Most weeks, the disaster that was the head of the FDA not having any understanding of statistics and not making any attempt to think about the world would have been the headline. Then the CDC decided to revise its guidelines on testing from being for it to largely being against it, under pressure from the White House, and suddenly it’s the B-story.

It seems that every day there is a new thing surfacing to enrage me. The difference is that early in the pandemic, every day something would terrify me. I’m still periodically scared in an existential or civilization-is-collapsing-in-general kind of way, but not in a ‘the economy is about to collapse’ or ‘millions of Americans are about to die’ kind of way. 

I’m not sure whether this is progress.

Either way, a reminder that I’ve started a sports, sports gambling and sports modeling substack to avoid cluttering up this blog, so check it out here if you have yet to do so and that is relevant to your interests.

That’s out of the way. Let’s run the numbers.

Positive Test Counts

June 11-June 1741976225107578717891
June 18-June 24662922679210722115446
June 25-July 1857613497416347216303
July 2-July 81038794013920286318226
July 9-July 151083955322925007220276
July 16-July 221175065779726522120917
July 23-July 291102196790324066726008
July 30-Aug 5910026446221294523784
Aug 6-Aug 12930426193118848621569
Aug 13-Aug 19808876338415699820857
Aug 20-Aug 26675456654013232218707

Apologies for the lack of charts last week – it seems they didn’t copy properly from Google docs to WordPress and I didn’t notice. 

Positive test counts are declining rapidly in the West and South, and slowly in the Northeast. Before, I would have considered that strong evidence that things are going great in those areas. Now, with testing on the decline, it’s not that simple, and we’ll have to dig into the positive test percentages and deaths to know for sure.


June 11-June 17778104012071495
June 18-June 2483185912041061
June 25-July 18586581285818
July 2-July 88945591503761
July 9-July 1513805392278650
July 16-July 2214696743106524
July 23-July 2917077004443568
July 30-Aug 518317194379365
Aug 6-Aug 1217386634554453
Aug 13-Aug 1915768504264422
Aug 20-Aug 2615037453876375

Deaths continue to tell the story that the South and West have turned the corner, the Northeast is still making steady progress, and the Midwest did indeed have deaths transferred from the Aug 6 – Aug 12 week to the Aug 13 – Aug 19 week and is still steady or slowly getting worse as of a few weeks ago. 

Each day, we see the 7-day average deaths fall slightly. If anything, it’s a little too consistent and makes me suspicious, but that is probably paranoia. Probably.

Positive Test Percentages By Region

Positive test percentages are actually up slightly in the South this week, down in the Northeast, and down slightly in the West:

7/16 to 7/222.49%5.13%13.29%8.56%
7/23 to 7/292.54%5.51%12.32%7.99%
7/30 to 8/52.58%7.26%12.35%6.68%
8/6 to 8/132.30%5.67%14.67%6.98%
8/13 to 8/202.06%5.62%9.41%6.47%
8/20 to 8/261.86%5.78%9.93%5.88%

Hospitalizations are slightly down as well.

A decrease in testing should slightly increase positive test percentages, so this isn’t inconsistent with the South’s situation continuing to improve, but it’s also definitely not a good sign. A reminder that you can check the state-by-state data in my spreadsheet under the Positive Test Percentages tab here

For the first time, the Midwest did more tests this week than the West did, and they continue to have similar positive percentages on tests, so it will be interesting to see how long it takes their death differential to close.

Hawaii seems to have turned a corner, or at least it’s no longer seeing unchecked exponential growth. Arizona and Florida continue to make steady progress, with Arizona continuing to go faster. Texas news is not as good, and looks ambiguous due to its dramatic drop in testing. 

South Dakota and Iowa saw big scary jumps in positive test percentages. North Dakota also got noticeably worse. Minnesota continues to be headed in the wrong direction and looks poised for a new wave of protests and unrest.

Test Counts

DateUSA testsPositive %NY testsPositive %Cumulative Positives
June 11-June 173,453,4404.6%442,9511.1%0.66%
June 18-June 243,686,3365.9%440,8331.0%0.72%
June 25-July 14,352,9817.1%419,6961.2%0.82%
July 2-July 84,468,8508.2%429,8041.1%0.93%
July 9-July 155,209,2438.4%447,0731.1%1.06%
July 16-July 225,456,1688.6%450,1151.1%1.20%
July 17-July 295,746,0567.9%448,1821.1%1.34%
July 30-Aug 55,107,7397.8%479,6131.0%1.46%
Aug 6-Aug 125,121,0117.3%502,0460.9%1.58%
Aug 13-Aug 195,293,5366.2%543,9220.8%1.68%
Aug 20-Aug 264,785,0566.0%549,2320.8%1.77%

After two weeks of what looked like stable test counts, we once again headed substantially in the wrong direction. The positive test percentage declining in spite of this is of course good news, but raises the worry of whether testing has shifted to where it is least needed while being shut down where it is needed most. New York’s testing keeps expanding even as its case counts decline. 

