All Episodes
April 19, 2020 - Dark Horse - Weinstein & Heying
56:54
E08 - The Evolutionary Lens with Bret Weinstein & Heather Heying | The Stealth Ecology of SARS-CoV2 | DarkHorse Podcast

The eighth livestream from Bret Weinstein and Heather Heying in their continuing discussion surrounding the coronavirus. Link to the Q&A portion of this episode: https://youtu.be/UsJhDdm-kcMSupport the Show.

| Copy link to current segment

Time Text
Hey folks!
Welcome to the Dark Horse Podcast live stream.
This is what, our eighth?
It's our eighth.
This is our eighth live stream.
I am sitting with Dr. Heather Hying, as has become our tradition.
We have been fending off technical difficulties of many different kinds, and I would say at this point, aside from the audio, the video, and our connection to the internet, everything is going smoothly.
It is smooth sailing, other than those three trivial issues.
I wouldn't call them trivial, but you know, they're part of it.
All right, so we want to start with logistics on how we're going to be dealing with the Super Chats.
Yes, why don't you take lead on that?
We'll try.
So we have been receiving many excellent, and some not so excellent, but mostly excellent Super chat questions, but it has become a bit onerous and we are not, we're giving short shrift to too many of them because we have been trying to, we've been reading all of them.
So this time we are going to try another strategy, which is those super chat questions that come in in this first live stream where we are just talking to each other, We will look at over the break between the two live streams and prioritize them based on both monetary amount and then try to do a very, very first pass at, you know, what questions are particularly interesting.
And we'll spend a half an hour only on those questions from the first, that come in during this first live stream.
And if we don't get to all of them, which we probably won't, that's unfortunate.
And then we'll spend a half an hour on the Super Chat questions that come in during the next live stream, which we will probably have to take in as they come in.
So we're going to limit the second live stream, the Q&A, to an hour and see how that goes.
And this is a moving target.
We're going to see how this goes this time and maybe modify it next time.
It's a learning process and a live fire exercise all wrapped into one.
So in any case, also do let us know.
We don't see every comment, but do let us know what's working, what's not working.
We discovered some sound problems last time.
Hopefully you will find that they have been successfully addressed this time, but every piece of information is useful to us and we're working on it and we're working on it alone because obviously nobody is coming to anybody's house to troubleshoot stuff, so...
We being the two of us on screen, but mostly Brett and our wonderful now 16-year-old son, Zachary Weinstein, who's doing the production end of things.
Zachary is on the production end, and Toby is on the wall fishing end.
He's becoming an expert at wall fishing, which makes me very proud.
You keep on assuring me you're doing no wall lobstering, but I keep hoping.
We have not been forced to wall lobstering.
Yeah.
All right, so I do have a Not a correction but an addendum.
Last time I mentioned a paper that I thought was well worth people's time written by one of the founders of Instagram and it's Kevin Systrom and we will put a link to the paper in the description but the short version is he argues that the are not term that people have become familiar with about how likely an individual case of the virus is to propagate to The term is R-naught, right, which reads as R-zero to people.
I just wasn't sure that that tracked.
Yeah, so on paper it looks like R-sub-zero, R-naught.
He argues that there is a more important term that we need, which is R-sub-T, which I think is just said R-sub-T, although it's not clear in the paper, which basically tracks the change in transmissibility based on the
responses that people have, the behavioral responses, and so this measure is important, for example, if we want to compare municipalities or something and see what's working and what isn't, not just simply assigning an R0 to the virus, but assigning an R sub T to the intersection between a virus and a population is very useful.
So anyway, it's a good paper, well written, I strongly recommend it, and we will link it in the description.
All right, I think that's all I have.
You have corrections?
Yeah, well, not a correction exactly, an addendum.
I wanted to add or clarify to our discussion about essentials last time, where we spent a while talking about flooring, much to the chagrin of some people and much to the enthusiasm of others.
We were specifically making an argument about why flooring, which has been included in the list of prohibited purchases in the Michigan stay-at-home order, might be considered an important psychological tool for people in such times and in times of lockdown, giving them something useful, productive, and creative to do.
We didn't speak to the more obvious things on the list, things like seeds and nursery supplies, because we've alluded to those in the past, and it's a really easy argument to make that people should be allowed to buy seeds with which to potentially grow their own food, especially as supply chains come to be at risk.
So, I'd like to reframe the discussion slightly, without spending a lot more time on it.
The concept of essentials, we were arguing basically, is a category problem, but it speaks to pre-existing category problems.
Nobody's being prohibited from buying anything like Anything that they might eat, actually.
You can buy flank steak and blueberries, you can buy mangoes and whole chickens, and you can buy tofu and kale, to list six things that we've bought during lockdown, right?
So long as you can find them in the stores, you can buy those.
You can also, though, at the moment, if you can find them in the stores, buy Cheetos and Pepsi and Pop-Tarts and Twizzlers and Red Bull and donuts.
