Covid 9/​9: Passing the Peak

Link post

Labor day muddies the data a bit, but it seems that Alex Tabarrok was correct and the current wave has peaked. We could well be facing another peak in December due to seasonality. We might also have issues from schools, although as you’ll see later they’re taking extreme precautions and also we didn’t see any sign of a school effect last year.

The primary question now is how and if we return to normality. It’s no longer a question of when. We’re going to be dealing with a substantial amount of Covid for quite a while, and a large number of unvaccinated people for quite a while, and our lives are ending one minute at a time. Whatever we are going to do to return to normal life, we need to start doing it, and if we’re not doing it, accept that actually we are and whatever we are doing is now normal. Either reclaim your life and the things that bring you joy, or accept you’re not getting them back.

Note: This week’s post was written using LessWrong’s editor. Hopefully this solves the issue people were reporting with being unable to load images. Be quick to point out any remaining issues. I’m working on being able to paste Excel charts properly with the new tech stack.

Executive Summary

Top points this week:

  1. Covid case numbers peak.

  2. Australia’s dystopian nightmare deepens, hopefully we can avoid this.

  3. Schools are acting crazy in name of prevention.

Let’s run the numbers.

The Numbers


Once again, I forgot about an upcoming holiday, in this case Labor Day, and its tendency to screw up reporting. I think it’s because holidays don’t seem ‘real’ to me as I’ve been working from home on my own schedule most of my life.

Prediction from last week: 1.1mm cases (+5%) and 11,150 deaths (+20%).

Result: 940k cases (-9%) and 10,272 deaths (+10%).

My guess is that the 10% drop on deaths is a reporting issue, and thus the cases also have a similar reporting issue and were about flat. This is still good news, but we shouldn’t expect a bigger drop next week or anything like that.

Prediction for next week: 940k cases (+0%) and 10,800 deaths (+5%).


Sorry that this is an image, couldn’t get this to look right otherwise.

This is an excellent deaths number, but the difference from my prediction is likely the result of Labor Day rather than less people dying. We’ll find out next week.


Again, sorry this is an image. Working on it.

Was this the peak? I think it likely was a peak, whether or not there’s another coming in several months. The question is how much of this was a holiday reporting/​testing issue versus a real decline, which always creates more uncertainty in the following week.

The Johns Hopkins data confirms this. The share of positive tests is down from peak, which makes me confident the true peak has come and gone.


How much did full FDA approval matter? This is one measure.

That might not sound like much but the answers to this question have been highly stable over time, so it’s a big deal.

I don’t think you can quite do this calculation…

Then again, you also kind of can do a version of that calculation. Ten million adults went from unwilling to vaccinate to willing. That doesn’t mean they run out and get it overnight, but if we assume their behavior shifts into the group they now place themselves in, a lot of them will now get the shot with substantial probability each week. I expected it to be high given the assumptions I assumed were being made, but a hundred dead people per week seems to be a good Fermi estimate.

About the distribution of our surplus vaccines, yeah, we might want to get on that.

Thailand develops robot to get 20% more doses out of vaccine vials. Presumably this will mostly be ignored.

Today’s edition of ‘this, but unironically’:

If I was trying to convince someone that the Earth was not flat, as opposed to being smug and angry about them not believing it was round, yes, I’d actually want to understand their reasoning and respond to their true rejections. What would be the alternative? Act smug and remind them how stupid they are being and how I have a lot of social proof until they give up? If that would work it would have worked already.

The point though is to know the actual reasons. When we find out some of those reasons are really stupid, though, we shouldn’t flinch from that.

Vaccine Effectiveness

Another study has come out showing vaccines greatly reduce spread of Covid even when someone is infected. Link to study here.

Statistics like this remain the most puzzling thing about the whole situation (study).

