many of the herd immunity theorists strike back and ask “where are the deaths“? But that is not the right question for testing herd immunity claims. Those claims were about transmission slowing down, and those claims should be true about Covid-19 cases whether or not more people are surviving in the hospital.
Why are cases spiking but deaths not spiking? Here is a set of hypotheses, in my subjective order of likely importance.
1. The strains that are circulating are less deadly.
2. The people who are getting it are less frail. See the discussion of “dry tinder” in Daniel Klein’s essay. And also enough folks finally got the memo about protecting people in nursing homes.
3. The treatments people get now are helpful, whereas six months ago they were ineffective/harmful.
4. Testing protocols are finding more of the milder and asymptomatic cases that they were missing before.
5. The long and variable lag between cases and deaths has become longer and more variable.
And note that the average daily death rate still stands above 700, which is outside of the range for a normal flu, at least on an annual basis.
Thanks, Arnold. I know you have a high opinion of Tyler’s work, but I don’t. Although I have been reading MR for a long time, I still find Alex’s work much better than Tyler’s. Tyler’s posts are very useful for the links to research papers, but you have to spent too much time to make sense of his ideas because he rarely discusses them within a reasonable analytical framework (his short-cuts ideas that pretend to refer to some obscure framework are a joke).
Take Covid-19. I always remember your post of April 19 “Sooner or later, Mild or severe”. A very simple framework that you use to say where you were standing on the choices you identified (btw, I have always thought you should have continued using this framework to illustrate why different people had preferred different choices as well as to discuss alternative government strategies in greater detail). Tyler has already written hundreds of posts on Covid-19 but without any simple framework that can help “to accumulate” knowledge for our responses –yes, we have been learning and our responses should be conditioned on our learning. To make things worse, Tyler has failed to provide a framework for discussing governments’ responses to Covid-19 and their economic consequences. Yes, I know we didn’t have that framework before Covid-19 (your standing on macroeconomics makes quite clear why we shouldn’t rely on it) but Tyler could have attempted to put some pieces together for a new framework.
Lately, I have also found that Tyler has been quite ignorant of relevant information — for example, he wrote the post you refer when this information had been available
https://www.citizenfreepress.com/breaking/cdc-publishes-new-survival-rates-for-covid-msm-goes-silent/
“I know you have a high opinion of Tyler’s work, but I don’t”
I actually find MR to be redundantly boring. I stopped following. Peak MR was circa 2011.
In Tyler’s favorite game of underrated/overrated, I’m going with overrated.
And, this leaves me with a puzzle. ASK is my favorite blogger, but his favorite blogger is Tyler. How to reconcile this? Oh well…
1. Mask-mediated variolation. Let me explain. More people are wearing masks in more public places, more often, and official and social enforcement is stronger now than many months ago. Also there has been more re-opening and slow but gradual normalization, increasing the chance of encounters between currently-infecteds and never-infecteds. There were also a few important holidays in which groups and families tend to gather together, and, ah … some other major crowded events. The bottom line is that masks – especially so-so masks, worn without much medical-level discipline, and re-used multiple times without sanitizing – don’t absolutely prevent spread of the disease, but do mean that one’s likely first exposure to the virus will have a much smaller viral load. Which is kind of like a policy of throttled and spontaneous mass variolation, on the way to herd immunity. That means more cases, but also, many more of those cases will be mild and asymptomatic, even correcting for age, health, risky behaviors, comorbidities, and drug use (especially immuno-suppresants like steroids).
2. We don’t actually know the historical case / infection rates. There weren’t enough tests, the tests we had were even poorer than the ones we have now, and the results came in too late to do any good. When we look at seroprevalence in certain populations and places, we are shocked at how quickly they got so high. We thought hospitalizations and severe cases were high, but it looks like they were only a drop in a population that can mostly brush the virus off. If you tried to reconstruct the likely historical pathway of all (not just tested, reported, or serious) cases in, say, Northern Italy or NYC, you would see a much quicker and higher peak, and a much faster descent, that would make today’s nth-wave cases indeed look like they validate a gradual progression towards herd immunity.
