Covid 8/12: The Worst Is Over

Good news, everyone! Andrew Cuomo has resigned, and Andrew Cuomo is the worst. 

I will damn well take it, because it’s not like he doesn’t also deserve to resign in disgrace for the stuff that officially got him, and again, also to say it for what is hopefully one last time, Andrew Cuomo Is The Worst. Hence, The Worst Is Over. Sing it high, sing it low. (HT: Meme source)

The title this week does not as reliably or fully refer to the Delta variant or the Covid-19 pandemic. Things are still steadily getting worse. But the turning point is plausibly in sight, as case growth slows, and I doubt we have more than one doubling left before things peak. 

Main event this week was continued arguments over mandates, both for vaccines and for masks and other NPIs. I’m making one last attempt this week to explain my reasoning on vaccination mandates, as I continue to get people disagreeing for a variety of reasons, and disambiguating the disagreements seems worthwhile; I tested it out in the comments last week and it seemed productive. 

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week: 855,000 cases (+45%) and 4,100 deaths (+40%). 

Result: 744k cases (+26%) and 3,725 deaths (+35%).

The case numbers are very encouraging. They’re still increasing, and there’s always the chance this was a data fluctuation that will be actively undone, but it probably wasn’t and represents either the control system kicking in or us nearing a natural peak. I still expect a similar increase next week, but I’d estimate a 35% chance that within two weeks we get about as high case counts as we’re going to see in this wave, and 55% it happens within three weeks.

I’m still predicting a +35% rise in deaths, as the cases from the last few weeks make their way through the system, and hope to be pleasantly surprised.

Prediction for next week: 900k cases (+21%) and 5,028 deaths (+35%).

[Note added 8/19: 5,028 deaths is actually a prediction of +48%, which would have been more accurate, but the intent here is clear from context, and I meant to do +35% and somehow gave an incorrect number. So even if it’s a better prediction, it’s unintentional, and I shouldn’t get credit for it. Damn. ]

Deaths and Cases

Grouping these together, because there’s a combined mystery to solve. 

Here’s the case numbers.

Date	WEST	MIDWEST	SOUTH	NORTHEAST	TOTAL
Jun 17-Jun 23	23,854	12,801	26,456	6,464	69,575
Jun 24-Jun 30	23,246	14,521	31,773	6,388	75,928
Jul 1-Jul 7	27,413	17,460	40,031	7,065	91,969
Jul 8-Jul 14	45,338	27,544	68,129	11,368	152,379
Jul 15-Jul 21	65,913	39,634	116,933	19,076	241,556
Jul 22-Jul 28	94,429	60,502	205,992	31,073	391,996
Jul 29-Aug 4	131,197	86,394	323,063	48,773	589,427
Aug 5-Aug 11	157,553	110,978	409,184	66,686	744,401

This is a substantial slowdown in new cases, cutting the growth rate in half. We might be peaking within a week or two. 

Here’s the death numbers.

Date	WEST	MIDWEST	SOUTH	NORTHEAST	TOTAL
Jun 24-Jun 30	550	459	706	186	1901
Jul 1-Jul 7	459	329	612	128	1528
Jul 8-Jul 14	532	398	689	145	1764
Jul 15-Jul 21	434	341	732	170	1677
Jul 22-Jul 28	491	385	1009	157	2042
Jul 29-Aug 4	693	477	1415	304	2889
Aug 5-Aug 11	705	605	2181	234	3725

(Note: California reported -352 deaths yesterday, I changed that number to 0. Last week Delaware dumped 124 deaths on us, presumably a backlog, which is why the NE number last week was so high. I’m choosing not to correct for that for now but might smooth it back for next week as it is potentially importantly misleading.)

The death number was modestly better than I predicted, but definitely not good. This NYMag article (sorry about linking to Topol so often, world is strangely small) highlights the negative perspective of the American death rates not dropping this wave the way they did in other countries. Yes, the CFR is down a lot, but the bulk of that is lag due to cases rising so rapidly. If you undo the lag, you see something else…

[Lagged = cases from 3 weeks ago. Lagged and averaged = cases from 0.1*(1 week ago) + 0.2*(2 weeks ago) + 0.4*(3 weeks ago)+ 0.2*(4 weeks ago)+ 0.1*(5 weeks ago)], chosen quickly to be not crazy. Originally had 0.5 instead of 0.4, which was dumb, it is now fixed.]

At the very beginning the CFR was much higher, but once we got adequate testing and reasonable care, things haven’t changed much on this chart. This is not at all what we see in the UK, where the CFR is clearly down a lot.

The uptick at the end represents the UK having declining case counts, rather than a higher death rate. The Netherlands looks similar to the UK but made the graph hard to read.