I’d quibble a bit with how they describe their results, but either way that’s quite a lot of antibodies. They say 20.2% of the population has antibodies from infection and the rest are from vaccination, which presumably is the result of giving vaccination full credit for all the antibodies of anyone who was vaccinated. Reinfections are currently thought to be extremely rare, vaccines clearly still work, and this is saying roughly half of the unvaccinated had a previous infection (with sustained antibodies) by May. The calculations can still work, especially if you presume that very little effective non-vaccine prevention is still going on.

This post goes over how effective the vaccines are. Mostly it’s good information and not far from my estimates, but it’s important to emphasize that no, we didn’t all know all along that vaccine effectiveness against mild but not serious illness would fade over time in about this way. I don’t even think we should be confident in that finding now. This is not a case of people ‘sending the wrong message’ or anything like that. This is a case of finding out something that was plausible but not expected, and then retconning to claim it was knowable and known beforehand. It wasn’t, and the line ‘of course it was always going to be like this’ is being used as a bludgeon in the hopes it will get everyone to meekly accept the new official facts – they’d rather pretend to have been lying because everyone knows they lie all the time already, I suppose.

If we develop new vaccines, or new versions of vaccines, how will we know if and how well they work? It is increasingly plausible we won’t, because the ‘ethics’ people and the logistics and regulatory departments will combine to prevent any useful studies from being run unless something changes. We need the ability to run comparison studies and we don’t currently have it.

You know what’s bad for your fertility? Covid-19. The latest data point on that is that getting Covid doubled the rate of stillbirths in Mississippi for the unvaccinated.

FDA Delenda Est

I could call this section ‘booster shots’ but at this point the alternative title seemed more appropriate. The FDA is determined to prevent booster shots. Several officials resigned last week over this issue, and the remainder are holding firm that one does not simply get a booster shot, one must get proper approvals through proper channels after the appropriate delays, and if you don’t ensure the delays are proper you are not following the science.

How is the FDA doing that? Didn’t the FDA promise quick and painless review for vaccine revisions and booster shots? Doesn’t mRNA provide the ability to update quickly, within days or at most weeks?

Oh, that won’t be a problem, all you have to do is apply standard procedure. Most people in the United States aren’t even legally allowed to be told what strain they were infected by because that requires extra approvals so cumbersome no one will ever get them, you think we’re going to let them get fully vaccinated that easily?

Mostly you should feel lucky we let you get vaccinated at all.

One tactic, once you’re temporarily done accusing people in charge of making decisions with ‘playing politics,’ is to ask explicitly political questions completely outside the FDA’s purview, because if you’re a veto point you get to decide why you want to veto.

You don’t need that though, as MR notes you can simply say no.

Top federal health officials have told the White House to scale back a plan to offer coronavirus booster shots to the general public later this month, saying that regulators need more time to collect and review all the necessary data, according to people familiar with the discussion.

Dr. Janet Woodcock, the acting commissioner of the Food and Drug Administration, and Dr. Rochelle P. Walensky, who heads the Centers for Disease Control and Prevention, warned the White House on Thursday that their agencies may be able to determine in the coming weeks whether to recommend boosters only for recipients of the Pfizer-BioNTech vaccine — and possibly just some of them to start.

What’s wrong with the Moderna booster application that they’re holding up for months?

Moderna wants to do boosters half the dose of its original shots, because (1) the original doses were too high, (2) the higher doses are even less necessary when you’re triggering a secondary reaction and (3) I don’t know if anyone noticed but there’s a shortage of vaccine.

Nice try.

Yes, he means 100ug and 50ug not 100cc and 50cc, and he’s very sorry.

The FDA responded no, we won’t allow that and we’ll hold up the whole process on that basis, because (1) if you do half doses then there might be waste, which is not going to get less perverse or absurd if you read it again or think about it for longer, I mean wow, and (2) because the half doses might ’cause confusion.’ It seems measuring different doses is beyond our medical system’s capabilities, so we’d better insist on giving people doubled doses (that cause higher short term side effects) to prevent that ‘confusion.’

