We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
cover of episode #78 Balaji Srinivasan: Exploring COVID-19

#78 Balaji Srinivasan: Exploring COVID-19

2020/3/13
logo of podcast The Knowledge Project with Shane Parrish

The Knowledge Project with Shane Parrish

AI Deep Dive AI Chapters Transcript
People
B
Balaji Srinivasan
S
Shane Parish
Topics
Shane Parish: 就COVID-19疫情的相关问题与Balaji Srinivasan进行探讨,包括疫情的定义、病毒的起源、传播方式、检测方法、以及政府和个人应该如何应对。 Balaji Srinivasan: 从个人背景出发,详细解释了COVID-19病毒的命名、起源(可能与蝙蝠有关)、传播方式(空气传播、接触传播、粪口传播)、检测方法(RT-PCR、测序),以及影响疫情发展的关键参数(R0、有效再生数、住院率、死亡率等)。他强调了无症状传播的危险性,并预测未来可能出现家用基因测序仪等技术来进行日常病毒检测。他还分析了政府在应对疫情方面的不同策略(集中式和分散式),并指出美国政府的集中式应对策略存在不足,建议政府紧急扩大“尝试权”法律,以加快药物和诊断测试的审批流程,并通过数学模型来评估感染风险,从而决定是否取消大型活动等。 Balaji Srinivasan: 详细阐述了COVID-19疫情可能带来的次生影响,包括供应链冲击、旅行限制、远程办公的普及以及由此带来的社会隔离和心理影响。他认为,疫情将加速远程办公、在线社交和虚拟现实等技术的应用,并可能导致社会结构发生变化。同时,他还分析了不同国家政府在应对疫情方面的差异,并指出新加坡和以色列等国家在应对疫情方面表现出色,这与这些国家政府中拥有较多科学和技术专家的背景有关。他认为,增长率外推法比基准率外推法更适合于预测和应对疫情。最后,他呼吁更多具有科学和技术背景的人士参与到公共事务中,以弥补政府在应对疫情方面的不足。

Deep Dive

Chapters
This chapter delves into the origins and nature of COVID-19, exploring its initial naming as the novel coronavirus and the reasons behind its current designation. It discusses the virus's potential origins in bats and its connection to wet markets in China, specifically the Huanan Seafood Market in Wuhan. The discussion also touches upon molecular phylogenetics, a method used to trace the evolutionary history of viruses, and the challenges in pinpointing the exact origin and initial transmission of COVID-19.

Shownotes Transcript

Translations:
中文

COVID-19 is the formal name of this disease. It was initially called the novel coronavirus and nCoV-19, novel coronavirus 2019, denoted both the virus and the disease. And now they have sort of factored those out into SARS-CoV-2, which is the virus, and COVID-19, the disease.

Hello and welcome. I'm Shane Parish and you're listening to a special episode of The Knowledge Project, a podcast dedicated to mastering the best of what other people have already figured out. This episode is special because we're going to deep dive on COVID-19. I haven't really done an episode about a timely issue affecting a lot of people before, so consider this a pop-up episode. We'll do these from time to time when we feel like we can add something to the conversation.

This episode is also special because it has no sponsor other than us. If you'd like the podcast with no sponsors and you want to support us, you can go to fs.blog/podcast and click premium. You'll save time, get transcripts and a host of other goodies and help support what we're doing. Today I'm talking with Balaji Srinivasan. I've said that wrong and I'm sorry, it's not easy to pronounce.

I want to preface this conversation by saying the opinions expressed in here are those of the participants and not Farnham Street Media. The participants are not medical professionals. I think doctors are some of the most incredible people in the world. And if you're having symptoms, you need to see your GP. Our guest is also not a pandemic expert. He is, however, one smart dude, and he thinks in a very multidisciplinary way. This episode is full of what ifs, what could be hypotheticals, what's going on.

It borders on sci-fi at some points. As you'll quickly find out in this interesting interview, I don't know anything about the subject, but I'm about to dive in and learn. I'll tell you more about our guest in a moment. It's time to listen and learn. ♪

The IKEA Business Network is now open for small businesses and entrepreneurs. Join for free today to get access to interior design services to help you make the most of your workspace, employee well-being benefits to help you and your people grow, and amazing discounts on travel, insurance, and IKEA purchases, deliveries, and more. Take your small business to the next level when you sign up for the IKEA Business Network for free today by searching IKEA Business Network.

So we're going to talk COVID-19 and I just want to preface this conversation by saying this isn't medical advice if you're experiencing any symptoms whatsoever. You should go see your GP. We're not doctors. I'm definitely not as smart as you are. I sort of represent, I think, the average person. I'm just curious about this subject and you seem to know a lot more about it than most people. Can you walk me through a little bit of your background?

Sure. So I'm a PhD, not an MD, a PhD in electrical engineering with an MS in chemical engineering from Stanford. And some of the specific work that I did in grad school was around genetic circuits and microbes. That's looking at how viruses and microbes are wired internally, their system diagrams, protein interaction networks, statistical and computational analysis of their genome sequences, you know, that kind of thing.

And when you think of that kind of work in genetic circuits or systems biology, that's at the intersection of electrical engineering, computer science, genomics, stats, biochemistry, et cetera. I also have some papers in clinical genomics, population genetics, and so on.

I then taught CS stats, bioinformatics at Stanford for a few years as a lecturer in the stats department. And then in 2007, 2008, I founded a genomics company called Council with a Y. I was CTO and co-founder of that. I was skilled to do more than a million diagnostic tests and change the standard of care for so-called Mendelian genetic diseases and was sold for $375 million.

Separate from that, I've been a partner at Anderson Horowitz, a multi-billion dollar venture capital firm where I did a bunch of investments and recruited people that helped build what became our crypto and our bio funds.

So that's some of the stuff that I've done in the kind of broadly bio space. Separately, I also taught a MOOC course, a massively open online course for 250,000 students. I've done a bunch of angel investments in both cryptocurrencies and companies, so Bitcoin, Ethereum, etc., but also superhuman, Soylent Land School. More recently, most people know me for crypto stuff because I've been active on Twitter for that topic, and I also sold a company to Coinbase and served as a CTO of Coinbase until recently.

So, you know, kind of jack of all trades, master of none. But actually my formal training is in genomics and I've founded a large diagnostic testing company. And diagnostics is an important dimension of this whole thing, as you can see from the recent, you know, FDA and CDC stuff that's been in the news.

So bioinformatics, diagnostics, statistics, those are things I do have expertise in. I'm not going to give you medical advice at all. But as I think you know, many papers that are written today in places like NEJAM or the Lancet, 11MD and a PhD like me as co-authors, one on the clinical stuff and one on the stats, bioinformatics stuff, and they're usually equal co-pilots on the paper.

So anyway, that's my background. I think I'm qualified to comment on aspects of this, especially those related to diagnostics and stats. But the topic's inherently multidisciplinary. Nobody knows pulmonology and infectious diseases and epidemiology and genomics and supply chains, all of these different things at once. So I don't think it's useful to be very narrowly credentialist, though credentials can certainly help to an extent when evaluating information. That's where I'm coming from.

Preaching to the choir here, man. You're one of the most interesting Twitter accounts I've followed for years. I've always wanted to have you on the show. I didn't think this would be the reason we get on it. We'll have to do another one later and learn more about some of your thinking around different topics. But I think it's important for this conversation if we can work to separate facts, opinions, and assumptions.