The numbers look like one would have expected them to look last week, plus a decline in tests. There is slow improvement in deaths and hospitalizations. The outlook continues to be that things will continue much as they are for quite a while, in terms of the virus itself.

Thus, the news this week lies elsewhere.

Head of FDA Fails Statistics Forever

See this tweet from my friend Andrew Rettek, noting that the head of the FDA is far from alone in failing statistics forever. 

(Or as the new cowards in charge of the page at TVTropes are now calling it, ‘Artistic License: Statistics.’ To me, that’s another sign of an alarming societal shift. It’s not ‘artistic license,’ it’s some combination of lying/fraud and a failure to understand how numbers work on a deep and essential level. There are times when it makes sense to use a little ‘artistic license’ with your physics or biology to tell a good story. But much more frequently, They Just Didn’t Care about what is true and what causes what in the physical world, the same as others who are making actual decisions with actual consequences don’t care either. If you don’t apply for the license, you don’t have one. Period. I call for the immediate and total restoration of failing forever.)

And there isn’t one for statistics or mathematics. You just fail. Period.

The mistake here isn’t one that someone who knows how numbers work would ever make. It’s also not one that people who think about what their statements actually translate to in the real world would actually make. 

It’s a statement one makes when one is doing word manipulations. Where one sees a 35% on a piece of paper, doesn’t notice at all that this is from 11.9% to 8.7%, and decides that means that 35% of all patients will go from dying to not dying rather than 3.2%. 

No, seriously, they doubled down on this pretty hardcore. Check this out from deeper in that thread:

And now the FDA's Twitter account is promoting Hahn's disinfo.

Full quote from Hahn: "Let me just put that in perspective… What that means is if… 100 people who are sick with covid-19, 35 would have been saved because of the administration of plasma."

— Susan Simpson (@TheViewFromLL2) August 23, 2020

He did eventually walk it back and apologize. But think about the mindset required to get things this wrong. To not stop and think ‘wait, I don’t think 35% of patients even die, something’s wrong’ and check again. Until verified, everything such people say has to be presumed to likely be complete nonsense or propaganda, with no relation to the physical world. 

That doesn’t mean the treatment itself is bad.

A Miasma of Convalescent Plasma

So does convalescent plasma work?

Wired reports that 97,000 People Got Convalescent Plasma. Who Knows if It Works? The article contains lots of good information. Clearly epic ball droppings occurred. No one wanted to fund the clinical trials. If you’re wondering, all we had was money, what could we possibly have done to help? That’s something you could have done to help. We had tests ready to go, with no ethical issues, all they needed was funding. They didn’t happen, so now ‘no one knows if it works.’

Of course, that’s complete nonsense. It works.

It’s not the major breakthrough the President claimed it was. The range of possible effectiveness is ‘somewhat’ to ‘actually quite a bit’ but it’s still expensive in several senses to administer and all but certain to be supplanted by superior treatments soon. As that second link explains, convalescent plasma is a classic emergency stopgap strategy that is used until better options can come online. 

Still, it works. If someone is a patient and the plasma is available, until better options come along, this is not medical advice because legal reasons, but there’s no reason I can see to not to take the plasma. As everyone says, it’s a historically safe treatment and seems clearly safe once more. In terms of how effective it is, we know that patients that took it earlier did better than those who took it later, and those who had a stronger dose did better than those who had a weaker dose. 

Which likely means that even the strongest doses were probably not as big as they should have been if we have access to unlimited plasma. More dakka!

People can protest that ‘there are no RCTs’ all they like. That doesn’t make Bayes law go away. This one is very cut and dry. 

The issue is that if one is doing a tradeoff between the logistical costs of treatment versus spending those resources elsewhere, it’s hard to know what value to put down, but a good estimate does not seem that hard. The 35% number seems large in the sense that it is somewhat cherry picked, but it also seems like there’s a lot of room for improvement in the protocol in terms of timing, quantity/quality and procedure, as we learn more. So if we can get our testing done in a timely manner and get the treatment to patients quickly, my guess is this is good for at least that much. Which is a big forking deal.

Of course, a better treatment for those who get hospitalized if anything makes it harder to squash the virus, rather than easier, given the resulting changes in behavior. So this doesn’t put us any closer to normal. Still, great news as far as it goes.