That. Is. Not. Science. This is something straight out of the CIA field sabotage manual, and should be treated as such. Burn the place to the ground. Salt the Earth.

The go to move, of course, is to ask for more data, but the “bioethics” people were on the scene to ensure that the previous trials weren’t extended, ensuring the option to demand data that takes arbitrarily long to gather, since without it there is ‘no compelling reason’ to require a booster shot, especially for J&J, where such a booster would be the same second dose everyone got in the first place and makes very little biological sense.

Eric Topol tried to convince such people that having had Covid should count as a dose of vaccine, since in terms of physical outcomes it totally does, and got nowhere. No matter the source, I appreciate the attempt.

The same issue is preventing revisiting initial doses, which were educated guesswork by various companies, and are now treated as Word of God, with no option for the good lord to change their mind on new evidence. Any sane world would run further experiments to find the right doses, but that is definitely right out.

Meanwhile, in Israel, power speaks truth and says to prepare for future boosters.

Another good note is that while the vaccines work well against Delta, a further mutation could change that, and updating the vaccines for Delta would help minimize that risk. That’s not how the FDA thinks, so crickets.

Vaccine Mandates

Biden to make it official and require vaccines for all federal workers.

The ACLU needs to change its name, since it seems to be taking an increasingly consistent set of anti-civil-liberties policy positions. It usually justifies this by saying that helping some preferred group of people is its new definition of civil liberties. Here they are on vaccinate mandates, as opposed to its previous position that coercion was bad actually.

That is… not how civil liberties work. That’s not how any of this works. The ACLU’s entire job is supposed to be to stand up and say ‘no you can’t tell people what they must or can’t do, I don’t care how many people want to do exactly that, we are supposed to have civil liberties’ even when the correct response is ‘no obviously we need to do this thing because of this compelling government interest and actually we have the right to do so’ and when that case is strong enough they’re supposed to lose but keep fighting the good fight. Or alternatively, if they wanted to argue for allowing private mandates, they could make the case (as I do) that allowing freedom of association, and choice in when to associate with someone, is actually protecting civil liberties, but they’re not making that argument. They’re making the arguments that (1) forcing people to do this causes good things to happen so it’s good and (2) if we force people to do these things we would stop taking away people’s other basic liberties, not that we’re trying to stop that, what was the name of our organization again?

We could really use an American Civil Liberties Union. It’s a shame that we don’t have one.

Vaccine mandates for NFL game attendance have broad support, especially among those who attend games, in the sense that they make people more interested in attending games far more than they make them less interested.

If you regularly attend games, and aren’t vaccinated, presumably a mandate makes you less interested, given you would not be allowed to attend. Thus, we can presume that NFL game attendees are mostly (>80%) vaccinated. And looking at all the numbers, it seems safe to say that among the vaccinated, requiring vaccination is very good for business. From what I’ve seen, this mostly extends to most other businesses as well.

Nate Silver did a highly virtuous thing, and laid out his views explicitly.

This would be even more virtuous if Nate’s views were more out there or less coherent. As it happens, I am very much in the same camp on pretty much everything on his list.

One important note is that if you are going to have a vaccine mandate, it’s important to let those impacted know as soon as possible, or you’re going to make the existing supply shortages that much worse. For example, New York school bus drivers, where there was already a shortage, and who absolutely should have a vaccine requirement and found out about this last week. Why is there a shortage? I’m guessing this attitude has something to do with it, and would suggest perhaps that if you want more people to do things like drive a bus around you should try paying them more money:

“This is a public service job,” Christopher said. “People don’t become bus drivers to get rich. They do it as a part time opportunity and many of them do it because they like kids, they want to help their school districts and communities. It is a community job that is important.”

NPIs Including Mask and Testing Mandates

Michael Mena tries one more time to explain the difference between rapid and PCR tests and what they are measuring. He’s been saying this for a long time over and over, I’m listing this because it seemed like an unusually clean explanation.