A simple theory is it has a lot to do with this, the positive test rate:

We went from 2% positive tests to 15% positive tests, which indicates our testing is not keeping pace with our cases. In fact, it’s not keeping pace at all?

Seriously, what the hell, people? It’s one thing to not keep pace, it’s another thing to be testing less now than we were when cases were at their lows. That means that, conditional on not having Covid, the chances of a given person getting tested have gone down a lot, despite their reasons to worry they have Covid having gone way up. Much higher chance they have a known contact with someone positive, and much higher background danger.

One possibility is that people are now increasingly using at-home tests first. We did this last week, when our four year old had a cough and we were asked to exercise an abundance of caution – it was annoying to stick that thing up his nose for a while, but it was cheap and quick, and got the job done. Whereas when we looked into getting a non-rapid non-home test, it was clearly going to be a pain in the ass to get. One of my friends posted a similar experience, where she couldn’t get her child tested for several days unless she got a test at a pharmacy. Both of us got negative results, and neither result counts in the charts above. If one of them had come back positive, it’s not clear how often that would have made it into our statistics either. If there was no need to escalate to official medical care, there’s no default mechanism to get those results into the statistics. 

Whether or not that’s the main mechanism, I am confident that a large majority of cases, especially asymptomatic or mild cases, are being missed entirely by the system. The true IFR likely has not dropped here quite as much as in other places, but our vaccination rates are not that much lower and our medical care is quite good and holding firm, so it’s likely falling almost as fast here as elsewhere. Our CFR staying high says more about our rate of case detection than it does anything else.

In some ways, this undercounting of cases is very good news, as it puts an upper bound on how bad the pandemic can get and moves us more rapidly towards the turning point. In terms of hopes for containment, however, it’s very bad news. If cases have grown much more rapidly than our numbers naively indicate, then our hopes for containing this in a meaningful way, rather than waiting for enough people to get Covid that things die down afterwards, are much worse. 

Vaccinations

Here’s Zeynep, on point as always, on vaccine hesitancy and the reasons behind it, which are far more reasonable than they’re usually made out to be. 

Also, it turns out that having a Bachelor’s or Masters makes you very likely to be vaccinated, but if you go on to get a PhD, you’re now in the least vaccinated group of all. 

This week’s thread on relative infectiousness of the vaccinated versus the unvaccinated. As you would expect, when you do random population samples, you find lower average viral loads in vaccinated people. 

The can-do spirit:

Also, perhaps we could try publicizing this more? Ask 24 out of 25 doctors if vaccination is right for you? Maybe take the crosstab of Republican doctors which is presumably still over 90%? Playing to win the game takes many forms.

Baltimore’s anti-Covid ad posters. No idea if they’re effective but at least they’re playing.

The can’t-do spirit:

This seems to me like it should be a rather large scandal. If people in Alabama don’t want the vaccine, and it’s going to expire, then it’s imperative to recognize that and send it elsewhere where people do want it. Worst case, you cycle out old vaccine to places cycling through what they have, and then move in new shipments so there’s still stock around when and if people change their minds. Also, ‘expired’ vaccines might not be good enough for us, but they doubtless mostly still work, so I’d still send ‘em off to whoever wants them. 

Yet I see almost no discussion of such matters. Shame!

The Isreali can’t do but therefore do what you can spirit:

And the can’t get paid enough spirit, I too blame capitalism for all these vaccines:

Vaccine Mandates

Nate Silver broke down people’s current stance on Covid into a categories, and I think this is mostly accurate:

Jacob offers an alternative theory of Group A:

My guess is this is indeed a substantial percentage of Group A, but I’d be rather shocked if it was anything like a majority.

One thing we can do is compare this to last week’s survey results on mandates, which found that 62% favored the maximally coercive policy of a universal vaccine requirement. That would roughly be Groups A+B, which is 55%, and some of Group C1, then only C2 and some of C1 are vaccinated but don’t then favor a full mandate on others. 

There aren’t that many people in the daylight between ‘willing to get vaccinated myself’ and, well, this:

I’d like to take one more stab at addressing various arguments against vaccine mandates, explain my perspective on it, and then move on to what’s actually happening. 

There are a bunch of different arguments against vaccine mandates, and it seems the only one that convinces a lot of people is the argument that (A) one shouldn’t take the vaccine, presumably (I’d hope, anyway) with the implication that the vaccine also shouldn’t be mandatory. 

(B) would be that bodily autonomy is super important, and thus vaccinations are in a different class from things like masks and lockdowns, and either (B1) allowing this is an escalation of authoritarianism, and will greatly expand authoritarian power and perception and lead to tyranny and/or (B2) this is going to piss a lot of people off, destroy their dignity, is generally bad on object level, and so on. And the claim that (B3) an employer mandate is authoritarianism and against freedom, as opposed to all the other things jobs require, and thus we should suspend freedom of contract or punish those who use it in this way. 

