and also caused a delay of probably weeks in delivery of that website, and also a delay of weeks in other organizations doing their own thing to impact the initiative because many of us found ourselves thinking, ‘well, we need to be in the room where it happens and we need to not upstage the federal government and we are getting signals from partners in say the White House that it would be a bad idea to scoop President Biden on the existence of a nationwide website’. So VaccinateCA, for our part, we did not launch our national site publicly until a minute after President Biden got up to the podium to speak about the government’s national website. That was our decision. I will take ownership of it and made it for reasons, but those delays were not necessary if we made better policy choices. We could have made better choices as a nation to bring more appropriate resources to bear earlier there.Adam:So when you say better choices and better policy choices, there was a statement you made during that answer that the concerns about the website were irrelevant, maybe because of the likes of Cloudflare, and edge cloud and so on? Is it just those concerns are less relevant now than they were during Obamacare?Patrick:So without going deep into the technical weeds here, the thing that made healthcare.gov difficult is that it was a single website that was interfacing with many, many dozens of different providers and databases and doing write-heavy load and needing to fan out this information to places like the Social Security Administration and the IRS and coalesce their information back from them, and do this on a user-by-user basis because everyone has their own insurance policy. They don’t share their insurance policy with every other American. Finding the vaccines on the other hand is not a write-heavy load at all. It’s a read load. You just need to get the same list or some subset of the same list that everyone gets. The website doesn’t need to know your identity at all because you’re going to be calling Walgreens as your next step. So that scales to the absolute moon. You could service the entire nation off of a server that costs $2 a month from Amazon. You probably wouldn’t do that, but directionally, that’s the amount of engineering work required. Not, 20 people drop everything for six weeks to implement a caching strategy and a thorough engineering architecture work here, which wasn’t needed during the healthcare.gov rescue situation.Adam:Sure. So the general model, just coming back to what we talking about before, where you’ve got the sort of waterfall of, there’s a high-level decision that’s made about an architecture, or do not let this website crash or what have you, often based on generations of scar tissue that you and I talked about last time we spoke. And then from there, everything is just downstream, which perhaps when you’re talking about designing and building a bridge and you want some degree of repeatability, you know, ‘We’ve done this a hundred times before, we want to be able to just do that again’ and setting those requirements, and the scope of those requirements is unlikely to change dramatically as the project is delivered. Whereas it almost seems uniquely terrible approach when it comes to delivering software because the solution might be completely, you might discover after five minutes that it’s completely ineffective, it’s gonna take forever and so on and so forth, but you’re locked in at that stage. Vendors aren’t particularly interested in necessarily telling you this, because they quote a price and they get paid.Patrick:To repeat something that my buddy Dave Guarino has said on number of occasions, in the software industry, we assume that 10% of the work happens before you launch the website. Then 90% of the work happens as you evolve the website in response to requirements that you learn as you operate it. But in government contracting, it’s the opposite. We expect 90% of the work to be done in the waterfall. Write all the test cases, write all the requirements documents, deliver exactly the thing that is required, and then reserve 10% of the total budget for operating, but you essentially cannot make any changes afterwards without a sign change order and another 18 months of rigamarole.Adam:Yeah. Makes sense for a bridge. Patrick:Makes sense for a bridge. You don’t want to change the number of levels of the bridge after your first two months of operating it. But, you do want to have contact with reality on, ‘okay, the exact problem we were dealing with in, say January with respect to the vaccination effort is, you know, the demographics which are most affected, etc. etc., and our guesses as to how they will interact with the site will not match what we know when, for example, by May, many places in the nation we are no longer supply constrained. So, okay, the operational focus should switch from rationing and doling this out to the appropriate people to attempting to make this process as easy to get through as possible. But if you’ve, you know, pre-committed a year earlier, like, ‘okay, we’re going to need a rationing system on day one, deliver me a rationing system’, then you will, like the state of California actually did, spend tens of millions of dollars getting consultancies to write rationing systems. Monstrous Crimes I want to both be polite about the fact and be honest about it. We, the United States of America, through our elected representatives and through civil servants who represent our interests, committed monstrous crimes in 2021, which are against the laws, traditions, and constitution of the United States of America, including aggressively redlining the provision of life-saving medical care in a way which was designed to cause racially discriminatory outcomes with the provision of medical care.Just throwing that out there as a statement. With that caveat, one of the things that we spent tens of millions of dollars on was that we want your consultancy to write a website which will enforce residency restrictions. A residency restriction is essentially, when we are under a supply constraint, there must be some method to decide which people get it, and some people don’t. We have, in our infinite wisdom as the government, decided that equity, equity, equity is one primary thing that we are focusing on. A thing that we think would be contrary to equity is allowing anyone who shows up at the clinic to receive the life-saving medication. The thing that we are specifically worried about is relatively well-resourced people from advantaged demographics will use their superior access to transportation and information to travel to clinics which have the vaccine available and take that instead of that vaccine being used by someone in the local community who we intend the vaccine to go to. Therefore, to get an appointment to go to the vaccine, you will need to go to the county’s website, which is delivered by Accenture or similar, and prove to the website that you reside within one of the zip codes that