And if we can try to preface things with this is true or this is what I think and here's why I think it or I'm assuming this to make this happen. Just so we're trying to get people the best information possible. So I want to start with something super basic, right? Like I'm not well versed in this. So what's an epidemic? So an epidemic is the spread of a new disease.

dictionary definition, widespread occurrence of infectious disease in a community at a particular time. But sometimes people use it to refer to like, you know, the obesity epidemic, which is a more kind of colloquial usage where at least we don't know if obesity is caused by an infectious agent. Most people think it's caused by eating, but maybe there's an infectious thing on top of that. So people sometimes use it casually, but at least in this context, occurrence of infectious disease in a community. And then maybe to your next question, a pandemic, you

is when there's an infectious disease that is occurring on multiple continents or the whole world and spreading. So it's, you know, pan in the sense of everywhere. And, you know, what's interesting, just about that term for a second, WHO, the World Health Organization, has not actually declared this a pandemic yet. They're still...

I think kind of, I don't know, in some mental model where they either think it can be contained or for political or sociological reasons are deciding not to call it such. Simply from like a dictionary definition standpoint, this is definitely spreading in an uncontrolled way on multiple continents.

So in that sense, it's a pandemic. What would cause the hesitation for labeling it a pandemic? I'm not an expert on the internal politics of WHO. I've seen people charge that it is due to pandemic bonds that would get paid out or WHO's relationship with China or various other kinds of things. I'm not privy to that. I wouldn't be able to give an informed opinion on why WHO has not declared it a pandemic.

However, I think a number of folks in the space who have been watching this, whether in diagnostics like my background or epidemiology or genomics, are kind of wondering why that official declaration hasn't been made because it's independently observable. You've been front-running the media on this for months. You seem to be well ahead of the curve labeling things and sort of like,

ascertaining what's going to happen. But what is it we're even talking about? What is COVID-19? So COVID-19 is the formal name of this disease. It was initially called the novel coronavirus, and nCoV-19, novel coronavirus 2019, denoted both the virus and the disease. And now they have sort of factored those out into SARS-CoV-2, which is the virus, and COVID-19, the disease.

Now, I think the reason they chose that is they didn't want to, and actually I think this is a good, you know, smart thing on their part, didn't want to stigmatize the region by calling it like the Wuhan flu or the China flu. Like Reddit has a community called the China flu. That's not helpful for lots of reasons. You know, like that area is hard enough hit. You don't want to hit it like, you know, for the rest of time, trying to blame it on them or anything like that. So I agree with their decision to kind of give it the sort of at

abstract name, but COVID-19 is the name of the disease and SARS-CoV-2 is the name of the virus. And so what is it? Like, where does it come from? I heard a rumor there's like a bat or something. So there's something called molecular phylogenetics where you can actually build evolutionary trees of not just, you know, humans and dogs and cats and so on, but also viruses and bacteria and what have you.

It's called molecular phylogenetics because normal phylogenetics, you're kind of just looking at macroscopic features of an animal, like does this have four legs or two? You know, does it have fur or not? And trying to classify them that way, right? Those are like features. Yeah. With molecular phylogenetics, the features are more like, do you have an A in this position or a C? You

Do you have a T or a G? In this case, it's an RNA virus, but the concepts are the same. You're not using DNA, but RNA. And the molecular file genetics of this coronavirus

You can put it in a tree with other related coronaviruses, and you can see that it's similar to these coronaviruses that came out of Yunnan province. So evidently there's some coronaviruses that were in bats, or at least labeled as such on GenBank, where they appear to be viruses that were resident in bats from Yunnan province in China.

So this particular substrain of virus appears to be most similar to those. And so that's why people have talked about, oh, did it come from bats? But we don't really know, I don't think. I haven't seen anything that actually nails, okay, this was patient zero and this was the exact moment. People think it's something in the vicinity of this Kwanin seed food market in Wuhan, but

I shouldn't say it doesn't matter. It does matter because if you know the origin of it, sometimes that can give you a clue for how you, for example, culture the virus or something. Sometimes there's a clue that's not obvious if you understand how it came about. Maybe the animal that spreads it is immune to it, and maybe there's some molecular biological feature of that animal we can clone for treatment, that kind of stuff.

But the origins, you know, people speculate it's a so-called zoonotic thing where it made the leap from animals to humans. And they speculate that it's something to do with people eating, you know, animals at this seafood market, which is not just seafood, but it has a bunch of other kinds of animals there. Is that because you're eating like raw animals or they're not cooked properly? Yeah.

Well, you know... Or just eating the animal in general or certain parts? I believe that these wet markets in China are selling... I have not personally visited the Wanan secret market. I don't think it's going to be a tourist attraction for the...

My understanding is they sell both dead and live animals. And so, you know, we're used to buying chicken, you know, but there you can buy a chicken or a fish and kill it yourself, right? And slaughter it yourself, right? And so there's folks who have written about how, you know, animals in that condition are very stressed out and, you know, their immune system may not be as strong and they're in close proximity to each other so things can jump back and forth.

I can't say how the thing arose, but that's likely... Let's call it the most...

She's the most plausible thing. That's the default assumption right now, right? That it came from that Wuhan and Sifu market. But with molecular phylogenics, I think eventually we'll be able to resolve that and trace where the most recent relatives of the virus are. But that's kind of an archaeological dig that may not happen for some time until things calm down. How did we find patient zero? So to my understanding, again, in literature, you see that in early December, you start to see patients in Wuhan.

And these patients were presenting with something that felt to the community sort of like SARS 2.0, because SARS was something people remembered from 17 years ago, from 2002 to 2003. So these patients were presenting and people were seeing more and more of them. By the way, patient zero in the colloquial sense means the first human to contract the virus, but in medicine it's usually the index case or the first documented patient in a disease epidemic. Those aren't always the same person.

Anyway, so the first documented patients were discovered in December, but if they were sick in December, then it probably means it got started sometime in November, got infected around that time because WHO has said the virus incubates for 1 to 14 days. So if your first case is December 1st, the person probably got infected in November, then things started compounding from there. I just want to walk through this in my head here. So a patient shows up at a hospital, looks like SARS 2.0, exhibiting symptoms. Then they do a, what do you do, a biopsy? Yeah.

Like how do you test for something like that? So the short version is that you take a swab, collect a sample from the patient's mouth or nose, you send that to a lab and you get back a positive or negative test result.

The longer version is that it starts with sample collection. You can take a swab, as I mentioned, for the nose or mouth. There's also other ways of collecting samples like aspirates or washes, which basically vacuum or wash material out of the patient's nose or throat into a sterile container. This is a respiratory or a virus that seems to target the respiratory system. So that's where people are typically collecting samples from, the nose and the mouth.

Then when you do that sample, there are a few different kinds of tests. An important one is the so-called RRT PCR protocol that the CDC has published, real-time reverse transcription PCR. As mentioned, you take these samples via swabs, then you extract RNA, then you turn that RNA into DNA versus reverse transcription, and then you detect the levels of DNA by amplifying it via PCR.

PCR is the polymerase chain reaction, a standard technique to amplify and detect small bits of DNA, even if they're present in a mix of other stuff. You can use it for things besides detection.

When you're doing this RRT-PCR, you also have some controls in there to check against various error modes. Now, if all the controls come up right and did the assay right, the virus is present, you should see the curve representing the amount of DNA copies of the original RNA rise quickly above a certain threshold. And that shows you that the viral RNA that you're looking for is abundant.

And the actual CDC test looks at this not just for one locus, but for three loci, meaning three different spots on the virus's genome. One of those spots, I believe, is for all coronavirus, and two loci are specific to this SARS-CoV-2 virus, the virus that causes COVID-19. Incidentally, the primers used to detect these loci have had bugs in them as being widely reported.

Now, the primary thing you get out of this RRT-PCR type test is basically a 0-1 presence-absence signal of whether the virus is present. You also get some quantitative information on viral load, a measure roughly of how much virus was present in the sample.