3. It increasingly looks like there was something akin to a ‘replication crisis’ in many of the early antibody tests, far too many Type I and Type II errors. As just one example, it seems that many people had similar coronavirus colds in the past, and either (a) were effectively immune to cv19 and so didn’t produce new antibodies specific to it and tested ‘negative’, even though they were likely exposed and ‘infected’, or (b) weren’t exposed, but tested ‘positive’, because the old antibodies they did have were pretty close to those people make for cv19. We also know that there has been a puzzling and troubling amount of inconsistency of these tests from different companies and brands, from different manufacturing ‘generations’ of similar tests, and even of multiple tests done on the same person over time.
But gradually the tests have gotten better, more are being done, and more people are getting them because results come back in time to be meaningful. So we are catching very many more actual cases in our net, less-garbage-in, so less-garbage-out, but the past data is full of unknown and changing amounts of garbage.
One might hope that eventually researchers will be able to reconstruct what actually happened to whom and when, but eventual mass vaccination will take the wall containing some of that important information and paint right over it. Ideally, everyone who gets a shot will also get a good test at the same time, and that mass of information about billions of people could be collected and analyzed for research purposes.
Tyler is mistaken that herd immunity necessarily implies cases fall. Cases are detected by real-time PCR which is highly sensitive and can potential detect as low as a single copy of a virus. Immunity means that when you are exposed to the virus, let’s say ten thousand copies, it only replicates to a limited point — let’s say ten billion copies, but no detectable illness — before the immune system gets it under control. Whereas in a person without immunity, the virus might grow to trillions of copies and cause sickness.
The threshold for a ‘case’ measured by PCR is different from the threshold for sickness or immunity (freedom from sickness), which is different from the threshold for death. You can have cases in immune people if your test is sensitive.
We might have a better idea of what weight to put on each of those possibilities if we had been doing random sampling and testing from the get go. It was madness that we didn’t.
Why didn’t any other country conduct randomized testing, either?
Things are looking up for us virus “kooks” here in Texas. Our 6 yo is in her fifth week of first grade, IN PERSON. She loves it.
Feared coronavirus outbreaks in schools yet to arrive, early data shows
Tracking infections over a two-week period beginning Aug. 31, it found that 0.23 percent of students had a confirmed or suspected case of the coronavirus. Among teachers, it was 0.49 percent. Looking only at confirmed cases, the rates were even lower: 0.078 percent for students and 0.15 percent for teachers.
“These numbers will be, for some people, reassuring and suggest that school openings may be less risky than they expected,” said Emily Oster, an economics professor at Brown University who helped create the tracker. She noted that the school coronavirus rates are “much lower” than those in the surrounding community.
Separately, the early data from Texas also shows low levels of infection. Since the start of school, 3,445 Texas students reported positive coronavirus tests — or about 0.31 percent of the 1.1 million students attending school in person, according to data released Wednesday. An additional 2,850 school employees tested positive, although a rate could not be calculated because it was not clear how many of the state’s more than 800,000 school staff members were working in school buildings.
Teacher’s unions in Texas that keep track of infections say they have been surprised by how low it was. In many parts of the country, teacher’s unions have resisted school systems’ efforts to return to classes, saying sufficient safeguards were not in place.
https://www.washingtonpost.com/education/feared-covid-outbreaks-in-schools-yet-to-arrive-early-data-shows/2020/09/23/0509bb84-fd22-11ea-b555-4d71a9254f4b_story.html
The bottom line on all the new data is that the various lockdowns/shutdowns and other restrictions are immensely harmful and evil with little to show for it.
For months now most people already started taking all the necessary common sense precautions relevant to them and the hissy fit by our managerial elites is embarrassing.