Not only are none of those things in the second list essential though, they are actually bad for human health, which by now everyone knows, even those of us who eat those things sometimes.
So nobody's attempting to draw that line between the things that we ingest that are arguably good for you and the things that we ingest that are quite obviously not.
If you can ingest it, you're allowed to buy it.
And because the fact that humans need to take in calories is understood in part because it is so easy to quantify.
And it's easy to quantify in part because everything alive does it.
And what we were trying to do by getting at the issue of flooring is get at those things that are essential to being human that are much harder to quantify but that are qualifiable necessaries.
Right.
So, because every evolved life form needs to take in energy somehow, if you're a photoautotrope like a plant, you convert sunlight into energy, into sugars, and then if you're a heterotrope like us, you either steal that energy directly from the autotropes when you eat plants, or you steal it from second-order autotropes when you eat other animals.
And we can look at it and we can count it, and sometimes we count the wrong things, and certainly we have been counting many of the wrong things, but it allows for this reductionist approach, and no one is trying to draw the line between that which you can take in to eat, which is necessary, and that which you can take in and it's not necessary.
But staying alive isn't the only goal here, is it?
Nor is keeping the economy intact the only goal.
If we emerge from this disaster, having mitigated it with fewer deaths than there would have been, had we not more or less stayed in place while hopefully getting outside, with fewer deaths, but we emerged broken and empty, this will be a large-scale depression that will be more than economic, that will afflict a massive part of humanity that will be far harder to emerge from.
Yes, and I would also point out we have kids who are growing up facing this unprecedented thing, and I know one of the things that gave me a lot of insight into where we are was in the early part of the lockdown, our older son Zach, who is manning the equipment here,
said in passing he asked me something to the effect of how often this has happened in my lifetime and I looked at him and I I said this has never happened in my lifetime or really ever this is unprecedented and it changed the way he saw things but of course you know a child who's seen a much shorter slice of history doesn't know how to calibrate this and the reason I raise it is
I can imagine a lot of frivolous things, things that really truly are frivolous, that might be actually arguably necessary in some parent-child relationship to keep a sense of normalcy.
And so the point I made last live stream was I defy you to draw that line well.
It's not that there isn't a line, but the point is that line actually has a lot of richness between individuals and special cases and, you know, if there's some candy that is, you know, a child's favorite and some parent is trying to deliver that normalcy and the fact that the candy was suddenly off limits would be jarring, I mean, I can easily see an argument for it, even if On balance, the candy is probably a negative with respect to its effect on immunity, for example.
Yeah, so really, we're pointing out the difficulty of drawing lines for which there are no easily quantifiable hard borders.
We're not saying that lines don't need to be drawn.
Some of them do, a hundred percent.
This is in large part what government is for, and we are fans of good government.
But some of the lines that have been being drawn are patently bad.
The closing down of beaches when people are maintaining social distancing is bad policy.
It's bad policy.
And in fact, it emerges fairly strongly, I would say, this week in several different papers, one of which we're going to get to.
But there is stronger and stronger evidence that the circumstances that are dangerous are indoor, close quarters circumstances.
And so, in some sense, by telling people you can't go to the beach or you can't go for a hike, we have been denying people something That they might do in favor of pushing them in the direction of, yeah, alright, you're not out at the, you know, nightclub, but you are in close quarters, probably indoors, where this thing is being transmitted.
In some sense we have to be very cautious.
We don't yet know how this thing functions and telling people you shouldn't do that because it stands to reason that you might get sick in a public park when in fact the data ultimately reflect the unlikeliness of that scenario is a problem.
And it also serves to create snitches out of people, right?
Which we are also beginning to see.
That people are being encouraged to tell on their neighbors if they perceive that they are going out too much.
And in fact, as I understood, the mostly good faith orders that were coming down early, early on there were no attempts to keep people in place.
It was, you know, shelter in place but go out for exercise, for fresh air.
And, you know, people with dogs had this easy excuse, which is, which is itself very interesting.
Why is it now that we understand that our dogs need to get out and move, but we are less likely to understand that we do as well?
Yeah.
So the whole idea of having bureaucrats do this from something other than a place of caution is, I think, legitimately frightening.
And it's resulting in backlash, which I think is itself frightening.
Frightening.
It's terrifying.
The backlash against the idea of controlling the virus through, you know, wearing a mask or avoiding social contact or all these things.
That backlash is obviously a danger to human health and even an individual who chooses to flaunt the rules is putting everybody who might be downstream of them epidemiologically at risk and some of those people are going to die.
So it's not a minor question.
But nonetheless, the incoherence of the policy is raising people's hackles with respect to governmental overreach and all of these things.
And so anyway, some sort of caution is clearly necessary.
Indeed.
All right.
Did you have a next topic you wanted to move to?
Well, ultimately we're going to be talking about that Santa Clara County serology testing paper, but it doesn't have to happen first.
As you like.
Actually, let's start there.
Many people will have seen this paper.
Do you want to introduce... Yeah, so Zach, maybe can you pull up the... I don't know how to pronounce his name, actually.