So, you know, there are other kinds of tests, though, they're worth mentioning, important ones, so-called sequencing assays that give you not just a 0-1 for the presence absence of the virus, but the full RNA sequence of the viral strain, the ACEs, U's and G's, rather than ACEs, T's and G's because it's RNA.

An example protocol for this is the Illumina protocol that was published to sequence SARS-CoV-2. This gives a lot more information than the RRT-PCR protocol as you get full sequence data, and that can be used for molecular phylogenetics, for looking at viral evolution, that kind of thing. Now, both of these tests, I should emphasize, are quite standard. People have been doing RRT-PCR and they've been doing sequencing on Illumina machines for basically a decade plus at this point.

The main difficulty here is that the SARS-CoV-2 virus is very contagious and dangerous, so you need to be even more careful with sample preparation than normal. Okay, that makes sense. So how does it transmit?

Okay, so first let's start with some definitions. The R0, which is basically an R with a subscript zero, is the basic reproduction number. The Australian Department of Health defines its average number of cases directly generated by one case in a population where all individuals are subject to infection, there's no immunity from past exposures or vaccination, nor any deliberate intervention in disease transmission.

So if the R0 is 3, the first person infected, it passes on to 3 people, who in turn each passes on to 3 more, giving 9, and 9 people infected and so on. So it's basically this exponential. Now, going from that same citation, the Australian Department of Health, the basic reproduction number, R0, is contrasted to the effective reproduction number, where there is some immunity or vaccination or some intervention measures are in place.

So while some people say this colloquially, at least by that definition, you can't reduce the basic reproduction number by human interventions like vaccination or quarantine or lockdown. That reduced number would be the effective reproduction number after human intervention. And so the R-naught is supposed to refer to what happens when a disease is tearing through a population. There's no natural immunity or vaccination against it, sort of like we have with COVID-19. Thus, we call it the basic reproduction number.

Even that basic reproduction number, though it's not fully constant, it can be modified by things like environmental conditions. For example, people think heat could theoretically make the virus break down outside the human body more quickly. That's true for certainly other kinds of microbes. So it's not a universal constant, even though R0. Now, with that said, we kind of care about the overall reproduction number, the effective reproduction number. So what can we do to reduce that?

So even before we have vaccinations, we can do things like canceling large events. You know, people say viruses love crowds. They also love choke points like door handles, bathroom faucets, things that lots of people touch. And insofar as we're advocating, you know, campaigns of handwashing, sanitizers, hand sanitizer use, face masks and so on, we are reducing the effect of reproduction a number of times.

Now, one important point though is that while we can tell that COVID-19 appears to be highly contagious, the exact reproduction number, whether basic or effective, is difficult to estimate precisely and has wide error bars. And let me give you an example. With Facebook, they can go to their database and they can poll every single person who invited every other person.

Right. So they can find this person went and invited 73 people and it happened at these specific times. They can pull that out of their database from an event that happened in 2007 or 2009 in September. Right. But to do that for a virus, you don't have that data. We aren't recording the data. You don't know exactly when it was spread. So you have to kind of infer it. And so it's necessarily kind of statistical. With that said, we have some very new tools, which are really interesting. So something Trevor Bedford has shown, this is the first time I think we're seeing like

public viral bioinformatics on social media. Trevor Bedford is a prof at UW. He's @TRVRB on Twitter. And his lab is something called Next, he has a project called Nextstrain, where essentially they've been, they've got a web interface where they've been sequencing viruses of different kinds of things and looking at infectious diseases and tracking them over time. And here's the awesome thing. In 2020, it is possible to sequence a virus

Get that into GenBank, which is like a database for all of these things. Analyze that, post the data online, and then post about it on social media and have that all happen in sub-24 hours. The cool thing about what Trevor's analysis showed is he was able to take two viral sequences, one from a patient around January 19th and one from a patient on February 29th, and

And he could show that it was very likely that the viral strain on February 29th descended or was closely related to a common ancestor of the one on January 19th.

And the reason that's important is it showed that there was likely a cluster of spread happening in Washington that was not contained. The alternative scenario would have been the patient on February 29th had a strain that was related to someone, to a Chinese strain, as opposed to a strain that had mutated and diverged. And every time it sort of infects somebody new, it mutates. But my understanding, correct me here, is like there's a core fundamental that doesn't change between them, which is what makes a vaccine possible.

Well, so actually, this is a sophisticated question. I would say it's not true that every time it infects somebody it mutates, but it is true that because viruses are like replicating so frequently, then this virus is actually quite mutable. So it does accumulate a lot of mutations. But you're right that there have to be certain conserved aspects of biology or else the symptoms start to change, right?

Sometimes, often, viruses evolve to be less lethal and more contagious because if it's like Ebola and somebody dies in a pool of blood, then they don't spread it very far. The Ebola virus actually sort of kills people before they can spread it.

But for this one, it's not obvious that it evolves to become less lethal because it's actually already super contagious and it's asymptomatic for a long period of time. So once you spread it, it doesn't care about you that much and might kill you or mess you up pretty bad. So that's bad because it's more like Spanish flu where it's super contagious

and it packs a punch, as opposed to being super contagious like swine flu or extremely lethal like Ebola. It's nowhere near as lethal as Ebola, but its scale is likely to be much higher than Ebola. And so the total number of people hospitalized or killed by it may be much larger. Are those other common variables associated with sort of

viruses like hospitalization rate, death rate, what are the other variables in that sort of equation that are the common ones people talk about? - Right, so there's the R naught, which does kind of influence your total number of infected, right, the cumulative infected.

There's hospitalization rate, there's death rate. You know, one thing we don't know, by the way, is what the long-term effects of this thing are. For example, you know, polio didn't kill everybody, but it did put, you know, some people, it paralyzed them for life, even though they survived, right? FDR survived, but, you know, he was paralyzed. We don't yet know what the long-term effects of this are. I don't think it's something which, something that's so severe as to cause death in otherwise healthy people within a few weeks. The folks who have severe conditions,

may have long-term impact that we're not aware of, right? Because it's so new that we don't know what that long-term impact is. Maybe they have impaired breathing. I'm seeing new reports that they, you know, and this is a may, I haven't looked at all the literature yet, so I'm saying may, but it may have Chinese social reporting that may have neurological effects, seems to maybe attack the nervous system and the immune system, not just the respiratory system. So severity is a continuum. It's not just dead or, you know, like bounce back. It's probably something you don't want to get.

I mean, there's lots of other variables that people look at. So this is kind of a derivative of other things, but the doubling rate, you know, how fast do cases double? And this one looks like on the error of 7.4 days. Then there's various intermediate time constants, like the time from when you're infected to showing symptoms, the time from when you're showing symptoms to when you die, if you die, right? Because certainly not everybody dies from this.

So like those vary a lot. WHO's recent report was reporting like one to 14 days from infections showing symptoms. And for those who do pass away, like two to eight weeks from symptoms to death. That's a large, you know, range. You know, those are some of the important parameters off the top of my head for thinking about this from an epidemiological or mathematical modeling standpoint. That strikes me as really interesting that you could possibly have a 14-day incubation period with no symptoms, but you're...

You're carrying it and I assume that you're passing it on to other people during that time as well. Yeah, so this is the really tricky part about this one. There's a lot of reports of asymptomatic spreading and that's really dangerous because if you're asymptomatic, you don't even... You don't even know. You're asymptomatic. Yeah, you don't even know, right? Here's the one good thing about being asymptomatic is...