Alas, to the extent that we do contact tracing, quarantining or worrying about infection, we’re not looking at levels of infectiousness with this kind of attention to the physical world. Even people I know personally usually fall back on heuristics that make them feel comfortable, rather than doing what would be effective. Even now, much of the testing we do carries multi-day delays, which destroys much if not most of the value they provide. Whereas in some other places, such as Germany, tests are everywhere in droves, and either free or almost free, often using American-made tests.

Meanwhile, our “experts” continue to be maximally paranoid and conservative:

The last one is distinct from the others, and in many ways the most interesting. I have yet to see a coherent rationale for no-second-dose-for-J&J-but-two-for-everyone-else position, at all, yet I would be shocked if even one such physician told those who got Moderna or Pfizer to stop after one dose.

The ‘experts’ also don’t understand that their attitude towards risk is not normal, or that regular people think life matters, or that people can connect dots and draw conclusions based on those dots even when those dots are misplaced. Nate’s hypothesis is that many people are assuming that the ‘experts’ are saying sane things rather than insane things, and drawing the appropriate conclusions:

It is a very understandable mistake for ordinary Americans to largely presume that their experts are not acting crazy, but at some point it’s important to notice that such presumptions are wrong. And yes, I do think that the people who say ‘avoiding such things is a small cost’ or ‘you can do all that at home, and cheaper too’ are at minimum being rather clueless about how people live. Basically this kind of energy:

Usually attributed to John Lennon, which is both too good to check and unlikely to withstand one.

Of course, the teachers are correct, whichever child said this did not understand the assignment. The difference is that they saw this as a problem to be solved.

The Tragedy of Australia

Australia continues to prove that there are worse things you can choose over getting Covid. There’s a new article in The Atlantic detailing how far they’ve gone, and the answer is way way way too far. You can’t enter, you can’t leave the country, you can’t leave your house without an Official Approved Reason, all enforced by the military.

Australia’s way of life has been put on indefinite hold. What did they get in exchange?

A year of house arrest to stop a 1 in 500 chance of death naively implies that the QALY value for a year under house arrest is at least 0.92 or so, and since Covid primarily kills the elderly that calculation is highly generous and it’s more like 0.96. Was it worth it?

More to the point, will it ever end? We’re getting mixed messages on that.

The calculation on having these types of controls in place, even for a period of months to a year, look terrible to me. If I had to choose between a year or more of this but no one gets Covid, or alternatively essentially everyone getting Covid before they’re vaccinated (over time, not actual everyone gets it at once and the hospitals crash), I’d choose essentially everyone getting Covid.

Meanwhile, Australian authorities have taken upon themselves full surveillance state powers, including the right to delete or modify social media posts during investigations (also known as ‘when they feel like it’) without informing anyone including the target. They can lock you out of your account, post new things, take down or modify other things, pose as you, and the only way you find out is when you can’t log in or notice the changes. So if you see a social media post or almost any other electronic communication by anyone in Australia, going forward, you cannot be confident that what you see is the communications of the person who you think owns the account, let alone that you are seeing all the things they intended to say, or that they will be the one who sees your reply.

Could it all happen here? What thanks do we owe the people, however frustrating and troublesome, that prevented something like this from happening so far?

Tyler Cowen explains why the defenses of Australia do not persuade him. He points out that Australia is not picking an efficient point on the safety versus liberty frontier, and making the best deal it can make. Instead, it is imposing these draconian restrictions and suspending normal life, while not doing basic cheap prevention or treatment including not pushing for rapid vaccination and not preparing for treatment. A series of email exchanges between Australia and Pfizer details now non-urgently they treated the problem.

I agree with that as an easy and conclusive rejection, but I don’t see it as the true rejection for two reasons. One is that one can point out huge, civilization-crippling mistakes but that doesn’t make actions conditional on those mistakes right or wrong if you assume that for whatever reason those mistakes can’t be fixed. This is far from a sufficient or general excuse, but that leads to the second and more important reason I don’t see this as the true rejection, which is that this isn’t necessary. If those additional interventions were impossible, would it change my view here? Not much.