So basically not buying into arguments like this:

There’s a lot of vocal support for both of these, both in my comments and in general. I’m sympathetic to these types of arguments, but not convinced, because mandating vaccinations to stop infectious disease has correctly been standard procedure for a while, and such precautions as a condition of employment or close physical proximity to many others are exactly how free people react to such situations. And if the choice is between ‘no indoor dining (or other X) for anyone’ and ‘no indoor dining (or other X) for the unvaccinated’ I know which one I’m choosing, and which one leaves me more free.

There was a lot of talk earlier worrying about tracking, but we’ve moved past vaccine passports and people just flash their cards, so that’s not a concern, and also a common alternative is ‘contact tracing’ which involves keeping a lot of records of exactly the type we’d worry about. Instead the concern here is the flip side of that, which is that (C) vaccine cards are easy to fake. That’s true, but also there’s a central database that knows the answer that already exists. I’d rather not push people to lie and commit fraud, but this doesn’t seem like that big a concern here.

A real concern is that (D) they got this one right, vaccines happen to be safe and effective, but it’s not clear that we wouldn’t be in a similar position in the future where the thing in question wasn’t safe and/or wasn’t effective. In this case, I actually don’t think this is true. I think both that the vaccines are safe and effective based on the evidence, and also that if the evidence did not strongly say they were safe and effective, we wouldn’t be contemplating such policies. The level of pushback we have now is when, scientifically, the case is overwhelming, and if the vaccines were instead not safe but still much safer than not getting vaccinated, we’d not only not make them mandatory, they’d be forbidden. We ran that experiment.

Finally, there’s (E) that more vaccinations won’t change the path of the pandemic at this point, so why are we bothering, choices have consequences, the unvaccinated will get sick but the vaccinated will mostly be fine, so all we’ll be out are a few medical bills (which will effectively get socialized, because insurance can’t discriminate and if you don’t have insurance mostly the government pays). I don’t buy this, there are a bunch of immunocompromised people, even for vaccinated people getting Covid is worth avoiding, and in practice if there’s lots of Covid out there the result will be lots of lost living of life. 

Another disagreement I’ve found is (F) the idea that an individual is only responsible (in various senses) for either literally themselves, or only those who that individual infects directly, rather than the marginal cost being every infection that results from your actions that wouldn’t have otherwise happened (there’s a control system, and some people would have gotten infected later anyway). 

Finally, there’s (G) that the mandates are covering people who already had Covid and thus don’t require vaccination. There’s the counter argument that vaccines are more effective than immunity from prior infection, but antibody tests could check and even worst case I doubt prior infection is that much less effective than J&J which we consider to fully count after one shot despite it not making much logical sense. I don’t find that convincing. I also don’t think the evidence that vaccination after infection is still worthwhile makes this sufficiently effective to convince me either, although it’s still a substantial additional reduction and I personally would still get the shot. What I do find convincing is that lots of people are wrong about whether they’ve had Covid, or would fool themselves or lie about it, and thus there’s no reasonable way to make this (otherwise correct) exception short of getting a confirmed antibody test, and the complexity costs and messaging and such involved make it not worth it. Sometimes you gotta suck up stupid things in the name of simplicity, but of course if anyone does want to make such exceptions that seems totally fine.

The high correlation of positions on all these points is, of course, both expected and suspicious, on all sides including my own. 

Other times, incentives matter, and people enjoy using hyperbolic language. I don’t really know what he was expecting. 

One cost of lack of vaccination is putting more strain on the hospital system, as they once again are forced to cancel elective surgeries, thus reallocating medical care from those who need it because life to those who need it because they are unvaccinated. The cost here is not purely monetary. 

In conclusion, I’m strongly in favor of employer mandates, and on imposing the kinds of restrictions we’d otherwise impose on everyone (e.g. travel, indoor dining and so on) only on the unvaccinated, although of course we should be smart about it, and I’m happy that so far no one I’ve seen is suggesting excluding the unvaccianted from beaches or playgrounds. 

I’m definitely in favor of letting gyms mandate vaccinations rather than masks, as opposed to being required by law to mandate masks instead as they are in Washington DC. 

For now, lack of FDA approval is holding many mandates back, even as increasingly many go forward anyway. 

If I knew the FDA was going to get this done in a few weeks, I’d be inclined to announce new restrictions but make them conditional on full FDA approval, so as to benefit from the cover that will provide and give them that much more of a nudge to hurry. Alas, while I hold out hope for it, I currently have no faith in that timeline, and we don’t need them to tell us what we already know. 