Bendevied et al.
2020.
What's up?
Um, let's see if I can do it.
I, for the first time today, I have stuff in front of me.
Let's see.
I may have it here, Zach.
Except I can't see my.
Nope.
Hold on.
Oh boy.
No, I don't, I don't have it here, Zach.
Uh, if you can pull it up, it's the one on the first page, Ben DeVita at all 2020 right below the break.
That'd be great.
Okay, so this came out at this point, I think it came out the day before we were supposed to do our live stream, so it must be three days ago, maybe?
I don't see, are you going to be able to show it to us?
Okay, because I will need to see it if we're going to talk about it.
The message is, basically, this is directly from the discussion, quote, the most important implication of these findings is that the number of infections is much greater than the reported number of cases.
The title of the paper being COVID-19 Antibody Seroprevalence in Santa Clara County, California.
So what these authors did was solicited via Facebook people to come to be tested for antibodies, which is to say past infection, not current infection.
It will not find current infection.
And then they had drive-through testing.
They were looking, they tried to control for three demographics, so they tried to get equal representation to match the demographic background data.
OK, well, we will find the link and put it in there.
I think, yeah, so this is the Santa Clara County Benavida et al.
2020 paper that is widely discussed right now.
So they they were looking specifically at, I think, age, race and ethnicity and zip code as the three parameters that they were trying to to match to the background demographics of Santa Clara County and were You know, shut down accepting applications for the serology testing when they got too many from certain zip codes or too many from maybe certain races or ethnicities.
And then did drive-through testing.
They also let people come with up to one child from a household.
So there was some non-independence among their data.
And let me just say that their conclusion is that, like I said, that they have, in Santa Clara County anyway, there's a far higher background in Santa Clara County anyway, there's a far higher background rate of infections.
In fact, I think the numbers they have, since I can't see the paper now, I can't actually pull up the specifics, 50 to 85 times the actual numbers that are being reported from the more standard testing in Santa Clara County.
Which means, if true, if those results are true, then the case fatality rate, which is defined from the Dictionary of Epidemiology as the proportion of cases of a specified condition that are fatal within a specified time, the CFR, the case fatality rate, would be far lower than the CFR, the case fatality rate, would be far lower than is And this is, you know, consistent with things that we've been saying for, you know, well over a month at this point, right?
That the CFR is likely to be lower than is being reported, but that does not change most of the things that we need to do in order to control this.
But let's just focus for a moment on the actual study.
You want to start by talking about it a little bit?
Well, let me just say, you're going to raise some concerns about how the study functioned.
Yeah.
It does mirror data we're seeing from elsewhere.
It's actually a fairly close match for the data from Scotland, and then there's another study, I believe from Denmark, that says the same thing, which is that the rate of infection is actually far higher than we know.
That when you test a population, instead of testing people who come through the door for one reason or another, you test the population at large to see how many people show signs of past exposure, you get a very high number.
Which doesn't change the number of people who have died of the condition.
And so what we find is that the transmissibility is higher than we think, but the mortality rate is lower.
And so it would not be surprising if Santa Clara County reflected this very same pattern.
On the other hand, it doesn't make the study viable.
If the study has methodological flaws, whether or not it comes to A conclusion that's mirrored elsewhere, you know, it's flawed.
So the question really is, what are the flaws and how vital are they in terms of the conclusion that was reached?
Is it valid?
Is it just noisy or is it misleading in some way?
Okay, good.
So there are several approaches that I could take to sort of critiquing and some of them bordering on nitpicking, but I think there's sort of two classes of critique of this paper that are large enough to warrant discussion here.
One is biases in the subjects.
So, as I said, the methods by which volunteers for serology testing were chosen was through ads on Facebook, and those selected arrived at testing centers by car for drive-through testing.
This means that those who were included in the study had both means and interest.
Let's take those one by one.
Means first.
Um, you're likely to be over-selected, uh, for relatively wealthy people.
At the very least, you're not including any homeless people if you can only get testing when you show up with a car at a testing site.
Uh, so, um, this, this though, um, so this, this is gonna miscount in maybe both directions simultaneously, which looks like noise, right?
So, if you, If you don't get some people in lower socioeconomic classes who haven't been tested because they didn't have previous means to get tested, and specifically homeless people, you will have missed a large number of people who were positive, and thus this could be actually an undercount, right?
Yep.
And I can see an argument for it going the other way, but it seems most likely that this means bias goes one way and actually suggests that these results in this paper are conservative, if everything else is correct.
With regard to interest, though, it clearly goes in the other direction, that those people who respond to a Facebook ad, A, are looking at Facebook at home.
That is to say, again, this is a question of means.
They have the means to have to have a computer at home through which to look at Facebook, but it stands to reason that those people who think they might have been infected in the past will have more interest in going to be tested than those who have no reason to think that they've been infected and, you know, other than an excuse to get outside, you know, why is it that they're going to go and potentially put themselves at risk for such a study?