It means that there are still some people who can go out and do things, right? But it may turn out that we need to test a very, very large number of people because there may be more asymptomatic spreaders than we think. And it may turn out that we need to make testing extremely routine.

which would basically mainstream genomics. You know, everybody would have, you know, right. Like the sequencing all the time, sort of like wearables, you know, your, your, your, your heart rate and your steps and so on are being tracked continuously. Sounds sci-fi, but it's actually not impossible to have, for example, a home sequencer that's plugged into your plumbing. So you spin the sink. This may not be the form factor. There may be other form factors. Okay. But you spin the sink or, or you, you know, put a sample into this somehow and,

and it does automated sample prep and it looks at what's there, right? Or maybe it's plugged into the sewage system or something. There's a ton of technical details to work out on something like that. It's a very high level kind of concept. Sample prep is important, distinguishing different people is important. But in theory, from that organic material, you could do sample prep and then you could basically do what's called environmental sequencing to determine all the microbes and viruses that are in that, right? And then you'd see that on your phone and you'd see if you were getting sick

or someone in your household is getting sick. And I kind of think something like that is going to be the types of things people will build both to measure and contain this and a sort of a hygiene 2.0 measure in the future.

So this will accelerate some sort of positive development around that. Is there, um, the test kits that they're, they're using now, are they real time? Like, are they, no, will you be able to go to the airport before you board a plane, like spit in something? No, they're not. Even though it's called real time PCR, it's not, it's not that fast. It's, um, the real time aspect refers to something different, which is a little bit, a little bit technical. It's a, you know,

you're basically looking at the ratios of curves you're not going to get it on the spot that like there's a different term for that it's like point of care diagnostics you're thinking of like like a pregnancy test or something and i do think we want to get this stuff into a we're actually point of care is a little different from pregnancy pregnancy test is the ultimate where you can do it at home but ideally you want to have something where the point of care is at home and you can kind of test yourself and plug into your smartphone and we're not there yet with this

Yet, I think we're going to have to get there rapidly. And there's potential tools that could be used for that.

you know, there's also proxies. You don't necessarily have to go to the molecular level right away. You can look at something like an Oura ring or even a thermometer and just measure your temperature. Right. And, uh, and so then that's symptomatic. It's not going to detect the asymptomatic stuff, but it's, it's maybe a cheaper and more widely deployable proxy because thermometers are well understood. Wearables are well understood. That's relatively easy to deploy. And I think the who reports that like 89% of people that were infected, uh,

had a fever. Is that somewhere around there? Yeah, I don't know. I don't remember the exact stat off the top of my head, but yes, it was incredibly high. Like it was. Yeah, that's right. Like, like basically it is, it's the most characteristic symptom. Right. And that's why, you know, in China folks are asking drivers to record their temperature and so on. So, um,

Oura rings or things like that. I mean, I'm not an investor in Oura or anything like that, but I know that it's pretty cool where it does monitor your body temperature all the time. Something like that, I think, will become very mainstream. You can see the variation. Yeah, that's right. Actually, there's another study which I retweeted a while back, which was wearables in general

seemingly give diagnostic rate information where you can predict the onset of illness from them. There's been papers published on that. So you might get farther with wearables and you might think you might need to go to so-called molecular diagnostics immediately. Are we going to a world where you can have a company issued or a ring or Apple phone and they basically say, don't go into the office today because we predict that you're likely contagious?

Oh, absolutely. Not with it, not with this particular virus, but with any virus in the future. Absolutely. I think this is going to, so let me give the silver lining to this whole thing, because this is a dark cloud. Okay. There's a lot of people, you know, we're just at the beginning of this potentially, you know, there's a lot of people ill, a lot of people infected. So I don't mean what I'm about to say in a casual way at all.

But the silver lining of this is this could be as much an accelerant for biomedicine as the internet was for software. For example, if you think about this from a funnel standpoint, every aspect of the system, from diagnostics to...

treatment to the analytics on the population, you need to innovate. I'll give five or six different things. Okay, we're going to want at-home diagnostics. We're going to want stuff that plugs into your smartphone. We're going to want wearables that give continuous sensing. We're going to want pandemic surveillance, like stuff that's monitoring effluent and sewer systems, both at home and at

and elsewhere. We're going to want, if I didn't mention telemedicine, we need that. We're going to want medical delivery robots so that people who are really contagious can kind of be in an isolation chamber and the delivery robots can deliver them the drugs without endangering doctors or nurses. We're going to need much faster clinical trials for drugs and vaccines. We're going to need like these instant hospitals that people have built. You know, we need to be able to add medical capacity in like a surge, you know, form to deal with this type of thing.

I mean, there's so, so, so much. You know, I think at the end of the day, what happens is, you know, an anti-missile defense system, when there's an incoming missile, it has to react so fast, it has to react at the speed of the missile, right? The human being involved may be too slow. So it has to track the missile, launch the counter missile and hit a bullet with a bullet to knock it out of the air and blow it up, right? Like the Iron Dome system in Israel.

A pandemic defense system, you know, the kind of thing the Chinese will build, I call this the, you know, the third great wall. And then somebody actually used that term in China as well, like, you know, a few weeks later. There's a great wall of China, the great firewall. And I think we're going to see the great bio wall of China, okay, where, you know, they'll be sequencing and testing the heck out of everybody, scanning the country constantly to see, okay, is a virus rising?

And then quarantine that area, vaccinate that area, you know, and just stop it from spreading. Because this is such an enormous scar on society. No one's going to forget this, that I think it's going to be a significant part of their, like, defense budget going forward.

And that's just the stuff I can see. Everything I mentioned there has been kind of happening like two months ago in China. You know, like, for example, the telemedicine hospitals with delivery robots, they built that in January, right? Like, they've been at a wartime pace of biomedical innovation since January, which is part of what got me, like, really interested in this because, I mean, you know,

The Chinese government and the Chinese people are serious people. They wouldn't commit economic suicide, you know, as my friend slash Twitter friend John Stokes said, you know, like to simply shut down their economy for nothing is not characteristic of them. So I took this seriously based on that, like,

extremely indisputable signal of reality. And the technological innovation that's kind of there has been amazing. And I think we're going to need to mirror that and learn from that in the West in order to deal with this. What would be the second order sort of consequences of a bio wall around a country, not specifically China, but just if like we do a thought experiment here and each country is now isolating itself from other countries and

the spread of disease maybe gets reduced, but does like human resiliency go down? Like what are the sort of consequences you could foresee? Well, so for one thing, before you travel to certain countries, they want to know that you've had this vaccination or that vaccination, right? You've probably seen that before, right? Yeah. Now that's going to go to real time.

Meaning, you know, within a day or an hour before you get on the flight, you know, imagine a next generation TSA that was actually competent, you know, in countries around the world, which needs your biological information before you can get on a flight or enter the country.

As in like you gave a blood sample or saliva or something? Yeah, that's right. I think that's where countries are going to go because, um, you know, screening is already happening for every traveler in India coming in. Like that's just, that just went live.

And so as countries react to this, they are going to require diagnostic testing. Just like you have to make sure you're bringing no guns on a plane or bombs on a plane, right? That could kill hundreds. They want to make sure you're not bringing a deadly virus into the country that could kill thousands or millions.

And so I think that kind of screening is going to happen. It's happening already in the sense of checkpoints within China, but that's going to be rolled out. And in a sense, it's just the generalization of, you know, OK, you need X and Y and Z vaccine in order to enter the country. Right. Like Singapore for a long time has required evidence that you've had yellow fever vaccine before you can enter the country. Right. So in a sense, it's not like completely unprecedented. Right.

But the scale of it is going to be totally new. And so this is going to be a gigantic shot in the arm for every diagnostic company in the world, every medical imaging company. All of those are about to get like defense budget level spending because you can think of this as a world war in terms of its scope.