Think of the Children

Consider the implicit assumptions here. Guidelines being commented on are here.

Lip service is being given to safety in class, but mostly parents are (correctly, if you believe school is good and important) viewing the problem as how to protect their children from being forced to quarantine. Testing your child isn’t a responsible thing to do, it’s an antisocial weapon because if the child tests positive it will require quarantines. The social pressure will actively be to prevent testing, at least in many cases, because the test has negative external value of information.

Still, if it does come to a quarantine, please do not threaten to zipline the principle. Although as we’ll see below, it will sometimes be tempting.

On the other hand, I’m not saying you should threaten to zipline anyone, but the rules are getting rather absurd in places. Imagine the reductio ad absurdum version, and then check this out.

Yes, that means that if any child says they have essentially any symptom at all, you lose an entire classroom for a quarantine period.

If you think you can run a school where a single child in a classroom claiming to have a headache forces the entire class into quarantine, well, good luck with that, and please gather detailed data on what happens and hopefully turn it into a sitcom later. I’m imagining the game theory and social dynamics involved and they do not involve learning or preventing infections but they are delicious. Is a child even a favorite to be in school on a given day?

Meanwhile, Biden is calling upon all schools to institute regular testing, of asymptomatic kids, and yes our child’s preschool is doing this, requiring a test every week for every child, and it’s exactly as insane as it sounds.

Is the point of childhood to prevent Covid? It seems like they’re acting as if the point of childhood is primarily to prevent Covid.

You’d think that by now, given all that, we’d have figured out reasonable things to tell someone who thinks they have Covid and is getting tested or has tested positive. Nope, definitely not here, sorry.

I will always be astounded the extent to which people think time stops if it does not happen to be 9am-5pm on a weekday. I will also be astounded of the extent that people think universities are supposed to be this special place where all this stuff is taken care of, how dare you ask college students to act like adults and handle their own lives, as opposed to the rest of the nation where (for example) contract tracing isn’t a thing at all and you’re supposed to take care of yourself. College really is a bizarre package of goods.

For an example of going completely off the rails in the other direction, may I suggest Connecticut College, where everyone is fully vaccinated. I don’t really know what anyone involved was expecting.

The most core error is here, in a classic case of one of these things is not like the others:

Worrying about students going to bars is highly reasonable. Worrying that they’re socializing in cars, friends’ rooms or apartments is different, but is being roped into the blame department. It’s also sending a clear message to students that the rules are about as worthy of respect as the rule that if they’re under 21 they shouldn’t drink.

Maybe We Should Try To Prevent Future Pandemics

Biden wants $65 billion for what he is calling a pandemic preparedness Apollo program. This is a great idea if it can be executed reasonably at all, even if only a small portion of funds end up going to this reasonable execution. The question is, can we execute, or are we too dead as players to do so? The first test is whether such a thing can pass at all even in the wake of a pandemic. The third test, if we get that far, will be to see if the money gets stolen (either outright through corruption and things that can’t possibly physically work, or reallocated) and the fourth test would be whether we then let even what we do build decay.

The fifth test, of course, would then be whether, if we had useful pandemic preparedness, would we be able to actually use it? Or would the FDA and CDC, combined with various partisan bickering, short term outlooks, accusations of racism and other such things, prevent us from deploying our tools even when we have them? Given what happened this time, this is not an idle question. We had our vaccines in a day, and could easily have known they worked within a month. So even if we did have a system that gave us our vaccines in a day and scaled them up in a week, would it even do anything useful?

You’ll note I skipped over the second test, which is to look at what’s in the actual proposal. Let’s take a look.

This is an immense boost in pandemic spending, the stimulus was only $6 trillion

Sounds great when stated like that, but there’s an obvious problem, which is no plan to pass test number five. What good is having vaccines you’re not able to use?