I now consider that out of the way, and won’t be discussing it further unless something changes. So, what are this week’s new mandates?

Here’s an interesting local one that turns out not to be the full corporate policy, and a reminder that this is how attention works on the internet, and it looks like about one out of every thousand people who saw the original post saw the clarification that it wasn’t the whole chain.

CNN’s mandate showed it has teeth. Which is how it has to be. Once you have a mandate, for many reasons it needs to be properly enforced.

Meanwhile, the question as always with children is, should we forbid vaccinations as horribly unsafe, or should we stop doing that and turn on a dime to mandate them outright, with not vaccinating as horribly dangerous?

Yelp is doing something interesting, which is that you can search restaurants by ‘All Staff Fully Vaccinated’ and by ‘Proof of Vaccination Required.’ I verified this, while also noticing that Yelp’s rating sorting algorithm leaves something to be desired (e.g. having a 5-star average on 2 reviews seemingly puts you in the top 10 for all of NYC, whoops). This is sufficiently cool that I’m going to play around a bit more with Yelp, since I’m working on my list of places to go in NYC once I get back anyway. 

The San Francisco Sheriff’s Deputy warns that when the full vaccine mandate is imposed, officers will quit en masse, whereas they wouldn’t if they only had to have a swab up their nose every week instead like the state suggests:

I wonder if this is a place where approximately zero people would be sad to see many of those 160 deputies go. The blue tribe locals are anti-police and will see these as bad actors and even outgroup members, and be happy to see them go. The red tribe will see this as the latest talking point in what they see as SF’s descent into crime and anarchy. The rest of us find out how many officers actually quit, which will be great data. Everybody wins?

In contrast to NYC’s teachers, the head of the American Federation of Teachers came out strongly in favor of a mandate.

France implements its Health Pass requirements with little fanfare, despite weeks of protests (WaPo). Sounds right.

Think of the Children

The request here comes from the American Academy of Pediatrics, representing 67,000 physicians.

As Alex Tabarrok puts it:

It is quite the rebuke. Well said. When concerned physicians tell you to stop demanding so much child safety data and Get On With It, that is the opposite of the mistake they are most likely to make, and thus this is strong evidence that a lot more Getting On With It is urgently needed. 

This thread points out the obvious, which is that anything picked up in the large sample, that wouldn’t have been picked up by the small sample, wouldn’t be big enough to make the vaccine not worth taking. Thus, the bigger sample is actively worse, because it is capable of finding rare effects that would scare either regular people into declining or scare the FDA into not approving, and such decisions would almost always be mistakes.  

When tackling the question of schools, if you are going to take the position that children are at sufficient risk from Covid that they can’t be put into rooms together with one adult, one should notice it’s strange to also forbid them from being vaccinated, and it’s heartening to find advocates that at least realizes that much (and of course wants to go directly from forbidden to mandatory, as I am confident we will do for children).

So at least there’s that. Of course, this is in the world where even fully vaccinated children can’t safely be put into the same room without masks.

I’m also happy to see air quality mentioned as a key issue (as a reminder, air quality improvement in schools would be urgently necessary and worthwhile even if it didn’t matter for Covid or the moment to moment experience of breathing the air). The core of the argument, later in the thread, is the risk of Long Covid. You have to raise the specter of Long Covid when talking about children, since the risk of anything else is clearly not worth worrying about even without vaccinations, and here the proposal is to worry about it even after vaccinations. 

As usual, the procedure in this thread was to gather together every possible symptom, of any severity, and any duration longer than a few weeks, that happens after someone has Covid (and that may or may not actually have anything to do with Covid) and count them all as Long Covid together, with no attempt to quantity what it means for someone if they get it. Also without any practical plan for how long proposals to avoid it might last or under what conditions they would be willing to stop using them. 

The New York Times was also its usual self and did its best scare piece on Long Covid in children, but it’s only one of a chorus of such claims. That doesn’t mean Long Covid isn’t real, it’s clearly a thing and the primary risk factor for younger people, but it must be kept in perspective. Here’s a thread pushing back, and the related post from Gaffney:

It could be worse, and usually is. The standard line is that Delta is even more dangerous for children because they have a higher percentage of infections, hospitalizations and deaths than they did previously. Which is absolutely true, and can be explained by the fact that they’re mostly not vaccinated.

From post in question, ya don’t say, might I suggest something we might do about that?

This CNN post is similar and typical. It cites increases in cases in children similar to increases in cases overall, then has to explain why anyone should care. Its first reason is that this is critical to keeping them in school, because if we don’t protect kids from Covid then we’ll have to take them out to protect kids from Covid, and then we’ll be forced to detain them at home instead. The second justification is that the kids might spawn new variants, which is technically true but seriously, come on. Then they hold up the specter of MIS-C, with a total of 4,196 cases, which is at least a specific issue rather than generic Long Covid, but again, math. 