But it's a world war against the virus. Do you think COVID-19 is the straw that sort of like pushes this over or the piece of sand, I guess, that piles on and topples the sand pile over? Well, I think it's definitely going to cause at least a recession, but likely something more severe than that. It's one of those things where there's so many secondary shocks. If you remember like the 2004 earthquake in the Indian Ocean, that led to a tsunami that devastated many countries in the region.

and and that's kind of how i think about this uh there's there's the badness of the virus itself and there's the second order kinds of things that arise from it for example the supply chain shocks right so you have you know all these people in china who can't go to work they can't go to work they can't produce you know maybe the screws for your to maintain your radiator or the you know even more dangerously like necessary components reagents for drugs right because supply chains are really complicated people don't

those to how many different parts go into, you know, like a printer or

or even a laptop for sure. Or even like a container of milk. You've got the label and the ink and the dye. Is that the disease causing that or is that the reaction of people? Like the flu infects a decent number of people every year and it may have a lower sort of death rate, but it's still very transmissible and it doesn't cause people to shut down factories or... True.

But the thing is that, you know, this analogy to the flu is very bad. I'm in no way attacking you. Let's deep dive on this. Yeah. No, no. I want to deep dive on this because it's the most common sort of message out there in the media. Like, why should we care about this more than the flu? You know, the flu is going to affect 40% of people over the next five years. I'm just making that number up, but it's going to affect a decent number of people over the next five years. And, you know,

We do have vaccines against it that are sort of like reasonably effective. Healthcare practitioners are trained to identify and treat. And like, so why is this? Well, we have to make distinctions, right? Like, you know, a common cold is not HIV, right? You know, there's infectious diseases. There's infectious diseases.

And so, analogizing it to something which is also an infectious disease is not helpful in this case. There is a similarity in the sense of it spreads like the flu, like strains A and B of influenza, in the sense that people are, it seems to be airborne, right? Which is, it spreads like the flu. People are coughing and what have you. But compared to the flu in terms of severity is not the case.

And there's a graph which I might be able to share with you. But wait, can I push back on that for a second? Like, how do we know that? Because I'm assuming, and I'm totally naive on this, so like I'm an idiot, but I'm assuming the number of infected people are astronomically higher than what's reported. Totally. And I've got a great one line. But if the death rate is accurate, wouldn't that reduce...

Like if the number of deaths as a result of this is reasonably accurate, wouldn't the death rate actually go dramatically down when you assume the number of people that would possibly be infected? Here's my short answer to this, which I actually tweeted a while back, but now I think it applies to like Iran and Italy and so on. Okay.

Wuhan was a normal city on December 1st of last year. And by January 23rd, all seven hospitals were filled. People were being sent home to die. The entire city was put under quarantine. And it was, you know, the largest, like, you know, quarantine in human history began.

That was not a bad flu season. So is that Chinese reaction or is that like a necessity? Like, do you think that... Well, here's the thing. Like, you don't... I mean, it's one of those things where...

First of all, Iran seems to have had a similar experience where – I don't know if you've seen this guy on Twitter, Ali Ostaad, and his tweeting from Iran or this guy, Del Luca Anna. He's giving stats from Italy and people are showing the number of beds. This is actually on Wikipedia from the Italian government, increasing exponentially. In the first week, the region of Italy that's been hit by this announced they had run out of beds. Okay?

When you're seeing things from China, Italy, and Iran, I don't think this is psychosomatic. I don't think it's a lot of people who are hypochondriacs going to the hospital. Like, it's a hospital, right? It's expensive. You know, even if you have universal health care or what have you, and, you know, I don't think that's – it's not a trivial decision to go to the hospital.

And and so so I think that in very different societies now, we have to kind of take the warning at heat. It just it's sort of beggars imagination to believe that the Chinese government would have 760 million people under lockdown and quarantine because they're scared of a sniffle. That's just not what this is.

This is simply not the most parsimonious explanation. You know, you can sort of wait for the storm to hit somebody before they'll listen to that. And some people will be like, I'm not saying you are, right? But all the evidence we have indicates that the absolute number of hospitalizations and deaths when this hits an area is very high. Even if it is true there's lots and lots of mild cases, the N is so large that it overwhelms the medical system.

And the number of people requiring intensive care. Exactly. The number of available beds, which is further. And how does it transmit? So it's airborne, obviously, but it's also surfaces. Well, yeah. So, I mean, it seems highly contagious. So, you know, it being airborne, which really sucks because you should wash your hands. You should avoid handshakes.

but that won't necessarily save you because it's if it's airborne right it appears that fecal transmission is another vector and that was something that happened with SARS which is a related virus you know 17 years ago but there's a famous infamous case of a

apartment complex in Hong Kong called Amoy Gardens, and about 300 people there contracted SARS because it gave some of them diarrhea. And this is gross, but some of them flushed and the sewage system wasn't perfect, and some of the droplets or whatever were coming back up. And so people were all infecting each other within the building. That's a more general thing, which is apartments are not biohazard containment units.

For example, if you think about your soundproofing in an apartment, it's not perfect. You can sometimes hear a thumping bass or whatever. It hasn't been tested for military-grade soundproofing or whatever the ultimate level is, anechoic chamber kind of soundproofing. In the same way, yeah, it's probably true that you're going to be less communicable in your apartment, but you're still connected to the same air ducts and your water supply is still the same and people are pushing the same elevator buttons. So

It's not 100%. You said you had some numbers before. Oh, yeah, sorry. I digress a little bit because we were talking about fecal transmission. Yeah, yeah. I gave the Amoy Gardens example, right? And so the reason I just gave that example is fecal transmission, most people will need to say, well, I'm not, you know. I'm not eating anybody's poop. Exactly. But unfortunately, in Amoy Gardens, people were. Mm-hmm.

that you realize it, number one. And number two, in many cities on the coast of the US, like San Francisco most infamously, people are definitely inhaling fecal matter as they're walking outside. They're definitely getting it on their shoes and so on and tracking it in, even in small quantities or whatever, right?

And that is something where if you have one infected person who's doing that, there's a guy, a professor from UC Berkeley who warned about this two years ago, actually, in an NBC report. He's like, you know, the streets of San Francisco are at a developing world level in terms of, you know, public health, right? So the warning's evolving there, but it's sort of like the city is kind of an immunocompromised city.

San Francisco in particular. And it's going to be pretty dangerous if and when. I think it may have already happened. We'll see. If uncontrolled community spread is happening in SF, like it seems to have been happening in Washington, Washington State, that is. Yeah, that won't be good. Why are kids more resilient than adults? Like it seems to almost...

to age, but younger is least affected to older being the most affected. Is that accurate? I mean, you know, one thing with all of these, these types of things is, um,

These are emergency conditions right now. So it is good that we have demographic crosstabs, right? We can see that, you know, the old or the elderly seem to be more affected, much more affected, especially as you get to 70 and 80-year-olds. And we can see that the young appear to be less affected. We can see it seems to hit men harder than women. But those are kind of macro-scale demographic observations. The molecular biology of that, I think, is probably still a hypothesis. Hmm.

Because it could be one of a bunch of different things biologically that are different between a kid and somebody older. It could be a bunch of things biologically that are different between women and men that are causing this. And I don't think we know the molecular biology of it. What are some of the second order impacts of what's happening now? So like

Around here, I'm assuming everywhere around the world at this point Purell is sold out. You know what's really interesting to me is I was at the store the other day and I was trying to get some Purell like everybody else. Why did you go to the store? Why didn't you go online? Because it's like 300 bucks on Amazon right now. Right.