FDA Fails To Approve Tests, People Die and Economy Collapses

A ‘mid-mortem’ came out this week on the FDA’s test approval policy. It mostly reinforces what we already mostly knew – the FDA made an active choice to shut down testing and force all action to go through proper channels, refusing to use its discretion, and then continued to require constantly changing actions and paperwork in order to get approval, and then did not prioritize the handling of said actions and paperwork on their end. Nor did it communicate its changing priorities and rules to those who needed to follow them. 

Many companies report waiting for months for word from the FDA, after filling out what they believed were the necessary requirements and paperwork, only to have their tests stranded in limbo. 

The biggest news there is that this process wasn’t only wholly unnecessary and monumentally destructive, but according to an HHS statement on August 19, much of it was not even within FDA’s jurisdictional rights to impose those restrictions. Many of its actions were without a legal basis.   

Once again: FDA delenda est. 

Musician Plays For Crowd

In ‘someone actually did any study of anything at all, so I’ll take it’ news, Coronavirus: Germany puts on crowded concerts to study risks

To quote the article: The concert study, called Restart-19, was created “to investigate the conditions under which such events can be carried out despite the pandemic,” researchers said.

As far as I can tell, what they studied was where people stood during the concert?

It’s not clear what else they could have done with this setup. It’s considered unethical to let anyone assume Covid-19 risk in order to help anyone else, so they made sure to test everyone for Covid-19 before the concert. At which point, they studied… the complete non-spreading of Covid-19 that took place, since everyone had tested negative? That does not seem super useful.

If you want to ‘investigate the conditions under which such events can be carried out’ and you’ve already decided the answer is ‘by testing everyone before the concert’ then I believe you have your answer. The good news is all we have to do is let people use reasonable and cheap tests, and then all that testing can be funded with less of an extra fee than the Ticketmaster ‘online purchase’ surcharge on tickets. 

Area Couple Does Math, Also Each Other, Probably Didn’t Actually Do Math

A common question I get asked, and also that I ask myself, is what activities are how risky (Google thinks the ‘are how risky’ is a grammar error and wants to correct it to ‘are risky,’ which explains a lot of how we’re in this mess!), and whether a given activity is a sane or responsible thing to be doing given the pandemic.

What about dating and casual sex, or sex in general? (link is to Washington Post)  

Several times there have been fun news items about municipalities trying to ban three ways, or encourage glory holes, or other similar things. I never thought any of those attempts would accomplish anything, but I was always both amused and delighted by such efforts. They represent an attempt to think about the physical world and the consequences of actions, and make decisions on that basis. 

Setting aside that casual sex and dating and the pursuit of happiness in general is considered by the Very Serious People to be ‘inessential’ making it bad to ‘take risk,’ how risky is it, and are there steps that one can take to limit that risk?

The upper bound on risk is that you’re completely exposed to one person. If they have it, you get it. It’s hard to think of something that would more reliably spread Covid-19 than an extended french kiss. That certainly has a leg up on ‘droplet spread’ and is a much scarier prospect than the sexual act itself. 

It seems very reasonable to pick your spots and then say ‘you know what, I don’t want to worry about this and it’s not clear in practice I can reduce that risk much further with precautions that would be useful, so we’ll just accept that risk and do whatever.’ Or perhaps you can make an effort to avoid anything touching anyone’s nose or mouth, or even leave the masks fully on while everything else comes off. To each their own.

But is it reasonable to expose yourself completely to one person in order to get lucky? If everyone is young and has been mostly isolating, and has no symptoms, this seems like it ends up pretty far down on the actual list of downsides to what you’re about to do. If the encounter would have been super exciting before, and again if everyone is otherwise being responsible and isn’t high risk, judicious amounts of this seem to me like an efficient use of one’s (and society’s) risk points. But keep in mind that doing it more with more different people dramatically increases the risk per encounter, so let’s not go too crazy. 

Let’s Go, Mets

My team, The New York Mets, missed five games because one player and one staff member tested positive for Covid-19. Of course, this happened in Miami. Out of an abundance of caution, the team returned to New York and quarantined for five days before resuming play.

It’s good to see baseball acting responsibly. If anything, this was more caution than necessary. 

Since more sports games are now being cancelled for purely symbolic protest reasons than for Covid-19 outbreaks, I’d say the restart of sports has been a smashing success.

The Simulacra of a Mask

While watching the disaster that is the Mets, I saw an ad imploring us all to wear face masks. It showed pictures of people wearing masks that said things like “for my husband” on them. I worry about the message ‘you wear a mask for other people’ given how people are sometimes, but mostly I don’t worry about this. Then the slogan across the screen: Show You Care. Wear a Mask.