we have allocated for those vaccine doses. Only then will you get the ticket, virtual or otherwise, which allows you to go to the pharmacy and get the vaccine. We spent tens of millions of dollars on that, targeting essentially a four-month window where we were acutely supply constrained. But we did not turn off residency restrictions on the websites after that four month window because we physically had no way to do that because that was not in the bid documents in some cases.Adam:Not only that, but the governor went on television and said, ‘anybody that skips the queue will be prosecuted. Any doctor that skips the queue will be prosecuted.’Patrick:Yep, and we will not just prosecute you, we will be aggressive about going after the reputational impacts to you, your institutions, and your licenses. That is virtually a verbatim quote from a governor.Adam:Yeah. So the consequence of this is you’ve got a supply constrained vaccine, you redline, saying you could only be in this [neighbourhood], you defrost a bunch of vaccines in the morning, not enough people show up that particular day. And so you throw them out because risking even saving people’s lives is not worth the risk of having your institution shut down and thus, you know, many, many, many more people die. It’s absurd.Patrick:Yep, absolutely the theater of the absurd. I really want us to remember institutionally that this happened. You know, there is a New York Times article which documents an actual prosecution of a Texas (I believe) family doctor for shoplifting. The medical/regulatory reality of these vaccines is they quote ‘go bad’ unquote after 12 hours. The physical reality is slightly different, but the paperwork says what it says. So if you have five, eight or 10 doses in a vial and you have three doses which have not been injected close to the end of the business day at your place of employment, either they go in the trash or they go to somebody. There was a doctor in Texas who made I think above and beyond efforts to find someone to receive the vaccine. Couldn’t find anyone, didn’t want to throw it in the trash and then administered it to his wife, I believe, and he was prosecuted for shoplifting and at risk of losing his medical license, etc. etc. The chilling effect of that was very real. When the governor got on the nightly news and said, we will end your medical institution if you enable anyone to skip the line, what a lot of pharmacies heard was, we could keep an end-of-day shots list. We could keep a list by the phone of people to call just in case nobody came in for the last two, three, five doses. We could extend life-saving medical care to them. However, we’ve been told that that will result not just in some consequences with regard to our vaccine ability, but that will result in this pharmacy, which administers thousands of drugs for thousands of conditions, many of which are critical to people’s lives and welfare. We will be shut down. So, okay, we will not risk that over one drug for one condition for a few patients. We would rather… I don’t want to throw the pharmacies under the bus for the following decision, but it’s important to understand that the United States of America made this decision. We would rather people die than violate the sanctity of this tiering system. That was the wrong decision to make, and we made it with our eyes open. We should seriously think about what madness led us to make that decision. Before even thinking about what madness led us to make that decision, I would love if we could openly acknowledged that we made that decision, but for very frustrating reasons about the US polity, no one is very interested in having that discussion. Sorry, I’m obviously passionate and frustrated about this, but we did make that decision. There are people who are no longer with us as a result of this. And I feel we owe it to ourselves, we owe it to history, and we, not to put too fine a point on it, owe it to the dead to be honest about what we did and for the purpose of doing this better the next time. If God willing, we’d never see a respiratory pandemic again in our lifetimes, but if it happens again, we must do better.Adam:Yeah.I want to just take one part of that, because it’s obviously an enormous topic and, like I said at the start, I think it deserves a lot more discussion. Just to take one small part of it: the decision that was made around this particular policy decision that then led to the development of a software package that then was locked in place. One of the things that is becoming increasingly clear to me is the idea that, again, unlike perhaps a bridge where there’s a business case process perhaps and we’ve identified the need for a bridge. Once that’s the case, and we go into design and build and so on, we’re not rethinking the nature of the transportation network that that bridge is associated with, and reasonably enough. Whereas when it comes to software, I think what’s often missed is this idea that there’s like, okay, well, software is some component, like you’re buying a thing off, whether it’s a large capital expense or a small one (it’s something that you’re buying off a shelf like a consumer product). Rather than simply the information layer of the implementation of policy and process. So if you’re thinking about software or delivering a software project, really it needs to be incorporated into the whole thinking of policy and process, right? So as you’re developing this thing, whether it’s a small scale, relatively straightforward piece or something significant, it needs to be thought of in the scope of the project needs to be, well, what policy decisions have been made here? Are they still appropriate? Is the process that we’re attempting to digitize here actually still relevant? And then perhaps, you know, even if that policy decision that we’re acknowledging is terrible, if we’ve got more of a flexible approach than at the very least when that finally changes, we’re in a position maybe to update the way that we’re actually implementing it.Patrick:Absolutely, yes. Software is certainly not a replacement for good policy. It’s a complement to it and it should be developed in the knowledge that policy can change and software can change and perhaps we should cause those changes to be mutually reflective of each other. There were some bright spots during the 2021 vaccination campaign where those infelicities that we baked into software, due to our RFP and similar process for procuring it, were worked around by people at say the ground level. One thing that happened routinely when VaccinateCA would call up pharmacists is say, officially a person like the one you just described cannot get the vaccine today because officially the county says we are still at stage 2C for example. However, unofficially, I do have doses available. So what you should tell people is tell whatever lie is required to get the county’s website to issue you an appointment. Come in for the appointment. I will not verify any of the information