So I think it's I was thinking about how retail everybody's like, oh, retail is moving quickly and they're catching up on this technology thing. And Purell was on sale everywhere because some guy like six months ago decided that they would have a sale this week on Purell. So not only is it out of stock and hugely in demand, it's on sale like two for one at tons of stores.

Anyway, that was just my retail observation. But what are some of the second order consequences of hand sanitizer use? Like everybody's just massively going to use hand sanitizer now. Well, I mean, some people have said, oh, you know, you're just going to get, you know, microbes that are resistant to hand sanitizer. That may be true. Probably true.

I think, you know, so the second or constants that I'm thinking about are at a somewhat different scale. So here's a few of them that I'm thinking about. Oh, I want to go through all of them. Yeah, yeah. Sure. So I mean, to downplay that, I do think it's possible that likely, in fact, that you're going to get microbes that are eventually resistant to hand sanitizer if it's widely used, for sure.

Still, let's use the ammo we have now and then come up with more ammo later for something like this. I think that's probably reasonable. So the kind of second order consequences I'm thinking of, here's a few of them. Number one, the supply chain shock, right? Which is to say all these places in China have just shut down. Lots of things are made in China. All these small companies in the US that depend upon Chinese parts, all these hardware companies. I mean, hardware companies...

are just going to die, many of them, because they just can't get the parts from Chinese warehouses. It's funny to use the reverse analogy. If Amazon Web Services went down,

A lot of sites would just go down. They just do not have a backup option. And this is like that, the physical cloud, so to speak, to kind of use a digital analogy to explain the physical role. So one, supply chain shock. Number two, travel shock. Airline travel, conferences, events, rallies, sports games, etc.

Gone, right? Like canceled in lots and lots of countries. And maybe for an extended period of time, depending on how bad the outbreak gets. Do you think they go away or do you think they switch to like conferences switch to virtual? Do you think like sporting all switches to TV now? Yeah, I think, well, I think what happens is, you know, historians of the future may write something like 2020 was the year that the internet actually began.

Right. Where, you know, I have this concept of the primary in the mirror. So in the mid 90s, when the New York Times first went online, the New York Times dot com website was just a tiny mirror and the physical paper was primary. There's a few articles that were online. Right.

Then over time, the website bore more and more and more and more of the load. And I don't think there's any particular announcement, but certainly today you would say NYTimes.com is a primary and the paper, the physical paper is just a printout at a particular time stamp. That's the mirror.

Right. So it gradually shifted from the physical as primary and the digital as a mirror to the reverse, where the physical is just a printout of the website. Right. Makes sense. And in fact, there's certain things that are digital native that you couldn't print out, like some of the interactive graphics or or links to tweets, embedded tweets. I mean, it's harder to print that out. Right. Right. It doesn't make sense. The data science exploration. So you start going digital native.

I think that this year is going to be – it's kind of like during World War II, many women entered the workforce. And then even after World War II, a new normal had been reached. This is what's called hysteresis in physics. It's like you apply a stimulus to a system. Then even after you remove that stimulus, the system is in a different state.

than it was before. There's some memory in the system, right? So what this is going to do is it's going to force everybody to look at remote work as not an auxiliary, but as a primary. This is one of the grand challenges, I think, for tech as opposed to biotech to work on. And it's going to have its hands full building what I call the remote economy, not just remote work. Like every job has to be remote capable, not just, you know, like,

legal work and programming and graphic design and all that type of stuff. We need telemedicine, we need autonomous delivery, we need people to be able to operate forklifts from a factory, all that type of stuff. It sounds like you're saying this isn't going to cause it, but it's going to accelerate these natural sort of trends or things we've been dabbling with because there'll be a huge health benefit or a perceived health benefit in the future.

to having things operate this. But then what are the second order consequences of that? Like social isolation? We need to feel part of something and part of society and connected to our community. What are the psychological impacts if everybody worked remotely? So what I think happens is you get two compliments that all the things that are canceled, people are going to want to bring back or bring back in some other form. There's a couple of options for that. First

is work from home is going to get really good. Things like tandem office, tandem chat, and so on. People are going to try, there's a hundred different things that you could argue are the reasons that people aren't as productive at home. And people are going to push that. And VR is an obvious

There's a bunch of other obvious angles to try. Maybe it's like a Facebook portal-like thing. The tricky part is anything that's hardware related may be tough because supply chains will be hit. So there might be demand for the VR headsets, but they may not have the ability to supply them, which would suck for the VR companies. They may figure out some workaround, though.

So one aspect is that remote work is going to become really good. The second thing is people are going to lean harder into social networks and eSports and all of these online complements or proxies for offline socialization. A third aspect, I think, is that people are still going to want meat space, you know, or physical space. I think you're going to start to see trusted communities where

where you're literally trusting the other person. You have to have very high trust that they are taking all kinds of hygienic measures. Right, so it's like a bio trust. Yeah, it's a bio trust, right? So that's a very high bar. Like, do you trust this person to take such good care of their own hygiene and health and to be so diligent about

that they're not going to put you and your loved ones in danger, right? It's a pretty high bar. And so, you know, I think that what you start to get is, especially if people are locked at home for a while,

I mean, all this homeschooling stuff, right? Basically, the educational system has been completely transformed with all of these children sent home and there's all these parents there. So I think you're going to have these community organizations form where a group of folks who know each other to not be infected or trust each other to maybe trade off and then they start educating the children amongst themselves.

We're using online tools and each parent has the kids for a day or something like that. Right.

This is in a weird way like a forced de-globalization, radical forced de-globalization because all the things that were benefits like economies of scale, large groups of people, big crowds and so on are now big demerits at least in real life. And so force is very small scale in the physical world and I think you're going to get scale in the digital world as a compliment to that until we get vaccines and other kinds of things and at least aspects of life come back to normal. Right. That makes a lot of sense.

What are sort of some of the other second order consequences you see to the economy, both positive and negative to the whole COVID-19? And one thing that I do want to point out that I like is that

a positive, you know, is that governments are handling this very differently and that will enable us to see how different things play out over time and get feedback on like how we could handle the next one better, I'm assuming. Well, yeah, I mean, I think of this as like the decathlon for governments because, you

You can't be ideological about it. You might need to, for example, you might need to mix a border shutdown and quarantine with universal health care for the vaccine with accelerated technological capitalism to push out the telemedicine and the delivery robots and so on. Right. So this doesn't fit neatly into ideological categories. It's just about competence, you know.

And so who's winning that international decathlon right now? Probably Singapore and Israel, right? I haven't looked at what Estonia is doing, but I wouldn't be surprised if they're very much ahead of the game. So what are they doing differently? What are they doing differently? Well, I mean Singapore's leader, Lee Hsien Loong, is Lee Kuan Yew's son. He's not just like good at math.

He is, you know, he was the wrangler at Cambridge. That means, you know, senior wrangler is a guy who has the top mathematics undergrad, right? Right. Who has a CS degree. And in fact, he has posted like Sudoku solvers on Facebook in either C or C++. And like, he still actually knows code, right? So this is somebody who has like a technical background.

In Estonia as well, they've got a very strong computer science engineering influence in their government because they kind of decided in the early 90s to become an internet country around the time they got independence. And so the critical point here is it's not so much that, oh, the prime minister needs to code the response themselves. That's not what I'm saying at all. What I am saying is there's a critical mass of people with scientific and technical expertise in government and politics and the media and so on in these societies who

And those folks are capable of doing what Tyler Cohen calls growth rate extrapolation rather than base rate extrapolation. You see Tyler's post on this? Yeah, walk the audience through the differences between growth rate extrapolation and base rate extrapolation. Sure. So very roughly, and this is maybe a little different than how Tyler himself phrase it, growth rate people model how the world can change. Base rate people think about how the world is basically going to stay the same.