So we’ve fully moved on to simulacra level 3

We could go with the Level 1 message that masks prevent infection. We could go with the level 2 message that they don’t, as many have tried at various points for various reasons, or that they are way more effective than they are, which has also been tried. Such messages are false, but they rest on the premise that people will decide whether masks help, and then do what is helpful and not what is unhelpful. 

Instead, we are being given the message that we must signal that we care about others, therefore we must wear a mask. 

The mask is being advertised as being as similar to not playing a basketball game to show you care about police violence. The basketball game does not cause police violence. 

The message is, hey, we know that wearing masks makes today a little bit worse, but it’s worth it, because you need to virtue signal. Bad things are happening, so you need to give up your nice thing, take away nice things and pressure others to give up their nice things, or people will think you do not care. And maybe, just maybe, if enough of us show we care, then someone will care enough to actually forking do something useful. 

This is very important, so I want to say this as clearly as I can: Fuck. That. Shit.

Wearing a mask is vital to preventing Covid-19 infection. Wear a forking mask. 

But only because it works. 

If you don’t think it works, don’t do it. 

You Had One Job

CDC decides third time’s the charm, decides to try harder once more with feeling, insists it needs to join the Delenda Est club.

It’s making a strong case!

You are literally called the Centers for Disease Control. Your one job is to control disease. 

So I want to know what the flying fork is going on over there, given the new guidance they issued this week. First link is the guidance, second and third are news coverage.

An attempt at a “good faith” interpretation of the new testing guidelines, that you ‘do not necessarily need a test’ even if you have been within 6 feet of a known positive for fifteen minutes, is that the CDC is lying. That they are doing the exact same thing with testing that was previously done with masks. 

The plan would be to say: 

  1. Testing doesn’t work for you! 
  2. Save the testing for those who really need it!

Because there aren’t enough tests, or doing too many tests is slowing down our turnaround times.

An attempt at an actual good faith interpretation is that these tests aren’t actually useful. That if you get exposed that getting a test does not have value until you have symptoms. The argument goes that getting a negative test doesn’t mean you won’t become positive later, so you have to quarantine anyway, so what’s the point in bothering. Because that’s how humans work, and that’s how human lives function. And no one else might want to know if they had been exposed in turn.

An attempt at a not-so-good faith interpretation is that this is the result of pressure from the White House, who want to suppress testing so the numbers look superficially better. 

Those were the three I could come up with.

And we have our answer, which I saw after writing the rest of this section. The directive came under pressure from the White House.

Then there’s the other half of the new guidance, which states that travellers no longer need to quarantine on arrival. As one would expect, local officials everywhere seem to disagree, such as in California. I would be floored if New York lifted its travel restrictions.

Basically, the CDC is taking the two things we actually know how to do to stop the spread beyond modifying day to day normal behavior. Our four tactics are, essentially: Wear a mask, socially distance, test and contract trace, and restrict travel to low-infection areas. The CDC is saying that contact tracing and a lot of testing isn’t worth it so don’t bother, and restricting travel isn’t worth it. Except that those are the only tactics that have been shown, anywhere, to actually work. 

We need more testing. Even the money congress allocated for testing, paltry as it was, mostly hasn’t been spent. Test counts are declining rather than increasing over the last few weeks, which is alarming. And while we need more testing, it would make sense to tell people who are low priority that they can’t be tested. But this is something else, and the stand against quarantines on travel makes no sense except maybe as some effort by spiteful people to get back at the blue states, either by tricking them into following it, or by getting to argue they are not ‘following CDC guidelines.’ 

The problem with putting the CDC in the Delenda Est club is that the CDC actually does have its one job to do. We should be doing disease control. The problem with the FDA is that it is doing its job. The problem with the CDC is that it is NOT doing its job. Big difference. So we need to fix the problems, but alas, I can’t let the CDC into the club. Maybe we need to burn it down and start over, but we’d then need to rebuild it, faster, stronger, better. Under no circumstances should we salt the earth.

But for now, we should treat the CDC as having been captured by the White House, and the White House as fully committed to misinformation. Thus, the CDC is now fully committed to misinformation. Going forward, at least until Trump leaves office, treat anything and everything from the CDC with extreme caution and as plausibly motivated as propaganda efforts to re-elect the President.

New Comment
16 comments, sorted by Click to highlight new comments since:

There are other CDCs, e.g. in South Korea, Taiwan, etc.

Could we just switch to following their advice, and ignore the US CDC?

I'm sure there's something stopping us, but I'm having trouble pinpointing what it is.

I'm sure there's something stopping us, but I'm having trouble pinpointing what it is.