So given that, given that interest was a driver, I can't, you know, it's impossible to quantify how large a driver, but a driver of how it is that the subjects got selected for this.
That is going to suggest that that's going to go the opposite direction of the way I think the means question would, which is to say it's going to have driven the number of positives up compared to the background population, which suggests that the case fatality rate that they're reporting is actually too high and that it should be lower.
So that's the first large class of critique that I have about the paper.
You want to say anything about that?
I would add one thing to it, which is a different kind of critique altogether, and it may well be that this is so well understood epidemiologically that it's a non-issue, but there is a question in my mind given, and I have studied immunobiology, albeit a long time in the past, but There's a question about whether or not the antibodies necessarily reflect a history of infection or do they reflect a history of exposure.
And my guess would be that the level of antibodies that is being detected would reflect a cleared infection.
That is to say, the immune system goes into a pattern where it becomes good at recognizing the invader that it's already seen and that that is what is being detected in which case it would reflect an infection.
But it is conceivable to me that especially in light of how much doesn't make sense about this virus or not doesn't make sense but let's say it's anomalous about this virus.
One thing we have to ask ourselves is what happens to somebody who has a high level of exposure but doesn't contract an infection.
So So, it's worth asking the question, and I would love it if somebody knows the answer, and it is clear that these antibody tests are necessarily reflecting an infected person after their infection has cleared up.
That would be useful.
Well, in fact, it raises the question of high level of exposure without contracting an infection, what would the antibody test read?
What then does asymptomatic mean?
Asymptomatic is a category that could mean either of those two things, actually, if they are in fact distinct.
Yeah.
So anyway, maybe somebody in our audience can clue us into what we need to see.
Yeah.
So the other major concern that I had about the scientific validity of this paper, even though again, and we'll talk about this more later, I tend to believe that the results are more true than the other results that we're seeing that show a very, very low Background infection rate in populations.
The authors demonstrate, they go through, you know, a fair bit of statistical song and dance about explaining how they assess the actual rate of false negatives and false positives in these tests.
And other people online have assessed whether or not they think that those are well done.
But they don't seem to deal with the problem that With a low base rate of infection in the population, even the higher rate that they report here, it's like the higher rate that they report is up to close to 5%.
You know, it's still very low, right?
So with a relatively low base rate of infection in the population, False positives potentially outnumber true positives in test results.
So I'm at considerable risk of losing a lot of the audience.
I'd like to walk us through just a couple of made-up numbers just to make this point.
Because I think a lot of people have been saying this, but with some made-up numbers that are simple, it might stick for people.
So let's imagine a population of 103 people.
And again, this is just to make the numbers easy.
Let me say first that the authors of this paper claim a false positive rate, well, a confidence interval and specificity between 98.3 and 99.9 percent, which is staggering.
So that would be a false positive rate of 0.1 to 1.7 percent, which is very, very low.
Most of the tests we're seeing suggest about a 5 percent false positive rate.
So just for ease of doing some fake math here, I'm going to use this 5 percent.
So in a population of 103 people, where only three people have been infected, that's based on this generous background infection rate of 3%, or 2.9%.
If the false positive rate is 5% for the test, that is, the test tends to give a false positive one out of 20 times, 5% of the time, maybe the three here are actually positive, test positive.
The flip is here that false negatives will be pretty rare in such a situation.
So the three in the population of 103 will actually test positive.
The remaining 100 though, all of whom are actually negative, Five out of those hundred will also test positive because you've got a 5% false positive rate.
So in that case, with those very particular numbers that I chose for ease of communication, you'll get eight positive results, in which three are actually positive and five are not positive, which means the majority of people who show up as positive are actually not.
So that's that's obviously difficult to wrap your mind around and let me just one more one more mathematical thing and then I'll drop the numbers.
If you take the same population and the background infection rate is now 15% rather than what I said before 3% Now you have 15 people out of the population of 103 who are actually positive, only 4 people out of the 88 remaining people who are actually negative, which means you got 19 positive results, only 4 of whom, only 20% or so, are false positives.
So as the background rate of infection in the population goes up, the risk that you're positive is a false positive goes down.
So I would also point out that there are two potential kinds of false positives here and which kind we are dealing with has a lot to say about how easily we can correct for this.
One is a false positive where the test comes up positive for no good reason and a second test would suggest something was wrong.
The second would be a false positive.
I guess this would be the case if people who had exposures but did not contract the infection had sufficient antibodies to trigger the test and we regarded them as it as having cleared an infection.
They would test positive each time they had the test even though they have never had the infection.
So it is possible And yet I would argue that those are people we want to include.
If they've got immunity, we want to include them.
Well, I don't know if they have immunity in this hypothetical situation, but let's put it this way.
It's not that that situation would be uninteresting.
It's that it's a signal for a category that we either are or are not counting.
And so we need to be very careful not to assume people into a category They are not necessarily in, and a lot depends.
You know, it may be that experts on coronavirus infections know very well what a particular titer of antibodies implies, and so they've ruled it out for good reason.