And so the growth rate people are often mathematical. They're used to exponential growth for things like iPhones or Facebook or Bitcoin. They're looking for large deviations. They're often investing money. They're quantitative, numerical, they're technical, they're scientific.

Base rate people are, you know, and they're often correct in the default that, oh, things aren't going to change. I mean, an institution is almost by definition a base rate thing because it's kind of, it stayed at STAID, right? Yeah.

But basically people are – they're about the conventional wisdom. They like mainstream things. They think, oh, that's weird when you point them to something which is at 0.01%. In a sense, anything that's below 50% of the population or a large enough thing that doesn't have a constituency, they kind of ignore because it's just not important enough to matter.

The problem is for them, I think, is if something is growing exponentially, then it's 1% and it's 15% and then it's 72% and then it's 100%. Like, you know, pretty much everybody in media, for example, has a Twitter account and just goes like this really fast before they have a chance to react to it.

And so I think the issue is that most of the people in politics are, you know, they're lawyers or they are, you know, career politicians. There's really a deficit of folks with technical engineering degrees. And the interesting thing is, you know, I was able to reframe this and think about this to myself.

I actually think that's, you know, on us in the sense of folks who have any form of technical or scientific background, I think need to, as like a civic duty, and this is, you know, both for this crisis and then for future kinds of things, we have to do journalism, citizen journalism, like a year of it, for example, per person, something like that.

We have to actually get involved in politics and things like this because otherwise there's going to be a deficit of folks with that training. Because one of the drivers of this is you can definitely make a lot more money outside of politics and journalism and whatnot by if you have a technical background by going to technology or to Wall Street. And that's definitely one of the factors.

But what that's meant is folks within those domains are making decisions without an intuition for math and an intuition for science. And you can't teach them all of that in, you know, the instant that they need to understand it and grok it to deal with the pandemic. Whereas Singapore sort of like takes the top students and then tries to fast track government, right? That's right. So there's more shared knowledge, right? If I say, you know, oh,

I can explain the R-naught and, you know, the Singaporean government, I don't have to explain to them. They understand what that is, right? Even if they hadn't heard it before, they've done enough spreadsheets or what have you that they can understand the implications of that. I'm not sure that, you know, our government in the U.S. is selected on that basis. What's your opinion on how should governments respond?

What should governments be doing? What should they do? This is a great question. So what I've been tweeting about and one of the things I'm writing up is something I call the decentralized response. So there's a centralized response and a decentralized response. So China's response is centralized. The Chinese government actually has a critical mass of folks with engineering degrees. I think Xi Jinping has a chemical engineering degree from Jinghua and Hu Jintao's predecessor, I believe, has a hydroelectric engineering degree also from Jinghua.

So the Chinese government has centralized response, which is quarantine things, you know, do basically surveillance in the sense of both pandemic surveillance and contact tracing. And the government was basically driving everything. You know, certainly like lots of private citizens were doing things, lots of companies were asked to do things. The government was kind of in the driver's seat.

I think in the US, that's not working. The centralized response has been very poor. In particular, FDA and CDC and HHS have not been doing a wonderful job. And that's because, I don't know if you followed this whole thing, HHS declared a public health emergency, which you would think would push things through radically. It didn't. Instead, it meant that there was now a higher bar

to get diagnostic tests out because FDA had so-called emergency use authorizations that had to grant to clinical labs before they could test. Net-net, a bureaucratic process inhibited the United States from testing for a critical 30-day-ish period from January to February and gave people a false sense of security that there weren't that many coronavirus cases in the U.S.,

So the centralized response so far has been a total, I should say, a total failure, mostly a failure. The decentralized response, though, I think has shown promise. So there's these folks at UW who have developed a diagnostic. There's that professor I mentioned who has estimated the number of cases. There's also the fact that the first patient who was written up in the New England Journal of Medicine in January was treated under compassionate use, which I believe used the more recent right-to-try pathway approach.

I need to double check this and verify this, but there's no mention of FDA approval in Gilead statement. They talked about local regulators approving it. And this is kind of a wrinkle, by the way, just to explain this for a second. So not everything in healthcare is FDA approved. Clinical labs, for example, are regulated by another entity called CMS, the Center for Medicare and Medicaid Services, under a program called CLIA. But CMS and FDA are under HHS Health and Human Services, which is a parent agency.

And in CMS, the CLEAR program for laboratories works with local state agencies to regulate lab tests. The way it works is that the lab itself gets cleared by a state agency like, say, the California Department of Health.

Then any so-called homebrew test or LDT, which stands for laboratory developed test, the laboratory itself develops or creates, can be shipped based on the lab director's approval. Now, of course, they need to prepare copious reports on the sensitivity and specificity of the test to develop the appropriate controls, all those things. And that lab director has to have a doctoral degree and maintain board certification. This has to be a qualified person. But in general, overall, you have far more flexibility to ship and update a lab test than a traditional medical device in a box.

Now, the FDA hasn't liked this because it's a path that's been outside them. For more than a decade, they've contended they have something called enforcement discretion, meaning they have just chosen not to enforce the law against laboratory-developed tests, but reserve the right to do so.

And they've constantly been seeking to bring lab tests under FDA's purview for so-called pre-market approval, which means among other things, huge time and cost to bring the test to market. And that FDA reviews every change you make to the test. If there's software involved, which there is for many modern assays, every Git commit would have to go through something called design control, which is like the sort of bureaucratic process around revisions to a medical device.

Opinions differ on this, but many lab directors believe that adding pre-market review on top of existing LDT regulations really slows things down. It adds a lot of paperwork, doesn't add much value. And so right now, those lab directors, usually they're regulated by this more decentralized system for lab disapproval called CLIA. It's a system outside the FDA that the FDA doesn't like.

Now, once Health and Human Services declared a public health emergency over this virus, FDA actually gained a new power, the power to grant so-called emergency use authorizations.

The idea is that it's an emergency seat, and so an emergency is "high risk". You need a lot more review for the test, and that means FDA needs to slow things down. It was the denial of these emergency use authorizations over the critical month of February that forced all the labs to keep their hands tied. Basically, all our planes were sitting on the ground while we got Pearl Harbor by the virus.

And why did FDA do this? Because they saw an opportunity to win ground in a bureaucratic turf war, to never let a crisis go to waste, to use the EUA to force lab directors to send tests through them. And FDA has a deep-seated belief that they're the only good actor in medicine and biomedicine, that anything outside them isn't the highest standard. At a cultural level, they basically don't care about the cost or time they impose on others for this ostensible high standard.

And nor do they think that significant pre-market review and bureaucracy may actually reduce quality overall due to less ability to iterate and far higher startup costs, which could lead to less competition, innovation, and substantial delays.

Now that's an argument that's been going back and forth for many years, but at least now the American public is seeing that in this case, the cost of FDA delays in test approval may be measured in needless hospitalizations and deaths. Because under this emergency use authorization system, lab test approval was changed from the more decentralized system under CLIA to a centralized system where FDA positioned themselves as a bottleneck for the EUA.

and no labs could get approval other than CDC itself. - Okay, and CDC itself was doing 12 tests a day, and that basically meant for a country of 300 million people that's testing 12 a day, that's a centralized bottleneck. What happened with this case in Washington is interesting, which is a different example, decentralized not diagnostics, but decentralized drug prescription.

So typically to get access to a drug, it has to go through an FDA process, but there are other ways to do it. One way is so-called off-label prescription, where a drug that's approved for purpose A, a doctor can prescribe it for purpose B, even if it hasn't been approved by the FDA for that purpose. However, the drug company historically has not been able to tell the doctor, hey, you can prescribe it for purpose B, because that would be considered so-called off-label marketing.