Presumably much of the usefulness of the CDC comes from data collection and reacting to that data; I wouldn't expect the Taiwanese CDC to be collecting data on American COVID cases.

See this tweet from my friend Andrew Rettek, noting that the head of the FDA is far from alone in failing statistics forever. 

I look at these links for about 5 mins, and I want to register that my impression is consistent with the FDA person having correct beliefs about the facts on the ground and just being imprecise when trying to explain it in simple terms to the public. If 100 people were going to die of covid, then the plasma thing would have saved 35 of them.

But I'm also having difficulty finding primary sources, so perhaps they did just say the straightforward absolute-risk thing.


I 100% listened and heard him get it very wrong, and lots of other people did too. And if he'd gotten it right, he wouldn't have had to apologize afterwards, so I'm rather confident he got it wrong.

(Which is why, sadly, apologizing is once again clearly the wrong strategic move, even when clearly the right thing to do. Sigh.)

Gotcha, you saw the primary source.

What a sad state of affairs...

The statement is in the first 20 seconds of this video:

If his sentence was 1 word away from right to super wrong I still think it's terribly unprofessional.

In any case, he should know that the sentence is absolutely critical. He should have made three sentences in a row to clarify just in case, because if he missed the word "dying" from your sentence the consequences are enormous.

It’s a statement one makes when one is doing word manipulations. Where one sees a 35% on a piece of paper, doesn’t notice at all that this is from 11.9% to 8.7%, and decides that means that 35% of all patients will go from dying to not dying rather than 3.2%.

Where does the 35% come from?

A reduction from 11.9% to 8.7% means a decrease of 3.2%. As a fraction of the 11.9%, 3.2% is about 27% of the 11.9, not 35%.

11.9% vs 8.7% is early plasma administration (0-3 days from diagnosis) vs late (4+ days).

13.7% is using low antibody count plasma, 8.9% is using high antibody count plasma. I guess this is the 35% reduction.


a decrease of 3.2%

Don't say that. Say a decrease of 3.2 points. Or "percentage points," for clarity. Worst case scenario is that people are confused, which is better than wrong.

I’m still periodically scared in an existential or civilization-is-collapsing-in-general kind of way, but not in a ‘the economy is about to collapse’ or ‘millions of Americans are about to die’ kind of way. 
I’m not sure whether this is progress.

It definitely is progress. If we were in the latter situation, there would be nothing at all to do except hope you personally don't die, whereas in the former there's a chance for things to get better - if we learn the lesson.

By strange coincidence, it's exactly 6 months since I wrote this, and I think it's important to remember just how dire the subjective future seemed at the end of February - that (subjectively, anyway) could have happened, but didn't.

It's surely time to start modeling the endgame, in the form of vaccination scenarios, starting with critical personnel, and extending to the broader population as more doses become available. Not predictions, there's not enough definite information yet, but scenarios: make some assumptions about how vaccines will work (e.g. are boosts needed every few months) and about when they become available, assumptions that are plausible and that are concrete enough to have definite implications - and then let's see what that looks like, let's see what those implications are.

I also don't think that the US has done especially badly, if judged by world standards. I think of its outcome as between Europe's and Mexico's. Very crudely, I think of the populations at risk from Covid as being, first of all, those who are ill or weak due to age, and secondly, younger people with an existing health problem that puts them at risk. Geographically and in terms of total population, the US is comparable to the EU, but would have more people with the kind of conditions (diabetes, hypertension, obesity) that have made the Mexican death rate so high.

I'd also be interested to know what you think of Alex Berenson's oeuvre. He's criticized a lot of policy and journalism regarding the pandemic, and I think some of his criticisms would be right and some wrong, but haven't taken the time to sift through them.

Any remark on the recent reports of reinfections? In previous pieces you were quite critical of such reports, IIRC you estimated a minimum of 4 months of immunity. However, the US reinfection case seems to have been reinfected after less than two months, and with a much more serious reinfection to boot.

Wearing a mask is vital to preventing Covid-19 infection

I’m wearing a mask because I think they are a reasonable intervention and in the hope that me wearing one encourages other people to wear one. (It sounds like they’re more effective at protecting everyone else than protecting the wearer). I‘m not sure which simulacra level this is (1.1, game theoretic axis?)

You're not limited to one simulacrum level per unit of information. What you're describing is just combining level 1 (reasonable intervention) and level 2 (influencing others to wear a mask).


Yes, also in the hopes that it helps encourage others to wear them. As Silver_Swift says it's fine to use that. It can even be Level 1 - you're providing a true Bayesian update to others about whether masks work and what the consequences of wearing one would be!