Or it may be that things are so novel here that we may discover that there was a category that we had lumped with another category and changed the interpretation in a way that was misleading.
Yeah, in fact, what I'm seeing is too much of the time, and this is forced by the fact that this is such a fast-moving, fast-emerging pandemic, that categories are being split into two, that we are making binary categories that really warrant many more subdivisions.
And so, in general, I think titer, you know, countable titer of antibodies counts as a yes.
And if you can't see it at all, it counts as a no.
And that's obviously not a fine enough line to get at some of the questions that you and we are raising.
But there's also, you know, there's so much important work to be done.
And this, you know, will seem like nitpicking.
Why aren't they splitting it into more categories?
Then you have this question, just like we were talking about with regard to what's essential, of where to draw the lines.
So, are there any antibodies present versus are there no antibodies present?
Is at least a line that everyone can agree on that's quantifiable, yes or no, but binary situations that are actually, in which you can have zero and you can have some to a lot, that's actually a continuum and it warrants a different kind of statistical analysis that we just don't have
At the very least we can say that the phenomenology of these infections is so varied and so confusing that we know there are features of this that are utterly continuous and leave you scratching your head.
So whether the infection itself and the tests we're using are truly testing binary circumstances or whether there are categories we're missing is an important question.
So have you gotten where you wanted to go with the the Santa Clara County Um, I guess, let me just say, I mean, this is probably a decent segue to, I don't know where you're going next, but I suspect it's a decent segue.
Um, so, all of those critiques, and again, I have more, um, which I'm not going to make here, all those critiques of this study, um, Do not mean that I think that the overall conclusion is likely wrong.
And this is based, as you suggested, on some of what we're seeing elsewhere and on what we understand to be a high transmission rate in places like hospitals and nursing homes and such.
So I don't find that this study demonstrates its conclusion to any high scientific standard.
That said, a low CFR, even if the case fatality rate really is quite low, you know, a lot of people are saying, you know, less than 1%, meaning doesn't mean no CFR, doesn't mean it's not killing people, and deaths are quite high in some places.
So that begs the question, is the CFR higher?
In places like New York City and Lombardy?
I don't know.
Or are background rates of infection higher?
And what is it?
This is where we need the excellent epidemiology, and it's just moving so fast that it's really hard to get to.
So some of the other studies that have reflected a similarly high base infection rate, and therefore low CFR rate, We're done with very different methodologies.
In other words, blood from blood donors, for example, was tested and that doesn't have the same bias of soliciting people who can drive in from a Facebook app.
So, it does suggest that there is reality to this interpretation.
Whether it varies by location, we don't know.
And, you know, the study is some sort of data.
It's just compromised by methodological flaws.
Absolutely.
So I guess, I mean, there's a lot more to say here, but the main thing I would say is case fatality rate gets at deaths.
And it may well be that the biggest lingering effect, health effect, put aside the economic effects, but the biggest lingering effect health-wise of this virus is among the survivors that we are seeing increasingly but the biggest lingering effect health-wise of this virus is among
And of course, we don't have any long-term data, because this isn't a long-term situation yet, but that people who survive have respiratory problems, have lung problems, have cardiac problems, may have neurological problems, that it's affecting all of these systems in the body.
And of course, why wouldn't it?
Why would we assume that this respiratory virus, that this virus that we have named a respiratory virus, would only affect respiratory systems?
There's really no reason to assume that, and so focusing only on the case fatality rate, as if that is a good measure of the effect on human health and well-being, is not wise.
Yeah, it's a hazard.
I'm particularly troubled by the fact that we don't have any good resolution on the so-called asymptomatic cases, whether they have damage or not.
And if you have followed the discussion here and on Eric's podcast about telomeres, the thing to remember is that In effect, damage pushes you closer to some threshold from which you cannot repair, right?
So that threshold at which you cannot repair some vital tissue is death, right?
You have a vital tissue, it's gonna fail, you can't live without it.
So things that damage you push you in the direction of death.
Now, a low CFR reflects people who have not died yet.
It does not mean they were not shoved in the direction of death by 15 years.
It also leaves open questions, which increasingly I hear circulating.
about whether A, this is a virus that can recur, right, without having to be reinfected, whether B, you have immunity once you've had it or it's going to behave more like colds and flu where versions of it will reinfect you periodically.
So to the extent that this is a virus that clearly gives many people enough of a shove to push them across the threshold where they can't get back, It's presumably giving many more people who do get back a shove in the direction of mortality.
And then what is it doing to the people who had a low-symptom or no-symptom infection?
What will they be like 15 years from now or something like that?
So anyway, all of these things are important questions.
And it points once again to the limitations of binary thinking.
Dead or not, not a sufficient analysis.
Yes, fully recovered is not fully recovered, and that's true for many, many phenomena, right?
We medically miscategorize this, or at least the story that we tell does.
So I wanted to suggest a couple things.
Places where we're falling down, and then places where we're succeeding.
Again, I am stunned by how vibrant, how interesting, and how normal the scientific discussion is.
Everybody has been reduced to amateur status.
Even the experts who are experts at coronaviruses or epidemiology or whatever are novices with respect to this particular virus.
So people are interacting in a way that is prone to error.
I've seen lots of false starts, lots of cul-de-sacs, but the fact is the scientific apparatus is actually functioning in some ways better than I've ever seen it function before, right?
And the fact that you can tune in, that you're not being excluded because, oh, you're not an expert enough to have access to the literature, is wonderful.
On the other hand, Even, I'll say, not only are most of these papers going up on preprint servers, but Elsevier, you know, one of the biggest blocks to actually sharing scientific data out there, because it's one of the big publishers, is making its COVID-19 related publications open at the moment.
Yes, this has embarrassed publishers into behaving as good citizens, which, you know, is all extremely positive.
The problem is we are hamstrung by the quality of what there is to be analyzed.
And there are a couple of things that I would suggest here.
One of them is that we are unfortunately being hamstrung by the political apparatus, which has a perverse incentive.
And I'm speculating here as to why this is the case.
The Trump administration, for example, has been very slow and reluctant about pushing testing, right?
And there's a question about why would anybody not want more testing?
How could information possibly be bad?
And the answer is that there's a very reasonable, if immoral, argument for eliminating that data if you are a politician in a position of authority.
And it has to do With what I call monkeying with the baseline.
If you are in a position of authority and you're going to make decisions, and those decisions are going to have life and death ramifications, they're going to decide whether or not we fall into a depression or don't, and you don't want to make a call and have it come back to haunt you, then leaving the data very vague maximizes the opportunity to Shift the narrative to push blame onto somebody else.
So I remember this first dawned on me after the Fukushima disaster in which the Shinzo Abe administration in Japan began doing all kinds of bizarre stuff.
Like, there was waste that you couldn't burn because it was too radioactive, and there was a very clear legal standard about how much radioactivity caused something not to be viable for incineration, and they justified the adding of non-radioactive material to this radioactive waste in order to get it below the threshold.
There's no justification for this.
All it does is add extra pollutants to the atmosphere, but they gamed their own standard.
And then, I remember they were moving waste around the country to burn it in different places and the only argument for doing this What I could see is that later on, if you wanted to find out how bad the Fukushima disaster was, you were going to have to figure out what the baseline rate of all of these cancers was.
Well, if you're burning the radioactive waste in different parts of the country, then the baseline will be artificially high, which will make the Fukushima disaster look artificially okay.
So anyway, monkeying with the baseline is one way to dodge responsibility.
And I'm concerned that what we are facing is a political apparatus that is playing that game and blinding us scientifically at a time when we need high-quality vision more than ever.
They seem totally willing and capable of monkeying with the baseline.
I like this phrase very much.
Fukushima, of course, like Deepwater Horizon as well, was different from what's happening here in that while there was an event, presumably, there was a patient zero, there was an origination of this virus wherever it came from, as we've talked about in past podcasts, our awareness of it in much of the world was this rolling awareness, right?
There was not an event, a tsunami, Or a burst gas line, in the case of Deepwater Horizon.
And it means that there was denial, followed by possible embarrassment, and then, sort of late to the game, I would think, comes monkeying with the baseline.
Hey, we can be opportunistic here and change what's going on.
It just doesn't look like the monkeying with the baseline thing is what was happening earliest, at least not in the US, not in the Trump administration.
You know, maybe.
Well, I think the problem is, I don't know how monkeying with the baseline shows up in the official discussions, but my sense is it's going to happen everywhere that responsibility is being apportioned.
I think it accounts for some of what we see in China and a lot of what we see in the US.
And when it first dawns on people that they don't want good data, I don't know.
And it may be that that becomes a stance that you even recognize as a general matter, that good data tends to put you in an awkward spot.
And so cloudy data is preferable if you're in a political position of authority.
But we, who have to face the outcomes here, Have an interest in forcing our political apparatus to do right by us.
And so I was going to make a suggestion for in addition to what you have been saying since literally moment one on this, which is testing, testing, testing.
That is absolutely the first move here is to figure out how widespread this thing is.
But People who have been paying attention to these live streams will remember we talked about the USS Theodore Roosevelt, the aircraft carrier that docked in Guam because it had a infection spreading on board and I remarked that this was a lost opportunity because this had been an isolated population which actually allowed you to get good data about how this stuff is spreading and what the rates of mortality and serious illness are per infected person.
That ship has unsailed as it were.
That ship has docked.
But what I realized actually in the middle of the night last night was... Was that what was going on?
That is what was going on.
Is that military bases might actually provide the opportunity that we need.
And I want to be very careful here.
I am not suggesting that anybody experiment on the military.
There is a very dark history of that sort of behavior and I'm not advocating that.
But what I am suggesting is that military bases function as effectively towns.
They're towns with a gate.
Where you know who comes in and you know who leaves.
You've got built-in track-and-trace.
Right, built-in track-and-trace.