And so that's a weird thing, right? A drug company cannot tell a doctor of a true fact

because it would be considered, you know, like promotion for a purpose that the FDA hadn't approved, right? That was basically most very recently in 2018, I believe, rejected on free speech grounds in U.S. versus Corona. And so like FDA's power to kind of centralize drugs kind of dropped a little bit there. Another aspect which is quite relevant to us here is these right to try laws that were passed recently, and I think in about 41 states,

mean that when someone's at death's door, you no longer need to pursue compassionate use going through the FDA. You can get, I believe, a local state, and the laws vary in different states, but a local state regulator to sign off on it and let it go through. And I believe that's what happened here with the remdesivir prescription. So that's another example of the decentralized response, not of diagnostics, but of drugs. Is that clear? Should I tighten that up or is that helpful?

I think it's clear. I mean, I just don't understand a lot about that sort of world. So, I mean, it made sense as an explanation to me. I just don't know the sort of like the context. Sure. Yeah. Going back to sort of like how should governments respond? Like what is the decentralized response? What is the right? What is it? And then what on the individual level can we do? So I think the single most important thing that the government should do is emergency expanded right to try laws. Right.

And what that would do is it would unblock. So we've already seen what happens with the diagnostics bottleneck, right? Making it something where it had to go through FDA approval and also the CDC's case definition. The FDA was blocking tests at the physical level and the case definition, which said only people who had returned from China are eligible for COVID-19 testing. That twin kind of centralization bottleneck were two bad decisions that were a central point of failure that's led to the, you know,

epidemic or really, you know, in the sense of the U.S. is part of a pandemic that we have now. So when you can't yell at institutions, like institutions that made that bad a call, you can't bottleneck more stuff through them.

Right. The alternative is do or one alternative, which I think we should do the most important thing we should do. Emergency expanded right to try, which gives every state the ability to clear anything related to the coronavirus. And so it's just approved at the state level rather than being boosted up to the federal level.

There's several reasons to do this. First is to remove some bureaucracy. Correct. That's right. And basically decentralize it. The reason I think this is a good idea is several fold. First, if this is pandemic, you know, when you have 50 states that are dealing with 50 epidemics, you just simply cannot bottleneck that through Washington, D.C. for approval on everything. Just doesn't work. Number one.

Number two is this is already how lab tests are regulated and they've been pretty much fine. Number three, yes, it is possible that there are some things that are approved that don't work or even that are unsafe, which is less likely but possible. It's more likely that it's just kind of a placebo and doesn't work.

But in this case, the cost of delay is so high that it's okay to have some false positives, right? It's okay because folks will die otherwise. That's to say you'd rather have had an imperfect test. I mean, the exact numbers do matter. We'd rather have had an imperfect test during February than have had essentially almost no testing.

because at least we would have realized the potential scale of it, right? With an imperfect test, there's things you can do to mitigate. You can do, if you know the false positive and false negative rate, you can sometimes do retesting. If the errors are random, you know, two tests, like, you know, I shouldn't say cancel each other out, but they'd compensate. You can use gold standards when necessary and do random resampling. An imperfect test is better than nothing for something like this. Not always, math matters, but frequently. That makes sense.

And so that emergency expanded right to try would give drugs diagnostics that are approved in states, I think would unblock a lot of stuff. And then information sharing would happen because if Washington state or Massachusetts or California or Texas or whatever figured out something, then that could be propagated to other states and you wouldn't have to bottleneck everything. - Okay, what else? Like should schools shut down? Should government shut down? Like how do you, like where do you draw the line on this? What's your opinion?

Great question. So there's some math to draw the line on it. So essentially one way of doing it, this is not the only way of phrasing it, but one way of doing it is for, given that you know a particular infection rate in the community, right? Let's say it's, you know, here's a good example. Let's say there's a hundred people in San Francisco who have the coronavirus and San Francisco's population is about 880,000 and you have a office that has a thousand people, right? And

What is the probability that one person in that office has the coronavirus? I have no idea, but you can walk me through it.

Yeah. So you can calculate that out and it's about 10%. So just to recap the numbers. So like 100 people in San Francisco that have the virus, population about 880,000, which is a thousand person gathering, right? The reason it's about 10% is even though the probability is very low, you have only 100 people who have the virus out of 880,000 in SF, right? The number of trials you're doing N equals a thousand is not very high, but pretty high.

And so you start to take a low probability event, but you try it a bunch of times, and then eventually you potentially get a spreader at the office, right? And the reason you might say, oh, well, isn't it paranoid to say one person with the coronavirus at the office will spread it to a bunch of people, right? That's the next follow-up question, right? Yeah. And the issue is there's choke points, right? So almost everybody touches the elevator button.

Right. Bathrooms, elevators, common areas like cafeterias. Doors, the fridge, you know, for food, right? If you had a security camera trained on it, right, you would see those things that 400 people touch.

And that's probably how if people go and, you know, these folks who've spread at conferences and so on, if you go back and run the security cameras and you look at exactly how it happened, I would not be surprised if it was something like that. You know, someone washes their face in the restroom, they turn the hand back, you know, they don't have to hand back the faucet handle back, right? And then 30 more people touch that, right? Ta-da, that's a great way to get super spreading. So...

So the answer to your question is we can use math to quantify this. You know, what's your risk? So just to go a little bit further, let's say there's now not 100 people in SF, but 1,000 people in SF who have the virus out of 880,000 odd people. And there's, again, 1,000 people in your office. Now it's more like, you know, more than 60%.

So I posted an infographic with this kind of calculation, so you can do it yourself if you want. The basic idea is that's the second major thing that people should be doing is they should be stopping large events, South by Southwest, Houston Rodeo, large conferences. What's a large event?

Is it like more than 10 people? Like, is it an international event? Like, is it an event, a local city event? Like what, define a large event. It's a good question. The answer is it depends upon the percentage infected in the people coming to the event. Which we don't know. You can, you don't know directly, but you can estimate, right? Now, if you use the analysis that Trevor Bedford had, right? You know, you see a bunch of people and he's able to estimate like a cluster of, you know, median size, like 570.

So 570 people in Washington have a case related to that cluster. There's other ways of estimating it. Based on 10 deaths in Washington state, there's probably thousands of undiagnosed cases out there. I can get into the math of that. So this is what I call Geiger counter mode. You can't just by feel tell whether a place is really radioactive or not. You need a Geiger counter.

And when you're dealing with an invisible threat, whether it's radiation or deadly viruses, you can't just kind of intuit your way to it and say, oh, that feels big to me or that feels small to me. You have to do it on the basis of, okay, what's the estimated infection rate? And then how likely am I to get it? Is there to be somebody at this that gets it? Plus, also, this is very important. We have empirical evidence of this. It's not paramount. Did you see what I've been tweeting on the parades in Philadelphia? Yeah.

Yeah. So, you know, we have at least two examples, one from basically 101 years ago, 102 years ago. That was the Philadelphia-St. Louis. Bingo. Exactly. Contrast between how they responded to the Spanish flu, where Philadelphia just sort of like went ahead with everything and St. Louis sort of like shut everything down. And it affected Philadelphia way more than St. Louis, correct? That's exactly right. That's exactly right. And the same thing actually happened in Wuhan for this virus two months ago. Right.

Right. Wuhan had a gathering of 40,000 families, which is a really big thing. Right. And that's a great way to just take something that was a viral fire and just make it go completely vertical. Hey, thank you so much. This was a fascinating conversation. Awesome. Cool.

Thank you for listening.