that is on your appointment card. I will administer life-saving medication to anyone who comes in for it. And I feel we owe a tremendous debt to people who put their lives, well, their professional reputations, their licenses, their careers on the line to say, I understand there are rules and I understand I swore an oath and where they conflict, I will favor the oath. Sorry, I a little emotional about this. But that ended up as a patch essentially over software which could not change on the day-to-day, week-to-week basis that policy was changing at. I also think there were so many infelicities with regards to how we communicated policy. The government, frequently at the state level and county levels, could not agree with itself. Not merely the state disagreed with counties, but like the county would disagree with itself as to what the policy was and had no channel for getting it out to the implicated individuals at safe pharmacies. So we heard from pharmacists very frequently that there is no blessed channel for us hearing, you know, like, where on the flowchart are we as of today from the State of California? I learned that if I happen to Google it or see it in my Twitter feed, but it’s not like I get an update every day from the county health department on move to 1C, we’ve, you know, passed yada yada in terms of our metrics. So a thing that VaccinateCA seriously considered doing for a minute was arranging to FedEx copies of local California newspapers to, you know, we will FedEx you a copy of yesterday’s newspaper so that you can learn what the policy was yesterday so you don’t learn what the policy yesterday was in two weeks when someone tells it to you around a watering hole. But given that one, you know, has abundant state capacity… Goodness, I hope we can arrange to have state capacity in the future, you could simply like, again, making phone calls to pharmacists is not an unsolved problem in material science. The state of California could have arranged to every day, like start the morning with a cup of coffee and a phone call to every pharmacist in the State of California. ‘Dr. Smith, thank you very much for your service, just letting you know you’re in Modesta County and the county says that it is on phase 2C today. Do you have any comments that you need to to give back to the county or anything that has been preventing the vaccination effort. Okay, talk to you tomorrow.’ A buddy of mine, Dave Kaston, who was a volunteer on VaccinateCA, made a comment to me which has stuck with me for the last few years. He said, you know, we have the National Guard available in the United States for a governor to say my state is under a state of emergency. Could be armed insurrection, could be a wildfire, and I have the authority as the governor to get 40,000 people in a room and, you know, ‘Put your machine gun on that side of the room. Now pick up this telephone and start making phone calls’. And we simply did not choose to order them to make phone calls. But had we done so, we could have had, you know, National Guard or reservists or similar backfill for any constrained part of the edifice that is the State of California, and just make it happen.Adam:Money was never a problem anyway, right? There are call centers everywhere that had a thousand times the capacity that would have been required for this.Patrick:Yeah, the right scale to think of is like a single high school class, if you had allowed them to make phone calls eight hours a day, could have made phone calls for the entire State of California. There are call centers which could not take that business because it is too small for them to contemplate. Mind boggling. VaccinateCA spent $400,000 or so on call center services over the course of six months, which is tiddlywinks by the standards of the budget of the State of California. Unused Capabilities Adam:Right. You raise the example of utilizing the National Guard. There’s a final quote that I have from your article about making use of this sort of capability. It’s the idea of a state [i.e. federal] government giving a call to the likes of Apple, Amazon, Google. You seem to employ many of America’s best experts in software logistics and marketing. I think it would be salutary if you put your very best people on solving this problem for America, immediately. I think you could probably have an initial proposal ready by tomorrow. If you think you would need to demonstrate moral authority or gain cooperation from any arm of government, you can put your project leader in a US Army colonel’s uniform and give them a piece of paper saying, do it, because the president says so. Well, you can’t do that, but I absolutely can and will. Does this sound reasonable to you? It’s a capability that was made use of extensively during World War II that we could have utilized again, that presumably the companies involved, they were unwilling to be involved informally without the sort of blessing of institutions, but to give them a colonel’s uniform, let’s say, and employ them in that way and pay them a salary of a the contract of a dollar, I think many would have jumped at the opportunity or the call to serve, but we didn’t do it.Patrick:I think a lot were, not so much waiting for the call to service, waiting for permission slash blessing, as you just alluded to. As I mentioned in the VaccinateCA piece, there was a real worry in the tech industry that causing the government to look ineffective in early 2021 would be held against them and their core businesses, due to the political implications of the wake of the January 6 events and similar. As a result of that, the call went out in the tech industry widely by government relations teams and PR teams saying, don’t get in the newspaper under any circumstances. We are putting our heads down. I believe that directly conflicted with the ability of public health teams at the tech companies, which by the way, that’s a real thing. There are 200 people at Google who work in the health department. But those teams were told to stand down on a variety of projects.I’m constrained a little about what I can say about things because obviously we made our partners in government and our partners in the tech industry some representations and I don’t want to cause anyone career damage by backing out of those handshake agreements, but I will throw out there that I’m aware of one major American software company which essentially had the project plan for VaccinateCA written up internally. The default state of any project plan is to go up to middle management, and then not achieve activation energy. [In this case] they simply didn’t achieve activation energy, but if you had tapped them on the shoulder and said, good news, your country needs you, I think we would have had a better chance of achieving that activation energy. Indeed, we were notoriously, in the last few years, tapping tech companies on the shoulder and saying the national government has some policy priorities and you should get aligned with them right