I am a long term reader of Marginal Revolution- all the way back to 2004 it has been a top line bookmark on every browser I have ever used. Yesterday’s post on herd immunity might have been the last straw for me, though. Cowen has repeatedly posted outdated or just wrong information during the last 6 months- information that it was easy to determine was wrong by doing just a cursory examination. Cowen has basically farmed out all of his thinking on this issue to the Chicken Littles of the political universe, and he repeatedly parrots them, too.
Of the reasons Kling listed above, it is probably a combination of all of them, but the most likely explanation is that the numbers of PCR tests and their increasing sensitivity is simply catching a much larger number of present and past infections. If you examine the data, in the US, as an example, is running about 8 times the numbers of PCR Covid-19 tests/day from June-today than it did in March-early May. Additionally, the numbers of cycles used/test to amplify the signal have been increased as testing resources became less constrained during the Summer- this raises the sensitivity of the tests run today vs those run in April. Additionally, Klein is correct- when you examine the CDC models and data, we have run consistently under total expected deaths for over 2 years prior to March 2020- those people who didn’t die during that time just got frailer and older- prime candidates to die from a respiratory illness- any respiratory illness.
On the last point- yes it is still at about 700/day, but those deaths are increasingly from the death certificate matching where all past deaths from the last 6 months are being compared to the positive COVID test list, and if you died of cancer in July at home, but had a positive COVID test in April during a doctor visit, then you are a COVID death in the present daily report. I don’t know how the CDC will weed these out, but we are getting close to zero excess deaths in the last 4 weeks.
“Cowen has repeatedly posted outdated or just wrong information during the last 6 months- information that it was easy to determine was wrong by doing just a cursory examination.”
Exhibit B: “We conclude that the Sturgis Motorcycle Rally generated public health costs of as much as $12.2 billion.”
His mood affiliation and virtue signaling overrode any common sense that this was completely bogus.
https://marginalrevolution.com/marginalrevolution/2020/09/monday-assorted-links-271.html
Yes, I pointed that out in the comments section on the first post he linked to that. Cowen’s critical thinking seems to have disappeared during this particular event. I have had problems with Tabarrok’s fetish for testing and tracing, but at least he is still doing his own critical thinking and analyses, and doing them pretty well that I have seen.
Another example from recent weeks was his “puzzling” over the difference between Pakistan and India’s- i.e. why was Pakistan doing so much better at containing the spread, and then he went into several hypotheses? It took me exactly a minute to demonstrate that India was running close to 1o0 times as many daily PCR/other COVID tests at the time- the most obvious explanation, and conspicuously absent in the blog post.
Maybe I’m being more charitable than you are but my impression is that when Tyler links an embarassing story without comment it expresses disapproval (with deniability).
For example if Tyler supported the radical left he would bury this kind of story, not give it its own headline: https://marginalrevolution.com/marginalrevolution/2020/09/solve-for-the-seattle-equilibrium.html
I’m inclined to agree with Ryan RE: Tyler’s links. Although it feels like part of the ‘Straussian’ game means you never know for sure. Which, like the (apocryphal?) professor who claimed all his lectures had an error, might be good to keep me on my toes as a reader?
I also agree with you Yancey, RE: Alex’s covid takes. It feels to me like he’s working with the most-mainstream axioms/givens, but applying real thinking to them (like calling for random tests, and for regulators to suck less).
But for both Tyler and Alex, the covid stuff has had kind of a reverse Gell-Mann amnesia effect for me – because I’m seeing them reason unimpressively about stuff that I’ve looked into moderately deeply, it’s making me downgrade my priors RE: the quality other thinking.
I’d like to ask for Kling’s view on Trump’s “School Choice” initiative for his second term. Kling has been a proponent of that type of initiative in the past. Do he think that Trump isn’t sincere, wouldn’t be effective, or Kling just doesn’t want to be associated with advocating for Trump?