Meanwhile, if you took Remote Schooling, treated it and its side effects as a pandemic, and ask what would happen if it was spreading across the country, I think the answer is full Australia-style stay-in-your-house lockdowns as needed. 

I do get that this is now a strange position to be in, and if you’re deciding on school policies independent of the FDA, you’re in the same position you’d be in if this was a wait on manufacturing and distribution instead of regulatory approval – the kids will be vaccinated, but can’t be now, and the fact that other people could change that if they wanted to does not give you that option. And if there are enough cases you will be forced to go to remote learning no matter what you know about its consequences. 

Tyler Cowen asks how many children are killed by school anyway. 

As for how many are killed by Covid-19.

That could maybe change by an order of magnitude before it’s all over, but that’s an upper bound. 

From an excellent post about the question of what to do for kids who it is illegal to vaccinate, here’s a chart of what happens if they do get Covid.

That death risk is consistent with what I had, and for those under 12 it will be lower still. I’m skeptical of the 1 in 50 line, but the word ‘any’ does strong work in such situations, so… maybe? 

Long Covid is real and important, but so is Long School. Most people I know are permanently traumatized by it, many people have nightmares about it, and so on. Suicide attempts drop dramatically for young children when school is out, in a way that suicides of older people don’t. It doesn’t end when you’re allowed to leave. There’s also Long Vehicle Accidents, Long Suicide Attempts, Long Cancer, Long Drowning and so on, as one might expect. 

We do the same thing with school shootings, where we force kids to take place in ‘live shooter drills’ and scare kids so much that they expect a school shooting to happen around them, whereas such things are exceedingly rare, schools don’t have more of them than the rest of life and the damage done by scaring everyone is orders of magnitude the bigger concern (and is probably doing far more to give kids ideas and causing such events than it is to prevent anything, if I had to guess). 

One could also compare the moral panics over ‘stranger danger’ to the fact that most kidnappings of children are by family members, and most abuse is by people they know well.

The problem, in all these cases, is that some threats are put in a special category where any bad events are unacceptable, whereas other things are part of life, such as getting lots of people into close proximity in this thing we call a ‘city.’ 

Kids are not in a good place right now. We’ve disrupted their lives and kept them socially isolated for over a year. Making their new school year largely about Covid, and forcing them behind masks, will make things that much worse. 

I am thrilled to see that lots of parents intend to home school their children for the coming school year. In-person school is bad enough, but choosing remote learning over home school is a true tragedy, either of misunderstanding the situation and/or a lack of sufficient resources to deal with the necessary obstructions, paperwork and actual educational efforts required. My heart goes out to anyone who knows and simply can’t do it. 

At some point we will need to learn to live with Covid, or make an extraordinary effort to somehow live without it by vaccinating everyone and then moving on. Or we could doom ourselves to a young adult dystopia in yet another way, the same way kids are no longer allowed to play outside and are told not to talk to strangers and pretend periodically to hide from gunmen roaming the building, plus whatever you consider a baseline ‘school’. That’s also an option. 

Mask and Testing Mandates

There’s an old improv game called standing, sitting, bending. May I present a new one: Eating, drinking, dancing:

Ministry of Truth

Even when the actual implementation in a given example seems fine, it’s important to focus on the reasoning, for this tells you what the ministry is looking to do next. 

Facebook’s War on Supposed Misinformation continues, and has produced the following ‘fact check’ of ‘misinformation.’ I wouldn’t be focusing on Something Wrong On The Internet, except this type of ‘fact checking’ from this exact source is being used to censor Facebook and Instagram. Although in this case, the post wasn’t censored, merely given a warning label (and one assumes also a massive Streisand Effect), that’s not always the case. 

The Covid survival rate is clearly over 99%, by the CDC’s own estimates. The CFR is 1.7% and the CDC conservatively estimates half of infections have been missed – I’m guessing there are at least twice as many as that, perhaps more. 

Saying this is known to not have a CFR under 1%, as your headline that is then quoted around, or that survivability is known to not be over 99%, is blatant lying and scaremongering. 

What’s this all about?

It’s about the new definition of misinformation, which as far as I can tell is information used to lead to a conclusion we don’t like. 

The fact check admits that the data comes from CDC modeling estimates, and then uses those best guesses as best guesses. But because you can’t prove those are the correct numbers, and the conclusion is one they don’t like, the ‘fact checker’ thus concludes the claim is ‘false.’

What’s funny is that the exact claim being evaluated I too think is actually false. I’ll get to that later. But that’s not a fact that’s correlated with any of their reasoning.