You have a system of authority in which people can actually be told, do this, don't do that.
And the question is, could these bases be used, A, as study locales?
Could we, what I was thinking is that we would have two medical teams on a base.
You would have the team that was dealing with treating the patients, And unlike is happening in the world at large, you would have an entirely separate team dealing with understanding what's taking place, right?
And that we could discover, A, how similar these these outbreaks are, B, you could potentially test different things, like different bases could have slightly different rules about whether you're allowed to go to the park, and we could discover if in fact going to the park is really a safe thing to do.
So anyway, I don't know how it would be done responsibly, but I do believe that there is a responsible way that would treat the members of our military with honor, but also allow us to take advantage of the fact that they already have these isolated communities And potentially allowing for, also, iterated serology testing.
You know, I'm just adding more and more to what I want out of the testing apparatus, but widespread antigen testing for current infection, widespread serology testing for past infection, which looks for antibodies, And at most, I think people have been tested twice, or if they got better and they maybe got sick again, maybe a third time.
But in order to assess whether or not your first positive was a positive, maybe, people, you know, it was a false positive or a real positive, people get tested twice.
But I would love to know what happens to those titers, what happens to the levels of antibodies over time.
Early estimates were suggesting that It's a good seven to ten days after active infection before you have any noticeable antibodies in your system, and that climbs for three weeks or so, and then remains steady for no one knows.
No one knows as far as I can tell, in part because the tests are just missing in action.
And this is partially due to errors in the early days, at least in the US, on the part of the CDC and such, but it's much more widespread than that.
I would love to see people who do show as having been infected.
I would like to know what happens to their antibody titers over time, which would then begin to give us a sense, possibly, as to whether or not immunity lasts and for how long.
Yep.
So imagine you had a military base in which everybody was being tracked from some, you know, moment zero.
Yeah.
And you could simply say, well, what do we know about asymptomatic cases and whether or not they show the same kind of ground glass lung damage that shows up in severe cases?
Well, that's an easy... you could test that question, I would imagine, in a few days if you had a population where you could simply identify Here are some people who match that description.
Let's give them an x-ray and find out how their lungs look and then, you know, track them over time.
So, in some sense, I'm heartened by the way the scientific apparatus is functioning.
I am disheartened by why we are scratching our heads over what seems like some of the most basic questions about the course of this disease, even here, you know, many weeks down the road.
Yeah, months even.
Yeah.
Oh, I agree.
Well, what else do you want to talk about today?
We've got maybe 15 minutes left.
I think we should talk about the subway study.
All right.
Now that I think I do have open here.
I think, hold on Zach, let me see if I can find it.
Yeah, this is it.
So you can put it on my, put my computer screen on Okay, so this came out of MIT.
I don't know this month.
I don't remember exactly when.
Oh, very recently.
So, The Subway Seated the Massive Coronavirus Epidemic in New York City is the title of the working paper.
Jeffrey Harris, who is an economist, and this paper is a very interesting paper.
Basically, it looks at New York and it tries to figure out why New York suffered the rather impressive rate of infection that it has faced.
And as it says right up front, it tested the hypothesis that the virus was being transmitted by the subway.
And it goes on to, you know, the paper is very careful.
It says, look, we don't know whether this is actually true because the subway is obviously a non-random network of trains.
Those trains are built to accommodate at least The historical understanding of where people needed to go from and to and so there are lots of other things that Travel at the same, you know in the same pattern however It does appear to be a very careful hypothesis test and you may have heard so the study is it says it's a correlational study But I believe it's actually a little too cautious on this front.
So the correlation being he's mapped cases.
I believe it's just active cases known in various parts, in zip codes, I think it is, in New York City.
And also then looked, actually, maybe I can pull this up.
He's got a lot of good maps.
Okay, Zach, if you want to put this back up, just so, sorry to interrupt, Brett, but just so people can see what one of the things he's done is he's got new reported cases in Manhattan as the red dots here.
And then the sky blue bars are turnstile entries.
So entries at turnstiles in the subway system from March 1st through April 4th, I guess, where you see that no...
No strictures were in place, I believe, for that first week of March, and then it starts to decline, and it declines quite a lot.
And later in the paper, he's got this divided up into the five boroughs, and also looking at differences.
He divvies it up a lot.
See if you can find the map.
He says, actually, particular lines are suspect, and he identifies... So is it this one?
That's the borough's one, not Anzac?
Keep going.
This paper is also very funny.
He uses humor in a way that you don't tend to see in academic papers.
Also, I shall say, that's one of them.
This is one line.
This paper is also very funny.
He uses humor in a way that you don't tend to see in academic papers.
Usually, presumably, it's peer-reviewed out.
Yeah, I mean, I guess this is part of what I'm reacting to, is that there is something that has become more honest about the way people are writing in this context.
And so anyway, keep scrolling a little bit.
I will just say in... You want a different map?
Yeah.
Oh, you want the map that intersects the two lines?
Sorry, we're probably giving people vertigo.
Yeah, we are probably giving people vertigo.