As a secondary question: what is Kling’s view on the Abraham Accords? Is that simply outside Kling’s subject area?
sorry, small typo: “[does] he think…”
I am not aware of any good evidence for (1) and it would be useful to point to it if you have it. (2)-(4) would all be tricky confounders to deal with in any attempt to isolate an effect of (1).
(5) seems like less a cause than a collateral consequence: it is what you would expect if (2)-(4) were working. When young people do die of COVID they probably take longer to do it; better treatments that prevent some deaths likely delay others; and earlier testing catching cases at milder stages will make the measured lag between case diagnosis and death longer.
In any given region, the only data you should really pay attention to are the numbers of people hospitalized with COVID complications and the numbers of people who are dying from it. Right now, the media are attempting to instill even more fear from the new confirmed infections in college students who have returned to campus. We already know those newly infected won’t die and won’t be hospitalized for it either. We knew this in March, too, by the way, but no one wanted to talk about it.
“We knew this in March, too, by the way, but no one wanted to talk about it.”
Fact check: TRUE
Remember when the rationale for Lockdown Socialism was “flatten the curve”? There has been very little commentary on the fact that this obviously changed, how no one quite announced or acknowledged that policy change in any official way, and whether there was any rational basis for the change – whether it made any sense in terms of some actually performed cost-benefit analysis of trade offs.
Originally, the rationale was to prevent healthcare resources from being overwhelmed, that is, to avoid deaths of people who could have been saved but for all the beds, machines, and personnel being fully occupied.
Under that rationale, you would throttle prohibitions and only force a level of lockdown restrictions to the extent you could keep the curve steady below some maximum threshold with some extra reserve slack on top just in case, like using prices to optimize electrical power plant capacity utilization. If the curve fell below that, you were restricting too much, or moving the goal posts to a completely different objective based on a different cost-benefit analysis.
A way to generate a very conservative estimate for that maximum threshold would be to infer it from the actual historical peaks, 60K hospitalized, 15K in ICUs, and 7K on ventilators (the use of which seemed to be mistaken in many cases and which were also said to be scarce and supply-inelastic which proved not to be the case). Today it is under 30K hospitalized (-50%), 6k in ICUs (-60%), and 1.5K on ventilators (-80%).
So, we have done much more than “flatten the curve”, it is far below that proposed target plateau, and yet the aperture for normalization (e.g., reopening schools for even mitigated in-person instruction) is not being relaxed proportionately with that.
Now the goal posts have indeed shifted to maintaining lockdown restrictions and keeping ‘cases’ down as much as possible until mass vaccination, even if most of those cases are not at all serious, (old sick people face *ten thousand times* more risk than young healthy people) and we now know extremely well who is most vulnerable and how to protect them.
Lockdown Authoritarianism apparently means never having to explain yourself. I propose Constitutional Amendments, at least at the state level, such that any period of emergency can only last 6 months before requiring public submission of the full case to an independent adjudicator, whether a court, a grand jury, or the people at large in a referendum.
“””Remember when the rationale for Lockdown Socialism was “flatten the curve”? There has been very little commentary on the fact that this obviously changed, how no one quite announced or acknowledged that policy change in any official way”””
I found that to be rather spooky – how places just seamlesses slipped into “zero Covid” mode after the curve was flattened.
One half-baked hypothesis I have is that some people got deluded into thinking zero-covid was possible because, post-lockdown(s), things never got really bad – no bodies piling up in the streets, etc..
My view is that this happened because covid is so much less dangerous and more widespread than initially believed. But if you didn’t hold that view, I could see you interpreting things as “wow, we got this scary/dangerous thing under control better than I ever thought possible. Let’s go the extra mile to get rid of it forever”
100%.
Here in Australia we have had the exact same transition. And I feel like I’m the only one who noticed. We went from “flattening the curve” which made perfect sense, especially given how little was known about actual IFR, to “eradicating the virus from the face of Australia” without any discussion of the change. Or any acknowledgment that this is the new policy. Everyone just fell into line. Most of my friends think lockdowns are absolutely necessary.
Here in Queensland, population 5M, we’ve had 6 deaths from Covid19. We aren’t locked down anymore but we have had borders closed.
In other news, the graphs of covid19 deaths are following influenza curves – northern temperate latitudes peaking in the winter and disappearing in the spring; southern temperate latitudes peaking in the southern winter. And lower latitudes and tropical regions having a much better time of it.
I recommend, Annual Review of Virology, Seasonality of Respiratory Viral Infections, Miyu Moriyama et al, 2020 for anyone interested in understanding this better. It seems highly likely that SARS-COV-2 has very similar transmission characteristics to influenza – based on what we see. And also it seems like government policy has little effect.
My nephew is sitting in his dorm paying $50k/year or whatever to take online classes (nothing in person). Apparently he just plays video games all day. Seems retarded. I asked him why he didn’t just take a gap year but apparently the college said that if you took a gap year this year your admission was revoked.
Infections not cases will tell you the progression of the virus. Just about everyone who attempts to measure infections uses an assumed ratio of infections to cases. Of course, as Pat mentions above, we don’t have a clue about that ratio. I think we can reason that the ratio has fallen over time because early on only people with symptoms had tests. Now, we test both asymptomatic and symptomatic individuals. So, we could have increasing cases and falling infections.
I don’t think the IFR (infection fatality ratio) has fallen significantly in the last 6 weeks, yet, assuming some constant ratio of infections to cases, we see a fall in the IFR. This implies, to me, a falling ratio of infections to cases. This could mean that infections continue to fall even though we see an increase in cases.
We should make a concerted effort throught random sampling, as Pat suggests, to get a handle on infections. We’re pissing up a rope without that.
Random testing was suggested, also, in early March by some people just to get a firm handle on the actual level of infections.
Is there even anything to explain? We know we only caught maybe 1 in 10 or 1 in 20 cases during the spring. If you correct for that, has the IFR changed much since then?
It’s important to remember that the story of NYC wasn’t one of a really deadly virus. We know now it was the story or a somewhat deadly virus that infected like 2-3m people in the city in the span of like 2 months. Until we see that level of infection we won’t see that level of death.
3. Yes, better treatment is available. In a recent study, intravenous dexamethasone combined with standard care reduced the patient time on mechanical ventilation by 2.6 days compared to patients who only received standard care. This saves lives.
There currently are 1,545 covid patients in the USA on ventilators according to covidtracking.com.
One wonders if the 700 deaths per day are due in part to unnecessary ventilation and or failure to use dexamethasone. The number of ventilated patients is low enough now that such cases should be overseen and monitored.
Mechanical ventilation causes additional lung damage but use of dexamethasone and blood thinners can get patients off ventilators. And the medication that must not be named keeps patients from getting hypoxia in the first place.
https://www.aier.org/article/open-letter-from-medical-doctors-and-health-professionals-to-all-belgian-authorities-and-all-belgian-media/
The great rush to put patients on ventilators probably increased the death rate substantially.
https://www.aarp.org/health/conditions-treatments/info-2020/ventilator-use-older-coronavirus-patients.html
One wonders if “first do no harm” has actually informed practice. One wonders too what the death rate would have been if affirmative action wasn’t a top admissions criteria for USA medical school admissions.
6. It’s summer time, and Vitamin D levels are elevated. So Covid deaths are down just like flu deaths are down every year. Deaths in Chile and Brazil peaked earlier in their winter and Argentina is peaking now. Stock up on your sardines and eggs now.
Florida Governor Ron DeSantis announced the next phase of re-openings in the state earlier today.
The response from the lockdown libertarians was predictable…
“Does the governor realize that if you kill your constituents, they can’t vote for you?”
https://twitter.com/asymmetricinfo/status/1309635671687155713?s=21