There’s so many different things dangerously wrong here. 

1. Unproven or unknown does not equal false, and by a sufficiently strong standard we “know” almost nothing, paging various philosophers. And by this standard, since I can choose not to offer proof, I can get them to say (almost?) anything is false, and thus by flipping the sign say almost anything is true, provided it serves their purposes. Neat trick.
2. Using the CDC’s numbers from their modeling is an excellent source of reasonable approximations. It comes (as per the fact check post!) directly from the CDC and is being used to predict things, so it’s a forward-looking estimation. When I disagree with such assessments because I think I know better than the CDC, which I do here, that’s because I’m the arrogant one who thinks he’s better than the CDC, not the other way around.
3. The general survival rate for Covid is clearly over 99% as discussed above via the CDC’s own estimate of the true case count. That doesn’t then automatically extend to ‘most age groups’ which is why I end up thinking the claim as categorized by the fact checker is false, but that’s not how the original post categorized anything, so the actual disagreement is over exact numbers for particular age groups. 
4. Evidence that isn’t of the correct form or from the correct source (even within exactly the correct overall source, the CDC) is being selectively dismissed when it doesn’t suit them, which makes it easy to find a ‘lack of sufficient evidence’ to find something to be false.
5. They do not cite what numbers they do believe, or any evidence for or against any numbers whatsoever except a general FUD about believing any numbers at all. 
6. Up front they are clear why they are doing this – it’s because the claim is being made in order to minimize the importance of vaccination. The fact is a soldier for the wrong side, ergo false.
7. Again, in their conclusion, they’re judging their characterization of the central claim as false, rather than disagreeing with any particular claim or giving an alternative model. 

Here’s the argument that if you have the best data available, you should ignore it, because there’s some factors it didn’t account for, and thus you should throw out all numbers and have no idea whatsoever. Which is a fully general argument against ever knowing anything at all: 

This is not how knowledge works, unless you are banning forbidden knowledge due to its Unfortunate Implications. Yes, of course you should use data on how many people have survived Covid-19 to predict your own chances of surviving infection. What the hell else would you use as a starting point? And the whole idea here is to then condition that on age, which is by far the biggest risk factor, and then condition on vaccination status (where I think their 94% number is somewhat low, but it’s well within the range of Numbers Used By Official Sources To Scare People The Proper Way Depending on Context, and also not a crazy estimate, I just think defense against death is somewhat higher.)

Whereas the post is indeed taking overall estimated forward-looking numbers, then adjusting them by age, and also listing vaccine effectiveness. If one wanted completeness yes there are other factors but they’re far less important – see my graph below for my ranking of the next two in line (diabetes and obesity), and how much less important they are than age. 

The advantage of telling people to throw up their hands is that you can simultaneously say vaccines are super effective and important (without specifying numbers) when telling people to get vaccinated, then turn around and tell them to be terrified of Covid afterwards anyway, even if they’re young.

Yeah, people are skeptical of authorities these days for some reason, can’t imagine why. 

Next, I’m going to actually fact check the chart, since I think my estimates are better than the CDC’s estimates. Are the ratios by age here correct? 

I actually think no, they’re too aggressive. Here was the result of my comorbidity work, which was pre-Delta and pre-vaccinations, and have younger people at more risk than this by an order of magnitude or so. 

To compare apples to apples you should look at the All Pop column on the right, and focus on ratios between groups, and also remember that there aren’t many people over 90 when combining the top group together. With that adjustment, my conclusion is that the post above is underestimating risk to the young by about a factor of 10. 

Now, let’s look at the actual post, and, huh, ok, I see it now…

(Note: That font and color scheme in the graphic is very recognizable as coming from Fox News, the outgroup’s relatively mainstream news source.)

Yeah, that’s… a very reasonable warning. This is indeed missing context and could mislead people, and the warning isn’t claiming it’s false, merely that it’s missing context. In particular, this is framed carefully to imply that the vaccine would replace the existing immune system rather than supplement it, and thus the vaccine would increase risk rather than decrease relative risk. 

So in the context of the post, the label is at least understandable. It sets a bad precedent even if the written justifications for it had been relatively good, so I’d rather not do it, but certainly one can understand it, especially when combined with the numbers here being so aggressive, although the extra 9 likely doesn’t change the message here much. 

As opposed to the reasoning in the justification post, which is… different.

I’ll also note that clicking on the warning doesn’t go anywhere when I tried it, which seems like a missed opportunity if one did want to communicate context.