Same problem with tests or treatments:

It’s all good, counting this caption I said Safety three times.

We can add a sixth test, I suppose, for ‘doesn’t fund a bunch of research likely to directly cause a pandemic’ and this reminds me a lot of the people who say the words “AI Safety” as if once you say the word Safety and decide that you’re going to take a Safe action instead of an Unsafe action – at some nonzero cost to you! – that suddenly hard part’s over and you’re Safe, as opposed to having no idea what an actually Safe action would actually be while some of us despair in the corner mumbling ‘none of this matters we are all going to die.’ Yes, I’d prefer to have some Lab Safety Measures instead of no such measures, but do you have any faith that they’ll be sufficient? Yeah, me neither.

Either way, to be very clear, trying to do this is very good, and it will be a good test of whether we can still be live players.

So you’re saying there’s a chance.

None of these concerns mean we shouldn’t do this. We should absolutely do everything we can, even if we expect the FDA and CDC are going to act next time around as if they want as many people as possible to die, and stall things as long as possible, because we’ll still be able to stop people from dying at least somewhat sooner and more effectively regardless, and who knows we might even do something to fix those problems.


The media has not, shall we say, covered itself in glory when dealing with all things Ivermectin.

The poison control story turns out to be rather exaggerated.

The mistake here, as per the correction Will links to, is very much not a good look.

Someone really, really does need to be fired for this. This kind of error is very much Not Okay, and results in the following very reasonable perspective, except about everything all the time.

How easy was this to catch? I kind of caught it purely because it didn’t make sense, and I think I should get a highly mediocre grade for not calling bullshit further. This is what I wrote last week:

Also note the numbers. One individual was told to ‘seek further evaluation,’ and 85% of the cases were mild. The definition of ‘mild’ can be whatever people want it to be, but if it’s ‘no need to seek further evaluation’ it seems like there were six poison control calls out of eight total calls? I’m guessing it’s higher than that, and please if you decide to take Ivermectin make sure you’re sourcing and dosing it safely and properly, but this isn’t an epidemic of cases, and this was going around enough it felt important to point that out, even if I’m highly skeptical that Ivermectin does anything useful.

I totally should have then jumped to ‘these numbers are not what they appear to be, whether or not they’re blatant lies,’ and I’ll work on making sure that’s how I react next time. I still highly doubt Ivermectin does anything useful against Covid, and certainly the fraudulent studies don’t help matters on that front (here’s a thread about what might be going on there), but would the media and health officials be acting differently if I was wrong about that?

In Other News

NYC Mass Transit continues to be down while car trips are up, I understand this trend to be typical of other regions.

Once habits shift they can be hard to shift back, and it now feels ‘responsible’ even if vaccinated to take a car instead of the subway, and less blameworthy, and now we’re stuck with it. People even take taxis over it, and I’m confident that if you’re cutting back on mass transit you should dramatically cut back on taxis. I wonder to what extent mask mandates on mass transit are making it sufficiently less pleasant that people are avoiding it. I’ve noticed a nonzero effect size there for myself.

Bryan Caplan points out that what we implemented under Covid can best be described as Federalist Dictatorship. Within each state, the governor essentially decided what to do and then that’s what happened, regardless of whether it made sense or was within his right to do so or any of that. As a result, we had a bunch of natural experiments, since many governors were in states that could have elected one from either party, and a bunch of them took idiosyncratic action.

There’s a recall in California that is remarkably close to succeeding and likely resulting in a Republican governor (and that if it did succeed would cost me a little money on PredictIt but as compensation also be hilarious), which is being caused by the current governor maintaining sustained onerous Covid restrictions while also being a giant hypocrite about it. So Obama steps in, describes the alternative as ‘putting kids at risk’ and even more nonsensically as ‘preventing recovery’ and a friend for whom such a rant is very out of character has some thoughts.

Not Covid

On the rules of the TSA, our future in all things if we are not careful.