now. [Reference to social media censorship] We can discuss some of those policy priorities in another podcast, but if we know we have the capability, and we know there are a lot of lives on the line, perhaps we should use that capability in service of something where the laws, constitution, traditions, moral sentiment to the United States of America is broadly in favor of us doing that, instead of using it for invidious ends or unimportant ends. I will mention for completeness that the statutory authority that allows the president to just pop a uniform on somebody say, you’re it, we’ve rediscovered that. So apparently some, I believe the CTO of Palantir got a uniform in the recent past. Patrick:The law is definitely still on the books. I’ll take no position as to whether the current administration is succeeding with its goals with regards to that uniform grant, but we can do it any time.Adam:Yeah, I guess one of the risks though that you and I discussed sitting down on the bench in Lighthaven though was if you’re in a situation where there’s increasing partisanship and a declining respect for the rule of law, let’s say, that, again, that may be a topic for another podcast, but in this situation, yes, Palantir is very clearly softly or informally politically aligned, shall we say, with the Trump administration and so is willing to take the risk of a change in administration and potential prosecutions arriving for that CTO. But one of the perhaps underappreciated consequences of this sort of world that we’re moving into is where there’s a much higher degree of risk that individuals would face compared to, let’s say, World War II, when DuPont was involved and did a massive amount of work for the Manhattan Project for a single dollar and was happy to take on that. If you’re in a position where every single time you’re asked to make that call of duty, that you think, ‘well, gosh, what’s the risk in it for me?’ then that makes it a lot more difficult to achieve those sort of results.Patrick:I hear that and I do acknowledge that the United States is in a period of a great amount of partisan disagreement and that sometimes that partisan disagreement is communicated in ways that are, let’s say, high temperature. However, I think we professionals who work in the tech industry or work in government should be rational with regards to the amount of risk that has been taken. I’ve made the allegation that we, as a society, through our healthcare system, through the government, committed massive, morally monstrous crimes. The total number of prosecutions with respect to those crimes is zero. Therefore, what is the real risk that a senior Google project manager whose job was setting up a database for a few weeks would find their tax return audited by the IRS? I think that risk is, one, minuscule. Two, allow people or corporations to run their risk analysis and see whether it’s within their tolerance or not. Three, we have options as the government if we want to make someone like durably immune to prosecution for official acts. We say what you will about the rule of law and goodness say, I’m greatly in favor of it and hope it continues. But we can change laws, we can write pardons, can etc. etc. etc., if this is important to us. So I don’t think we should catastrophize about the likelihood of, well, if the tech industry does things which are aligned with the interests of Administration 49, then Administration 50 might prosecute individual tech industry executives.Adam:Far cry from that yet. Let’s hope.Patrick:Yep, and I have a lot of faith in the American system of government and in our judicial system and in the Constitution and some impaired faith with regards to norms over the course of last few years. But we are talking about well-resourced, relatively powerful individuals. speaking only for myself, you know, I was willing to leave my family for six months during the middle of the pandemic to travel to California to run the VaccinateCA Initiative. Would I also have been willing to, what’s the phrase people use, catch a case over it? Yeah, sure. I will hire a lawyer. I’d be happy to show up in a court of competent jurisdiction this year and say, you know, I’m going to assert privilege under the First Amendment for publishing true things about publicly available information.Indeed, I made that argument to other people at Vaccinate CA that were worried, what if we embarrass the State of California by saying true things about their policies? And I said, well, we’re a very specialized newspaper publishing true facts about the COVID vaccine, and newspapers are all but inviolate in the United States of America. We have the First Amendment and very subtle jurisprudence here. And I think that argument will carry the day 100 times out of 100. Software That Works Adam:Right. So one of the things that’s interested me is the dynamics of civic hackathons, of which you could say, VaccinateCA was a particularly successful one. My experience is that they’re often filled with great people and a lot of enthusiasm and technical expertise, but that they’re perhaps too low context of the situation within, let’s say, city hall, or what have you, to make significant impact. Part of that might be because of the sort of non-integration with requirements and process and policy changes. But I guess the, the question I have for you is: how do we recreate the VaccinateCA on demand, or a similar institution? Or, failing some complete answer to that question, how would we make it easier next time for such a thing to arise?Patrick:So I’ll say it’s a strength of the system and nation that we were allowed to do it as an informal effort, and then to snowball our success from the informal effort to get some level of institutional legitimacy, and then be able to partner with government and larger non-governmental institutions like, Google. It would have been a lot faster if we had been a Skunk Works project within any part of the American civic system or a Skunk Works project within a large tech corporation. The tech corporations had their own reasons for not approving Skunk Works projects of that nature. So there’s one micro-patch that you can make. Simply heal the wounds that exist between the tech industry and the government and other centers of power in the United States and cause them to be willing to do things together again. But, you know, the Digital Services Agency, which has been rebranded recently I hear, and other initiatives have been attempting to make government more nimble about delivery of technical services over the last couple of years in the United States. There’s no reason in principle that a project very like VaccinateCA couldn’t have been run out the DSA directly or run out of the California Governor’s Office or run out of the California National Guard or similar. I think on one level, it’s a simple question of will, and I call it the will to have nice things. But an awareness that we do have state capacity when we want to, like, oh goodness, when America decides to do something, we really do it. So we should just decide to have nice things in some places. I think it’s a bit of a question of making policy decision makers more aware of things that they are already capable of. So dispel that ‘ugg field’ that we have around software. Get some people who are more software-adjacent or internet-native into positions of authority and not merely that they use Signal for discussing war plans. I’m sorry that was an aside, but some awareness of the fact that websites are not in fact the Moon Landing in terms of technical difficulty. Then, okay, what falls out from that? Is this a problem shaped such that we just need to publish a website is actually responsive to the needs here. VaccinateCA was in material part not a website, it’s a website and a calling operation. But neither of those is an unsolved problem in management science.So those are the bigger things with respect to getting uninvolved technologists more aware of how politics is conducted. One is increasing inter-connectivity in places like this space, Statecraft, etc., you know, make people aware of each other on the internet. One which I wish we did more of was simply have engineers watch people do work in government, and note what they see. Stripe does not endorse the following, but I will give this because I think it is instructive. Much operational software in the financial industry is very broken, and it causes the teams that are in charge of operations to have a very poor experience of their lives and jobs and causes mistakes. One canonical example of this is the bond at Citibank, which caused a team of three people in a particular part of the Citibank edifice to wire out a billion dollars of the bank’s capital because the software that they were physically on did not make it obvious that they were wiring out a billion dollars of the bank’s capital.I watched a discussion happen between two senior decision-makers at Stripe. From the top of IT, the view was, ‘Well, good thing that could never happen here because our operational software is excellent’. And the Head of Ops [i.e. operations, the people solving day-to-day problems] said something which rounded to, ‘well, I can think of three places where it isn’t excellent immediately, bang, bang, bang’. And the Head of IT’s face goes white like a sheet. And he said, ‘why am I hearing about this now? Why haven’t you been screaming that from the rooftops for the last couple of years? That sounds like something we could get one engineer to fix in like three hours’. And the Head of Ops said, ‘well, I’ve been working ops in the financial industry for a long career at this point, and I’ve never had software which isn’t broken. So I’ve learned that all software is broken all the time, and ops just has to roll with that. And being a relatively new employee here, I’m obviously not going to incinerate my political capital asking for crazy concessions like software that works’. So, you know, ‘heal all the wounds of the world’ is small step, but we have to have a culture of understanding that we can actually ship products that work into the world. Then reacclimate to ‘we should expect when you join a new organization in the government that you get issued a laptop on the first day and that your laptop boots up in five seconds, and not 30 minutes’. As Dave Kasten, a buddy of mine, often says, you learn early and often in government that everything is broken, all your technical systems are broken all the time, that’s just the default state of them, and you work around that. And so your laptop takes 30 minutes to boot up in the morning because it’s loading Citrix environments and blah-dah-blah-dah-blah, and that just is the way it is. We can’t accept that. There’s some combination of regulation, some combination of legislation, some combination of vendor management, etc. etc. that causes your laptop to not work the way that every laptop does in Silicon Valley. Because they’re physically the same laptops, guys. This is a choice. Let’s choose to not be broken.Adam:Hmm. So one of the things you’ve said previously was that for government software, and I guess in this case hardware and the interface, the procurement usually has a paper trail carefully testing that it is neither graft nor waste, which receives far more intellectual effort than the shipping of the software product itself does. [And elsewhere] that the optimal levels of fraud is not zero, and that paperwork trail makes project rigid and unable to learn and adapt as it goes. Is this a fact of life that we just simply have to accept? I don’t know. What I’m trying to say here is I think anybody with any sense would agree with the statement that you’ve just made. There’s been a lot of great books over the last five, 10 years about how we are failing in the way we’re doing things with respect to software development and procurement in the public sector. Yet, whenever I read them, I think to myself, gosh, there’s a very complex web of incentives, and we talked about scar tissue and so on previously, that makes the proposed solutions [i.e. in the books] much more difficult in practice than you might think.The various forms of risk avoidance and so on that a private sector employee that’s responsible for their bottom line, let’s say, you know, within their business unit just simply doesn’t face, you know, the public sector employees have very little exposure to upside risk and extreme exposure to downside risk. So as long as we can continue to make sure that, you know, I can’t be blamed, because I’ve done all the process right. So there’s very great incentive towards this continuing, but you seem quite optimistic and I just wondered if you had any comment on that.Patrick:I think we can certainly do tinkering around the margins. One of the most useful things for senior decision makers in the United States to have counterfactually done in 2021 was get on the nightly news and say, you know, ‘we’re in a state of emergency. I have legitimate authority granted by the voters to express to you the moral sentiment of the American government on behalf of the American people. My expression is that the overriding priority is saving lives via effectively administering healthcare and not paperworking things. So if you are ever in a position where there is a conflict between saving lives and paperworking things, I would like you to choose to save a life. If you need specific guidance or specific authority or similar, talk to your usual partners in government and I want those usual partners to hear the same message and use any authority available to you under statute or otherwise to fix things here’. I think without changing any regulation anywhere, simply that it’s not even a grant of authority, it’s just a statement of principles, could have done something in an emergency situation. We could consider doing formal grants of authority. We could consider going to the binders of regulations for things and saying, ‘OK, we’re going to do a carve out for the binders upon binders right now’. In fact, we don’t have enough binders. We’re going to put one more binder of carve outs available, and that binder will be one, if there’s a presidential finding of blah, blah, blah, then you get to bypass things. Two, blah, blah, blah, bypass things. Three, blah, blah, blah, bypass things. Perhaps that’s something that policy people might be willing to explore. Perhaps this is something that, you know, 19 out of 20 departments will not get comfortable with, but we can generate some evidence of outstanding success in department number 20 and then socialize it over a period of decades. Because I don’t think we got into the situation over one week, or one month, or one administration. I think we’ve backed into the situation over decades of our operation of the government and fixing it will likely take a while. So we should put in the hard yards of doing that.Then, whew. Policy work and rationalization of existing policies. I understand that there are people who’ve devoted their entire career to this and I understand many of them end up quite, I don’t know, cynical is quite the right word or burned out is quite the right word, but you are up against a hard problem for your entire professional career. I will say on behalf of of us who work in private industry, I also had a compliance department. There are all sorts of things that I would like to do that are illegal. Shucks. Well, what things that I want to do are legal? Then do more of those things to expose myself to the upside of doing things in the world. And again, I think when we allow ourselves to have a will to do wonderful things that are productive in government, we frequently discover the capability of doing wonderful and productive things. So we should be aware that that is possible and pursue it with some vigor. The Twitter Message Bus Adam:All right, we’re coming towards the end of our time here. So I’ve got some quick fire questions and then a few predictions before I sadly have to let you go. You once described Twitter, prior to the Elon takeover anyway, as the message bus for the federal government institutions and media. Personally, I’ve seen it act as something of almost like a release valve, allowing messages from lowly Amazon users, let’s say, to reach senior bureaucrats or the chief executive in a way that perhaps the official channels don’t. Is that kind of what you’re talking about, or is there more to it than that? Is the functionality still intact given the, you know, schism with Bluesky and the changes that Elon has made?Patrick:So that isn’t primarily the mechanism I’m thinking about, but if I can say a few words about that mechanism, there’s a concept that’s called escalations in many institutions where you have a retail-level user, a person who exists and requires services, who has a unsatisfactory experience with the bureaucracy and wants to get their concerns listened to by someone else. There are ways to achieve those escalations to not just the retail-facing part of the bureaucracy, but decision-makers that are higher up. One historical way to do that is engaging a lawyer. So your lawyer talks to their lawyer, but obviously lawyers are expensive. One less expensive way is getting a local journalist to take your side of things and to call the PR department and it looks like you’re foreclosing on Miss Mildred’s house, but Miss Mildred showed me some canceled checks for her mortgage. Do you have a comment on that? Then a bank will discover very rapidly that it is capable of listening to Ms. Mildred again. This works in government too. Obviously, there is the case that you can call up in the United States, we call it Constituent Services at one’s local member of Congress, and then the local Constituent Service representative will call into the agency and say, Senator blah blah blah would very much appreciate if you could get an update to Mrs. Smith by the end of this week. If you can’t get that update to Mrs. Senator blah blah blah would very much appreciate if you updated this office as to what the blocker is. Thank you very much. Twitter is a pathway to escalations, which doesn’t require you to understand that constituent services exist in the world and doesn’t require you to engage a lawyer. That is a wonderful thing. But the specific thing that I am thinking of is that there are policymakers in various places of the government who have formal methods of coordination and informal methods of coordination. One of those informal methods of coordination is that the sentiment of the administration is said by people who are known to be proxies in places like editorials in the New York Times. Other bureaucrats read that sentiment of the administration with respect to what the administration’s core priorities are. Then they orient their own lives around that sentiment of the administration. But, low-level bureaucrats, proxies, aligned individuals within civil society, etc., etc., are on Twitter all the time passing these messages around. And sometimes those messages travel extremely fast on Twitter. One occasion where this happened during the last administration was, Elon Musk had some comments with regards to like individual flight plans of people available in some disaster relief scenario. And the Secretary of Transportation, whose name I unfortunately have never heard pronounced. [Pete Buttigieg – Adam]. Buttigieg, Secretary Buttigieg was literally replying to him directly on Twitter, “I don’t understand that that is in fact the case, but I will look into it immediately.” So know, fast resolution of those concerns. Also, you know, that is a signal seen by other departments on, ‘you might have thought this is a non-aligned individual and therefore do not extend them, say, the benefit of the doubt on this issue’. But you are now getting someone who is obviously an aligned individual to say, ‘oh no, this is something where the administration’s priority is very clear. We are absolutely 100 % behind the disaster relief effort here and will take reasonable efforts to expedite it’. I would expect that other parts of the American government seeing that in Twitter in the palm of their hand or seeing it reported in the American media, were like, ‘OK, message received, I should do that’. And that’s why I think it is the message bus, which is a computer science term for the same thing. One person telling many people on behalf of the system that this is our new operating model here. [Here is the thread Patrick is referring to. In further evidence of the ‘alignment’ dynamics discussed above, the situation was widely reported after the fact as Buttigieg ‘clapping back’ about Elon’s conspiracy theories that his relief flights were being stopped by the bureaucracy. Buttigieg himself however made no such claims, and between the ‘message bus’ working as described, and Buttigieg’s direct intervention, the situation was quickly resolved—as Buttigieg confirmed later.]Adam:Yeah. I understand there’s something similar formalized in China as well. You’ve got these sort of various messages that work their way down through the system. It’s a very interesting example that you raised there…Patrick:For what it’s worth, there have been many changes in Twitter over the course of the new management. Not all of them for the better. I shake my fist that my podcast links get deprioritized. But it is still the case that many members of the American journalist class are quite tied to Twitter. It is still the case that people who want to have a career in Washington are quite tied to Twitter. It is still the case that CEOs of American tech companies are extremely, extremely, aware of things that are said about their companies on Twitter. So that is where the action is, and I’d encourage people who are decision makers to understand that is where the action is and take appropriate actions based on that. Customer Service Adam:Sure. So one question before we get into some quick fire predictions, you obviously spent a significant period of your career, necessarily obviously to everybody, but you did in Japan, working with bureaucracies and private businesses. What’s something that a process, an idea, a practice that could be simply adopted or relatively easily? Some things are obviously highly culturally dependent, but there’s got to be some things there you go, gosh, if we just copied what they’re doing there, copied and pasted, that would be a significant benefit to our public.Patrick:I’ll make an obligatory disclaimer here, which is I am an individual that has an immigration status in the nation of Japan, to which I intend to return in the future, and I am therefore careful about what I say with regards to the operation of the Japanese government. That out of the way. One thing that is very different in Japan versus the United States, which is the country whose polity I’m most informed about that isn’t Japan, is that if you go to City Hall with the question on anything that is within the sphere of operation of the government. [In Japan] if you have no idea, as a citizen who might not have a lawyer or a competent regulatory expediter at your beck and call, who I talk to about this. You just go to the reception desk and ask, ‘I want to open a restaurant. Who are the open the restaurant people?’ The reception desk will always know who to point you to. So simply caring a lot about what the retail experience of government is, and what is the equivalent to the reception desk? And how do we say, you know, descriptively, Japan puts some of the best people in the Japanese government at the reception desk and not at the senior levels of policy. That’s a choice. There are pluses and minuses to that choice. But could we [in the West] care an acute amount about what someone’s first experience is with the government when they want to open a restaurant? Can we make sure that the reception desk or the website knows, if you want to open a restaurant, here’s the first place you talk to. Can we make sure when they get to that place? Opening a restaurant is not the highest frequency thing that happens in say the City of San Francisco, but it’s also not a rare case for the City of San Francisco. And it’s not an unknowable case for the city of San Francisco. Obviously people will attempt to open restaurants next year. Obviously there is like some amount of procedures that they will go to, and we can know those procedures in advance. Let’s write down a list of all of them and give people the, ‘So you want to open a restaurant? document’ when they attempt to engage us about opening those restaurants. Then let’s go over ‘okay, what are the things that are extremely high salience that are high volume?’ Like, should we make a checklist that is interactable? Should we expedite this with software? Should we make people available to walk you through questions about this, et cetera, et cetera? These things round to free. Well, obviously some money to put someone out there but there is no great intellectual content in just saying, you know, if you walk into the city hall in Chicago and ask and who generally should I speak to about taxation issues, making sure they have an answer for rounds to free relative to the, you know, multi-billion dollar budget of the City of Chicago. Like, let’s make that happen versus saying there is no reception desk anymore.Adam:One of the things that I learned when I was in Taiwan researching their public sector was that they had a very long running ‘ease of doing business’ program where they’ve just quantified various friction points throughout the system—interconnecting new wastewater and blah, blah, blah, blah, blah, all these taxation issues, permitting, so on—and have just been steadily working their way through the list to make it easier and easier. This is particularly targeted at small commercial businesses. I’ve written about some of the difficulties they have with permitting, and particularly for large industrial, even TSMC, which has a direct line to the president, has difficulty building new fabs. But leaving that aside, I think this sort of general process of just accounting for all of the different things that happen, like you say, the commercial businesses, and then just sequentially working through them could be very easily broadly adopted.Patrick:Yeah, if I can make a related comment to expand on that, as someone who employed people in a variety of US states for VaccinateCA, it was maddening to me as a small entrepreneur who had a lot of things on my mind, how difficult it was to simply get spun up with our first employee compliantly in a particular US state, because I had to correlate between here’s what the Department of Labor wants, here’s what the Taxation Department wants, and here’s what a third department wants from me, and there was no single document that listed all of those and they each have a different identification number that you have to apply for, etc. etc.It seems very clear to me, particularly in an environment where we’re nationwide employment via the magic of remote hiring, that there should be a state employee whose job it is to say, ‘you want to hire the citizens of this state? We want to help! What do you need to make that as easy as possible?’ How can we decrease the total amount of entrepreneurial effort to get set up, to hire your first employee and start paying us taxes and, you know, payroll, unemployment insurance, etc. Like, how can we reduce that from a two-day process to 30 minutes and make it much more deterministic for… You know, it shouldn’t require a PhD and HR to successfully employ someone in a state. So, yeah.Hopefully with the remote hiring boom, there’s something of internal market where the states that are better at doing that see the benefits of it and other states are rushed to copy. A thing that my once hometown of Ogaki in Japan started to do, which I thought was both beautiful and something that many cities across the world should do, is if you’re considering moving to Ogaki, which they would like to encourage because there’s been outflow from the Japanese regions to Tokyo over the last many years. They will happily have a city employee get on a Zoom call with you for 60 minutes and hear a little bit about your situation. ‘So what’s relevant to your interests? You have school-aged children? Let me tell you a little bit about the schools in Ogaki and then I will read some PR the town has written and tell you about my own experience’ and then we go on our happy days together. Given that if a professional moves to a city, the tax base of that city increases, it seems very obvious to me that almost any town in the Western world should be able to math out, yes, it definitely makes sense for us to have a bespoke conversation with any person who wants to move into this town. Let’s make that easy. Again, not rocket science. You have a compassionate individual be willing to talk for a few minutes about your challenges and then like, read a script, and perhaps do a few Google searches on your behalf. Like, perhaps you have a question about, I don’t know, ‘my son has a learning disability. I don’t know what Ogaki does about learning disabilities, and I don’t know how to look for that because I am a civilian.’ Well, you as someone who works for the city government, is capable of using Google and capable of Googling the obvious things and saying, ‘OK, you know unsurprisingly this city has programs with regards to individuals with learning disabilities and if you and your son were to live here in Ogaki we would be happy to help you with that and one place that you could have as an entry point there is your school, another place is the following office, I can reach you their phone number right now if you want and you can have a call with a responsible individual’ and you know just be willing to be out there on behalf of citizens and residents.Adam:There’s a lot of tacit knowledge as well that can just come through passively about how different things, places are organized geographically or what pinch points might be that just makes it a lot easier for people to move and make that jump.Patrick:I think even if organizations did this for a week or two and then asked the people who were doing the calls, ‘what do we hear a lot on those calls?’ You would get immediate information that you can impact back into the policy arena on, ‘okay, what isn’t clear on our website right now? What are emerging sentiments of people that are considering this town strongly? They are like quite concerned about civic beautification, whereas maybe our town has been institutionally concerned with infrastructure development. Maybe we should rationalize those priorities, etc. etc. Or maybe do a better job around the things that people mention are priorities where we think we’re very good at that. But apparently, people still worry about it. So there is a communications gap. Let’s rationalize our communication strategy around it’. As the tech industry would say, if you’re a project manager or engineer and aren’t routinely getting on the phone or Skype or similar with your users, you’re doing it wrong. There’s so much low-hanging fruit there, so we should grab it. Predictions Adam:All right. Okay, so just to close this out, some quick fire predictions. You can feel free to either answer just in the affirmative negative or do the superforecaster thing and give a percentage if you want.Patrick:I’ll try to give percentages, but I’m often less numerically inclined than previously on the math team and have an engineering degree would predict. But OK, ready for it.Adam:Fair enough. Okay, so firstly, following on from the discussion we had about procurement and state capacity, the California High Speed Rail Authority aims to connect Gilroy, which is roughly a 40 minute drive south of San Jose to Palmdale, which is an hour north of LA, by 2045. Do you think that this will be complete by 2050?Patrick:5%.Adam:Okay! Do you think that San Jose and Hollywood, let’s say, so this is in the very South of the Bay Area, not up in San Francisco, down in the South of the Bay Area, and let’s say Northern Los Angeles, do you think those two locations will be connected by rail this century?Patrick:This century? Okay. This century is a longer time. 60% and increasing, I’ll say, but weakly held.Adam:Okay, all right. So an AI question, you think, do you expect the number of public servants to increase or decrease as a percentage of the population as a result of AI adoption over the next decade? So if we did some regression about the degree of AI adoption within an organization, do you think you would see an increase or decrease in the number of public servants as a result of that?Patrick:I do not think that will causationally cause an increase or decrease.Adam:Just unchanged?Patrick:Unchanged.Adam:Regarding centralization: 10 years from now, will public systems and jurisdictions be more or less centralized than today? So thinking about, for example, the United States, if you had commentary around, let’s say, Japan or other places you’re familiar with, that’d be interesting too. Do you think that gonna be, power will be more concentrated in the central state or more devolved to, let’s say, local governments and other jurisdictions?Patrick:It’s complicated, the policy environment being complicated. I will say in the spirit intended, there will be more power in the central government rather than less.Adam:What would be, do you think, the main driver behind that?Patrick:The scale advantages of the internet as a technology, the ability of small numbers of individuals to work their will upon the world by understanding the points of leverage within the system, and also scale advantages to IT generally and to AI specifically will cause like much of the devolution of power to local governments has always been about the subsidiarity argument that you can only understand things when you’re in close physical proximity to the problem. But if there is a surfeit of understanding in the world that can be solved by having arbitrarily large amounts of cognition to throw at it, and we will soon have arbitrarily amounts of cognition at, well, we status quo as of October 2025, have arbitrary amounts of cognition at various levels to throw at problems. As that gets metabolized by government, a lot of things that were previously only possible if you had 30,000 people across the United States doing them will be possible by a small team of people in Washington, and I think it will be incentive compatible for some of those small teams to actually grab that power that is available.Adam:Right, that makes perfect sense. The landscape becoming more more legible over time and our ability to influence it only becoming more so. So we’re sort of beginning and ending this podcast with discussion of Seeing Like a State. Patrick, thanks so much for chatting with me.Patrick:Thanks very much for your time and attention. And if I can ever help folks out in any capacity, whether you are a central bank or just somebody else on the internet like me, drop me an email.Adam:Wonderful.