More generally, the official reasoning remains the supremely broad claim that any disagreement with health authorities is not allowed. As we are periodically reminded, this is despite the health authorities changing their opinions over time as (A) the facts change, (B) we get better evidence and (C) they update to take into account new information and incentives and priorities. Usually their truth tracking improves over time on a given issue (and stays the same on average because they add new issues), but not always. Also you can’t contradict multiple health authorities, including the WHO who still refuse to admit Covid is airborne, and those different authorities frequently contradict each other. In the case above, the CDC’s numbers can’t be used in a way that wasn’t intended. By the standards that are being used to censor a United States senator who is raising a perfectly valid scientific hypothesis in the link earlier in this paragraph, you could censor not only at least most of my Covid posts, but almost anything remotely useful anyone might say, whether they were trying to provide useful information or trying to figure things out. If I wasn’t Against Facebook I’d probably be banned from it by now, and I’m curious to what extent they mess with those who post these weekly updates there.   

Provincetown Follow-Up

Nate Silver offers some thoughts on exactly how hard it is to compare vaccinated to unvaccinated people, even in relatively ideal conditions for such comparisons.

Provincetown was the opposite of an ideal situation.

It seems very clear at this point that the Provincetown study did not mean anything like what the CDC was representing it to mean. This was a situation filled with activities that carry extreme Covid risks, among a unique and often immunocompromised population. The outlier results at most describe what happens in circumstances like that, and also fail to control for the population baselines appropriate to that situation.Despite that, there were zero deaths, only seven hospitalizations, and most vaccinated participants were not infected. The vaccines did their job. In hindsight, while I had a strong prior that the study wasn’t going to mean what the CDC was claiming, that prior wasn’t strong enough, and I gave the whole situation too much respect.

Thus, the focus shifts from the study and its claims to the actions of the CDC and media, and updating on what they did in response to this information. How much of this failure to update was due to people being afraid to point out the nature of the gathering, given today’s political climate, until enough others had done so first? Or was the narrative what everyone wanted to go with anyway, so it was too good to check regardless? Or was it that the CDC is no longer capable of reading scientific studies and analyzing data about the physical world in a reasonable way? Perhaps this was all kayfabe at the CDC and they knew exactly what they were doing, and were simply lying, and the media went along with it to show their deference to power and get clicks? Or was it something else?

Could the CDC actually be this bad at communicating about risk? 

There’s being bad at talking about probability and risk, and then there’s treating everything as an absolute. 

No matter what the cause, it is yet another reminder that the data all fits together into one physical world that runs with one set of physical laws and biological properties. Something that doesn’t fit and contradicts the data and results observed elsewhere must be treated with extreme skepticism, and any model must explain those other results and data points. 

Delta Variant

Vaccines work on it, J&J massive study edition, as in n=500k participants:

One J&J shot remains very good protection against death and good protection against infection and hospitalization, but not as good as two mRNA shots or one J&J plus one booster, and the logic of getting the second shot of mRNA is the same as the logic for an mRNA booster after getting J&J. Of course, if you request an mRNA booster after having had J&J, it might be tricky to get it, because the FDA and others are tying themselves up in knots denying the obvious is sufficiently obvious.

This isn’t the exact thing we most want, which is how effective the mRNA vaccines are against Delta, but it suggests only a small decline in effectiveness is likely. 

Also, about the way the CDC goes about the business of gathering its data:

I would like to hold the CDC to higher standards here than I can afford to have writing these posts, but that option is not available at this time. 

In Other News

What’s the difference between an EUA and full approval? Hundreds of thousands of pages of paperwork and a bunch of site inspections, among other things.  

About a week and a half ago, Scott Alexander wrote a righteous post everyone should read on how horrible the FDA is and in particular how they are way, way too slow to approve drugs and also getting them approved costs eleventy billion dollars each (realistically something like 100 million). One thing that caught everyone’s attention was the infant fish oil story, where the FDA for years let children get sick and die rather than let fish oil get added to an infant formula, as he detailed in his first follow-up, then he wrote a second follow up when a critic pointed out that those involved in that story praised everyone at the FDA. Scott points out that yes, the individual people at the FDA did their jobs in this situation, but that doesn’t make it better, the system was working as designed and the design sickened and killed a bunch of babies and that’s the thing to be focused on. If anything that’s worse, if it was the people letting us down we could go fix that. He’s being careful not to outright say FDA Delenda Est, but as his alternative he’s holding them to the impossibly high standard of ‘better than the man on the street.’  

I do think this concise statement of the argument goes slightly too far, but only slightly. 

It’s no wonder that no one wants to be FDA commissioner:

Well, no one who counts, anyway, where one who counts would be someone without trouble getting approved. I certainly get it, in the sense that when I think of my life if I was made Commissioner of the FDA, in terms of my lived experience, oh my would it be infinitely worse. I’d happily do it anyway, because someone has to and if I don’t do it then someone else will, and also it would open doors to do additional important things after, but I have to assume it would be a nightmare. 