Yeah, so... Okay, so you want Zach to show this one?
It's a little hard for me to parse which map this is, but it doesn't really matter.
The point is we have a correlational study.
You have probably been told that correlation does not imply causation.
That is a lie.
I am sorry somebody hurt you that way.
The truth is that correlation does imply causation when that correlation maps onto a pre-existing hypothesis, as it does here.
It does not say that this locks down the question, that this is causal, but it does say that this is strong evidence of a causal link, because the hypothesis was being tested by the paper.
Just with regard to correlation does imply causation when it maps onto a pre-existing hypothesis.
That does not require that the data didn't exist at the point that the hypothesis was formulated, but that the person formulating the hypothesis was not aware of the data.
Exactly.
The information was out there.
This guy, Harris, said, I've got an idea.
I've got a hypothesis.
I think that it was about subway ridership that is Related to, that was causing New York City to be such an epicenter of outbreak and he then went, having made the prediction, into the data and used the data to effectively, effectively test the hypothesis.
Although he doesn't make such a broad claim here.
Yeah.
He's a little more cautious than I think he needs to be, but he even discovers patterns within that.
Not only are particular subway lines implicated, but the local trains are apparently a particular source of infection, if this hypothesis is correct, and the reason is basically that local trains keep you in contact with other people longer.
So a local train It's one that stops at each of the stations along the way, and so if you're going a long distance on a train and there are no express trains, you get on the local train and you're on it longer because you're stopping at each thing, which also means you're being exposed to more people.
Maybe this is obvious to people, but it wasn't to our children when we were talking about this last night.
If you're going from A to G and an express train takes you, you get on at A and it doesn't stop until you get to G, That's quite different from a local train, where it'll not only take you longer, which is why people usually don't like the local trains if they know they're going all the way to G, but if it stops at B, C, D, E, and F on the way, there are many more possible transmission points along the way.
Yep.
And you will have been exposed to many more people, right?
That's the transmission points.
Oh, transmission points.
Okay.
People are points now.
In any case, what he concludes in this paper is that the reaction of the transit authority doesn't blame them because they didn't know what we now perhaps know.
But he says the reaction was the wrong one, which was to make all trains local and to reduce the number of trains to discourage ridership.
And the point is, if the mechanism here is right, then the best reaction would have been to increase the number of trains and express trains beat local trains because you're less likely to be infected in them.
He also points out that the fact of trains interacting with the epidemiology here is also strongly suggestive of economic factors and people have been trying to chase down what are the economic factors that have certain populations more afflicted than others and here would be an obvious one.
Mass transit.
Yeah, if you're condemned to mass transit and you have a job that you can't afford to not go to because, you know, you're depending on it economically week to week, Then you have a recipe for disaster in some parts of town and not others.
And apparently the Transit Authority started cleaning trains twice a day.
His point also in the article is, you know what, more trains.
Still with the hope that there is no, that there is far reduced ridership over time.
But at the end of each line the train should be being cleaned completely.
Yeah.
Now, last point on this.
I did read one report which reflected something.
I had been wondering about whether somebody needed to invent this, and apparently in Asia there are already UV robots and sanitizers that are being used to do things like clean buses and things.
There it is.
Things like that could be very helpful.
I also strongly suspect that HVAC systems are going to turn out to be a major villain here.
Oh, 100%.
We actually have begun already to see that information, and it's utterly predicted by the information from tuberculosis outbreaks from a century ago.
And how easy a problem is that to solve, right?
Because basically to the extent that air is being captured in one place and filtered through some system and then pumped back out elsewhere, that air is under your control and what you basically need to do is expose whatever passes through it to extreme enough circumstances that viruses can't survive, be it UV light, be it electromagnetic charge, whatever it would be,
Again then, so we have mass transit, which now is mostly subways and trains and buses, but before the massive lockdowns was obviously planes.
We know that planes were the major transmission agent, at least between continents and probably between states as well.
But yet one more reason, go outside.
Be outside when you can, because your HVAC systems are going to be pushing stuff around in ways that we do not yet have control over.
Yeah, so all of this suggests that there will be arbitrary features of life that differ between places that will have important epidemiological impacts, and one way to think about it is that there are different kinds of reproduction for a pathogen, right?
The fact that Is it more true, if this paper is accurate, is it more true to say that SARS-CoV-2 is transmitted largely through aerosolized particles or by subway, right?
The fact is it actually, it's different scales of transmission, right?
A virus that's really successful, like this one, is one that is utilizing many different modes of transmission over many different scales.
And yes, it always has to go person-to-person, except in the rare case that it goes to some other creature.
But person-to-person, when that person is in transit over an ocean, is obviously its own hazard.
Absolutely.
We need to start thinking at these multiple scales simultaneously.
All right.
Well, that's about an hour.
What do you think?
All right.
I think we've done it and we'll be back.
Okay.
We'll be back in maybe 15 minutes or so to do an hour of Q&A addressing your Super Chat questions.
All right.
See you then.
Export Selection