If you’re someone who ‘you won’t have trouble getting approved’ then presumably you’re looking to run the FDA the way the FDA is traditionally run, which means someone has to but also someone else would if you didn’t, and if you’re doing it honestly it’s a giant paycut, so why take on all that trouble? 

Did the NIH do better (MR)? Here’s the WSJ, here’s the full report, here’s part of MR’s summary, note the top line especially:

So, not that great, only a handful of billions while missing entirely many of the things we most need studied. The grant process isn’t working. In other distributional news, happy to see Aging get this attention, although it’s telling that it’s right behind Rare Diseases, even if they’re not in cute puppies.

Again, could be worse, you could be the WHO analysts and still, this week, be telling people Covid isn’t airborne. Delenda est indeed.

Obama had a birthday party, outdoors, with vaccinations and Covid tests required, but didn’t require masks, so naturally a bunch of Justifications are required for this living of life as if physical reality was exactly the way it is. 

What happened was that Obama has a brain and occasionally thinks about physical reality, but to explain this in those terms would destroy the rest of the narrative, so what they say ends up sounding not great. 

But seriously, how’s it going out there?

You can, of course, get an anecdote to say almost anything, for example “Idaho Covid ICU patients are already at an all time high” when the stats say that clearly isn’t true. Treat local reports with generous helpings of salt before generalizing. 

When cancer survivor Anthony Rizzo was traded from the Cubs to the Yankees, life was proven unfair, and also there were many who noted that he was unvaccinated, which turned out to be unrelated to his cancer – he simply declined the vaccine. A few weeks later, he’s on the Covid injured list. 

Israeli data seems to show that previous infection is not only highly effective at preventing reinfection, in their samples it looks even more effective than vaccination. This is the opposite of what is found in other reports, but definitely worth keeping an eye on. 

Perspective on Louisiana hospitals being full. Looks like this is essentially by design and didn’t require that many Covid patients for it to happen.

You’re about to spend trillions on ‘infrastructure’ that is mostly transfer payments to people you like, to be paid for by people you don’t like, in the wake of a huge pandemic, and aside from potentially banning large areas of software development by requiring theoretically impossible tax reporting, how are we doing on spending on actual pandemic preparedness? Oh…

Not Covid

I’m putting this here because from time to time, it will be needed.

I finally tried Storybook Brawl this week, and it is excellent. Highly recommended to anyone who likes playing games of any kind, give it a shot, it’s already Tier 1 even though it’s in Early Access. I definitely have thoughts on it, but we’ll see when I get around to writing those down. In the meantime, great fun.

I’ve also been greatly enjoying Across the Obelisk. This is a unique roguelike deckbuilder, in that it is trying and largely succeeding at being like a lightweight D&D rather than being a lightweight Magic: The Gathering. You have a lot more control than in most such games over what your deck looks like, so it’s up to you to decide how to keep it fresh after a while, but there’s a bunch of viable options and this is great stuff, again even though it’s still in Early Access. I’d put it at Tier 2 in its current state. If you’re up for what’s being described, check it out.

I finally saw a movie, Black Widow, in a theater for the first time since the pandemic began. It would have felt ritually impure to have my first movie back be anything else. By waiting for several weeks, I got a mostly empty theater, so social distancing was excellent. As much as I was looking forward to it, I didn’t realize how much I missed the movies until I finally got to go. Excellent experience, can’t wait to go again, don’t think I have time to see Free Guy or Suicide Squad tonight but I’m not ruling it out. My review of the movie Black Widow is: Exactly meets expectations. 

In Free Britney news, the system is even worse than you think. There was serious risk of having Britney Spears committed involuntarily, because of the supposed mental health strain of trying to free herself in court, and the strain of having her father as her legal enslaver against every scream she can muster:

A judge then refused to expedite the hearing, of course. One cannot rush such proceedings. The FDA would approve.

I’d also like to point out another parallel of horribly inefficient action that got highlighted this week, which is the War on Bags:

I am reminded of when my teammate Patrick Chapin went to get croissants from a gas station in Belgium (which were really good croissants, Europe has its advantages) and some other food, and was given an obviously horribly inadequate number of terribly flimsy bags. When he offered to pay unreasonably large amounts for additional bags, he was chastised for how little he cared about the planet. Then was forced to spend an hour getting back as the situation fell apart on him multiple times. 

You know what costs vastly more energy and carbon to produce than paper bags? Food. Even a tiny risk of food being wasted is much worse than using extra bags. Yet what happens when bags break? Food containers break open, food is dropped on the ground and made dirty, and both lead to food being thrown out. That’s in addition to the hours upon hours of lost time. 

And finally, in case you missed it, too good not to share and also insightful: