(Writer’s note: Due to the utter trainwreck that is WordPress’ new editor, this week’s post was written in Google Docs and then cut and pasted into WordPress. This is an excellent example of the principle that any given thing is likely getting worse even as things in general improve. As an experiment, I’m opening the draft document to public comments – I think commenting on the side as the post goes is actually a better method, and I encourage you to read it and comment on it there.)

The Covid news is not as excellent as it looks at first, but it is still good.

The positive test percentages, and the number of people testing positive, are dropping fast. The number of people hospitalized is dropping as well. The number of reported deaths mostly continues on its plateau, but that almost certainly reflects a dropping number of actual deaths by this point.

Those positive test percentages look less impressive when broken out by region, which we do this week for the first time, but improvement is still improvement.

There’s even a rapid test freshly approved by the FDA, thanks to the NBA. More on that in its section, but don’t say sports never did anything for you.

There’s also a lot of noise out there. There’s some new official lying for our own good I need to complain about.

It’s still important to celebrate the good times, when the times are good. Even if those times aren’t as good as we would like or hope.

Thanks to some C# code, I am now downloading and parsing the .csv files from CovidTrackingProject, which makes the data far easier to analyze.

Let’s run the newly expanded numbers.

Positive Test Counts

Date	WEST	MIDWEST	SOUTH	NORTHEAST
June 4-June 10	35487	24674	55731	22693
June 11-June 17	41976	22510	75787	17891
June 18-June 24	66292	26792	107221	15446
June 25-July 1	85761	34974	163472	16303
July 2-July 8	103879	40139	202863	18226
July 9-July 15	108395	53229	250072	20276
July 16-July 22	117506	57797	265221	20917
July 23-July 29	110219	67903	240667	26008
July 30-Aug 5	91002	64462	212945	23784
Aug 6-Aug 12	93042	61931	188486	21569
Aug 13-Aug 19	80887	63384	156998	20857

Here we see a hopeful continuation of previous trends. The South and Northeast continue to rapidly improve. The West has resumed improvement after last week’s setback. The Midwest continues to be stuck in place.

Death Counts

Date	WEST	MIDWEST	SOUTH	NORTHEAST
June 4-June 10	743	1297	1230	1936
June 11-June 17	778	1040	1207	1495
June 18-June 24	831	859	1204	1061
June 25-July 1	858	658	1285	818
July 2-July 8	894	559	1503	761
July 9-July 15	1380	539	2278	650
July 16-July 22	1469	674	3106	524
July 23-July 29	1707	700	4443	568
July 30-Aug 5	1831	719	4379	365
Aug 6-Aug 12	1738	663	4554	453
Aug 13-Aug 19	1576	850	4264	422

This shows a large increase in deaths in the Midwest, and a decline in the West. I am guessing that some of last week’s deaths in the Midwest got delayed, and things are much closer there to being steady. The South saw noticeable improvement, which makes sense, but it’s too early to know for sure. The Northeast is back about where it makes sense for the Northeast to be.

Test Counts (Old Chart)

Here’s the chart I make from the spreadsheet each week:

Date	USA tests	Positive %	NY tests	Positive %	Cumulative Positives
June 11-June 17	3,446,858	4.6%	442,951	1.1%	0.66%
June 18-June 24	3,638,024	6.0%	440,833	1.0%	0.72%
June 25-July 1	4,331,352	7.1%	419,696	1.2%	0.82%
July 2-July 8	4,461,980	8.2%	429,804	1.1%	0.93%
July 9-July 15	5,196,179	8.4%	447,073	1.1%	1.06%
July 16-July 22	5,481,861	8.5%	450,115	1.1%	1.20%
July 17-July 29	5,757,290	7.8%	448,182	1.1%	1.34%
July 30-Aug 5	5,079,828	7.9%	479,613	1.0%	1.46%
Aug 6-Aug 12	5,014,573	7.5%	502,046	0.9%	1.58%
Aug 13-Aug 19	5,229,319	6.3%	543,922	0.8%	1.68%

A small bounce back in capacity is very good to see given the alternative. The drop in positive rate is large and highly welcome. It’s the best looking statistic we’ve seen in months. But see the next section.

New York’s positive test percentages continue to drop. I worry that this effect is mostly ‘we have more testing capacity and marginal tests are all coming back negative’ rather than a drop in infections. Of course, in important ways that is excellent. It means testing truly is adequate, and the true infection rate might not be much higher than the positive test rate.

Positive Test Percentages and Hospitalizations By Region

Current Hospitalizations (average daily value over period):

Date	Northeast	Midwest	South	West
3/26 to 4/1	10977	116	1524	1625
4/2 to 4/8	24021	1716	3937	3835
4/9 to 4/15	33757	9068	6160	6466
4/16 to 4/22	34173	9753	6639	7408
4/23 to 4/29	31065	10467	7383	7609
4/30 to 5/6	26378	11058	8579	7174
5/7 to 5/13	21709	10505	8376	7250
5/14 to 5/20	17801	9462	7975	7178
5/21 to 5/27	14333	8718	8015	7073
5/28 to 6/3	11859	7826	8053	6950
6/4 to 6/10	9077	6740	8160	7232
6/11 to 6/17	6688	5661	8642	7286
6/18 to 6/24	5308	5032	10203	8266
6/25 to 7/1	4446	4920	13462	10324
7/2 to 7/8	3949	5075	17846	12427
7/9 to 7/15	3899	5630	28559	13892
7/16 to 7/22	3904	6127	33428	14523
7/23 to 7/29	3875	6594	33706	14255
7/30 to 8/5	3657	6672	30839	12863
8/6 to 8/13	3412	6703	27897	11379
8/13 to 8/20	3141	6639	24275	10214

This is the first time I’ve presented the hospitalization data, so I figured it made sense to go back a long way.

I’ve been skeptical of using the hospitalization data due to all the ways that data can be warped, but at this point it seems like a long view over time offers enough insight to be worthwhile. It is another perspective, and it suggests strongly that the corner has been turned.

Positive Test Percentages (Note: Washington has been given 100k negative tests per week because it stopped reporting negative tests over two weeks ago)

For individual states, see the chart in the Positive Test Percentages tab on the spreadsheet.

Percentages	Northeast	Midwest	South	West
3/26 to 4/1	32.17%	16.70%	11.55%	11.84%
4/2 to 4/8	36.20%	16.16%	12.97%	8.11%
4/9 to 4/15	35.03%	16.49%	11.27%	9.45%
4/16 to 4/22	30.92%	17.28%	10.66%	5.55%
4/23 to 4/29	22.07%	13.78%	7.41%	7.56%
4/30 to 5/6	17.88%	12.44%	6.80%	6.29%
5/7 to 5/13	9.92%	9.40%	5.88%	5.01%
5/14 to 5/20	8.49%	7.19%	4.71%	4.33%
5/21 to 5/27	5.41%	6.72%	5.77%	4.20%
5/28 to 6/3	4.24%	4.61%	6.18%	5.04%
6/4 to 6/10	2.65%	3.96%	6.20%	5.08%
6/11 to 6/17	2.04%	3.32%	6.97%	5.35%
6/18 to 6/24	1.90%	3.82%	9.49%	7.24%
6/25 to 7/1	1.78%	4.08%	11.61%	8.32%
7/2 to 7/8	2.20%	4.71%	13.15%	8.59%
7/9 to 7/15	2.11%	5.12%	13.77%	8.53%
7/16 to 7/22	2.49%	5.13%	13.29%	8.56%
7/23 to 7/29	2.54%	5.51%	12.32%	7.99%
7/30 to 8/5	2.58%	7.26%	12.35%	6.68%
8/6 to 8/13	2.30%	5.67%	14.67%	6.98%
8/13 to 8/20	2.06%	5.62%	9.41%	6.47%

This week’s dramatic drop was driven by the South, which went from an outlier high number that likely reflected data errors, to an outlier low number that probably also reflects data errors. My guess is that some negative tests got time-shifted from last week into this week. If you integrate the last two weeks, you get a positive rate for the South of 11.7%, which is a small but noticeable improvement over what came before that. We also see the Northeast and West reach their lowest levels in a while.

Note that these numbers came from using C# code to parse the CovidTrackingProject’s .csv files, so there is a larger than normal chance of meaningful errors creeping in. We manually fixed Washington. If you find an important error, please do let me know.

Putting all these perspectives together yields a consistent picture.

Putting the Numbers Together

Things are slowly improving in the Northeast and West, and improving somewhat more rapidly in the South, but not rapidly enough that we can be confident putting kids back in school or colder weather won’t reverse those trends. In the Midwest, things are mostly holding steady.

My presumption is that the Midwest is not improving because it is not getting sufficient benefits from herd immunity.

Going state by state, we see dramatic improvements continue in Arizona, where the positive rate was over 20% from 6/18 to 7/30 and is now down under 10% and falling fast. Hospitalizations are also down by more than half. Alabama and South Carolina and Arkansas are also recovering fast, so fast that the drops in positive test rates look suspiciously like data errors. Florida is recovering far slower, still with a 15% positive rate, but down substantially from peak and falling. Texas has too many data errors to tell for sure.

We see substantial improvement in California, although that could largely be them fixing data errors.

The only state where things are clearly rapidly getting worse is Hawaii. From mid-April until mid-July they held positive rates under 2% but now they’re headed up steadily and are over 10%. Hawaii has all the signs of an uncontrolled wave that’s going to look like what recently happened in Arizona, Texas and Florida.

Looking forward, we should expect the death rates to slowly drop, and the hospitalization rates to also slowly drop. Positive test numbers are likely to rebound upward a little bit, because this week’s rate was likely artificially low, but the real positive rate is likely to continue to drop.

Alas, none of this progress looks to be fast. Until we get technological innovation, that seems unlikely to change for the better. Which means that for all practical purposes, you’ll be stuck in this same limbo for many months.

The possible exceptions are the heart of the Northeast region plus Arizona.

Arizona may have overkilled the herd immunity threshold sufficiently that it can squash things within a few months. It would be unsurprising if that’s the safest place in the country by the end of the year, and the place most confident that it’s going to stay that way.

The core Northeast, meaning New York, New Jersey and New England but not Pennsylvania or Delaware, has strong testing in place, strong existing herd immunity levels and low current infection levels. New York City is, as Nate Silver noted this morning, arguably the safest place in the country.

So while things will likely remain mostly static, in those areas, that’s good. Static works. If I knew things would remain static, I’d be looking for a new place in Manhattan to return to. Instead, I’m waiting a few more months to avoid the uncertainty of the winter and school, as well as the election, before taking that step.

C.D.C. Once Again Reapplies to Delenda Est Club

I am going to reject their application once again. They are almost certainly under extreme pressure from the White House that has crippled their ability to respond, and have mostly done (as far as I can tell) excellent work in the past. On this particular issue, they’re saying something rather than saying nothing, and I don’t want to make the perfect the enemy even of the lousy, let alone the good. They have my sympathies. I want to cut them as much slack as possible.

But no more than that. Please don’t push it.

This week’s application is in the form of an advisory on those who have had Covid-19 and recovered.

In that advisory, they say that for three months, such people do not need to quarantine, but they do need to continue mask wearing and other social distancing measures.

This, of course, ended up being used primarily as another way for the Very Serious People to warn that immunity is going to fade Real Soon Now and can’t be relied upon, so we need to keep sacrificing to the Gods and punishing anyone who doesn’t, lest we invite the Gods’ wrath.

If you say that someone is safe for three months, but not for four months, you are conspicuously saying that they are not safe in month four. So that became the headline.

As I somehow have to keep saying: Four months from March would be July, so we know that immunity not lasting four months is utter nonsense. At a bare minimum it almost always lasts five.

Then there’s the matter of still requiring such people to continue socially distancing. Quarantining when called upon is far more efficient and important than socially distancing. If you don’t need to quarantine, you don’t have to socially distance.

Unless social distancing is about sacrificing to the Gods, rather than about not spreading the virus. And the CDC is saying you have to put your life on hold because otherwise people might see you not putting your life on hold.

If that’s what you mean, come out and say it, you lying cowards. See if people agree with you.

If that’s not what you mean, then I am here to inform you this is not how this works, nor is it how any of this works.

This blog’s official position (not medical advice!) is that if you are sure you have previously had Covid-19 and you are sure you have recovered, then you can presume you are effectively indefinitely immune. You can act as if you can neither be infected nor infect others, until such time as we get confirmed media reports of large numbers of reinfections. When and if that happens, you can do the math on timing and figure out how long you have. Until then, your actions are almost purely symbolic.

Which matters! You’re still setting an example. Consider setting a good one. But do it on purpose. Eyes open.

Tyler Cohen echoes this with his reminder about T-Cell immunity, which he calls the dog that didn’t bark. As Tyler says, such responses are common and well-known to those invested in immunology and public health. Yet we effectively didn’t hear anything about them until August. If there is potentially good news, the Very Serious People see it as threatening to undermine people’s willingness to sacrifice, and thus the news is suppressed. It’s not suppressed by some sinister conspiracy, but rather by making sure that it is in no one’s personal interest to talk about it, and letting nature (and Nature) take its course.

Could You Detect Covid-19 Via Voice Changes?

That’s a preliminary theory coming out of MIT. It sounds like their sample sizes are still super low and putting any value on this is still very premature, but I figured I’d share it anyway because it’s a pretty cool concept.

College Without Football Poses Risks To Student Athletes

University of Oklahoma head football coach Lincoln Riley announced Saturday that nine of his players tested positive for Covid-19. The results came back after the team returned from a one-week break from preseason practices.

Said Riley: “You know when you give players time, there is risk in that. This isn’t the NBA, we don’t have a bubble. We all have to continue to work to do a better job by all accounts. We’re still confident in the plan we have.”

In similar news, UNC Chapel Hill cancelled its in-person classes after finding multiple Covid clusters and running a positive test rate of 13.9%. Which to me seems like an overreaction, and another example of reacting to the inevitable result of a difficult trade-off with apologies for one’s lack of sacrifices and a chorus of cries that “I’ve made a huge mistake.” A common theme here is that if what you are doing is seen as reckless, then a mostly ordinary rate of positive tests will be seen as highly blameworthy, whereas if what you are doing is not seen as reckless, it’s considered fine.

Coach Riley’s comments remind us what is at stake. It’s not safe for students to be on campus not playing football. They’re going to break the rules, and they’re going to get infected. What they need are some semi-professional sports that allow them to follow proper distancing rules and get the frequent testing they need to stay safe. The good news is that a large portion of those who never played sports are probably huge nerds, who both follow rules and are already socially distant to begin with. So together the jocks and nerds can work together and beat this thing. Unity 2020.

Or if that’s not good enough, send everyone home entirely. Because then, when they get infected regardless, at least it won’t legally be in any way your fault.

World’s Most Important Scientific Research Performed by NBA

A collaboration between the NBA and Yale, funded by a grant from Emergent Ventures at the Mercatus Center at George Mason University, has led to the creation of a saliva-based test. That test has now been approved for emergency use by the FDA.

The test can be done in two hours, from saliva, without specialized equipment. Its method is being open sourced, so any lab can perform it. They anticipate it will cost about $10, and it is roughly as accurate as a PCR test.

Over time, this could allow us to test everyone periodically, which would in turn be sufficient to squash the pandemic entirely. We could then return to our old lives even without a vaccine.

The next section will talk about the role of Emergent Ventures here. The role of Yale was as a place that is still capable of doing at least some actual science when conditions are sufficiently good and the funding is available.

The key thing to understand here is that this could not have been done without the NBA and its bubble.

The NBA bubble had three necessary elements.

First, the NBA bubble created a population that could be physically constrained, monitored intensely and tested frequently, via multiple methods, without the need to consult with ‘medical ethicists’ and go through an IRB. It was a natural experiment. This allowed tests to be compared and evaluated in a way they couldn’t be elsewhere.

Second, the NBA bubble created a large direct financial incentive, justifying and paying for most of the expenses involved, and thus allowing the amount of scientific funds required to finish the job to be relatively small.

Third, it created enough of a sense of need and urgency from others that the Very Serious People and medical ethicists, who under normal circumstances never let a good voluntary cooperation towards the common good happen on their watch without going through all the proper channels, didn’t seek to shut the whole thing down in order to show how very serious they are. There was little or no talk about how ‘wasteful’ this was, or how ‘inequality’ was involved, or any other nonsense like that.

This is how we are allowed to figure things out in 2020. We can use a special case like the NBA bubble to create a world-saving test as a side effect, and then maybe be graciously allowed to share it with the rest of the world.

If there is a fully natural experiment going on where all the work happens to be done anyway, then maybe it’s ethical to analyze it and figure something out. This is from ESPN’s article on the test.

“My goal is not to test athletes,” Grubaugh said. “That’s not my target population. My target population is everybody. There were concerns about partnering with the NBA when all these other people need testing. But the simple answer ended up being the NBA was going to do all this testing anyway, so why not partner with them and try to create something for everyone?”

Good thing the NBA was doing all this deeply awful testing anyway, thus allowing science to take place. Ever since we banned that highly unethical thing called ‘scientific experiments’, it’s been rough out there.

World’s Most Important Scientific Research Funded By Blogger’s Personal Fund

I wrote this post two years ago urging my readers to apply for grants from Emergent Ventures. In it, I predicted that Emergent Ventures was likely to have a bigger impact than the two billion dollars that had recently been given away by Jeff Bezos (which, as I noted then, was a premature misdirection of funds from the world’s greatest charity, which as we all know is Amazon.com, but was definitely directed towards a good if inefficient cause).

I predicted this because Tyler Cowen would be distributing the money personally, based on quick turnaround decisions to find those who could use the money to study or do things with potentially high impacts, without a bureaucracy or worries about how things looked weighing him down. Also without the need to prove effort or hours worked or anything else weighing down those who got granted money. The ‘secret sauce’ was to make sure not to ‘keep the crazy off his desk.’

Two years later, the fund has exceeded any reasonable expectation with this one grant alone, playing a key role in a potential dramatic advancement of Covid-19 testing. It is likely that Emergent Ventures will, by the end of 2021, have had a meaningful impact on the world’s overall economy, health and well-being.

If you are in the business of using money to attempt to make the world a better place, or even if you are in the business of pretending to do that in order to feel better about yourself or look better to others, or to help others in doing any of that, please don’t ignore this data point!

For those whose charitable focus is anything other than exponential risk, and who don’t want to take direct action themselves or fund individual projects directly, if your money’s first stop is not ensuring that Emergent Ventures is fully funded, that seems like a mistake.

This is how you give your charity the chance to save the world. Not with a whimper or a committee. With a renegade who plays by his own rules. With a bang.

Take Me Out to the Ballgame

The St. Louis Cardinals, in the wake of their Covid-19 outbreak, had 41 people traveling with the team from St. Louis to Chicago.

They rented 41 cars and had everyone drive themselves there.

If you want to be safe, be safe.

Nothing is impossible if you have heart.

Tactical Vaccination

A paper that came out this week showed a result that seems superficially strange to many, but which struck me as intuitive.

Suppose we have a vaccine. Should we give it to old people first, because they are more vulnerable? Or should we give it to young people first, because they spread the virus more?

Their conclusion was that if we had a highly effective vaccine (>50% effective) we should give it to young people, because it can stop the spread of the virus. Whereas if it’s not as effective, we can’t stop the virus that way, so we should instead vaccinate the elderly so they don’t die.

That seems… obviously directionally right?

With any pandemic, there are essentially ‘success’ worlds and ‘failure’ worlds. Failed worlds rely on herd immunity and let things run their course, regardless of how fast or slow that outcome is reached. Successful worlds contain the virus before that happens. If you have a chance to move from failed to successful, or make yourself successful faster with many less infections, that is far more important than other marginal strategic improvements.

However, if you’re going to fail either way, then protecting the most vulnerable and otherwise mitigating the damage becomes the priority.

Shifts between justification via ‘this is for mitigation’ and ‘this is to actually succeed’ have been a big problem.

If High School Must Exist, Teach Statistics In High School

Here is this week’s ‘what do the school numbers actually look like’ exercise. This time it comes from Mississippi.

Ashton Pittman gives us the numbers on Mississippi schools: https://twitter.com/ashtonpittman/status/1295462052757221378

He meant it as an obvious scandal. Look at all these outbreaks!

Google says: In 2013 Mississippi had 493,650 students enrolled in a total of 1,063 schools in 162 school districts (versus 82 counties).

The first thing to note is that there are 71 school districts with at least one infection, and the total infection count is 444. The average ‘cluster’ is 6 people, if that cluster counts everyone in the district. Which does not seem remotely reasonable. So it doesn’t look like we’re seeing huge outbreaks.

There are 493,650 students. 199 of them have tested positive. That’s 0.04%. Hard to be concerned about that, or think it means kids are catching Covid-19 at school at an alarming rate.

There are 32,311 public school teachers. 245 of them have tested positive. That’s 0.76%, which is far more concerning.

That’s a While ‘dozens’ of schools just had their first day of class (which, of course, means that at those schools the infections we’ve found didn’t happen at school, yet somehow are always used as a further condemnation of the safety of schools!), there are a thousand schools in the state, so let’s say that on average they’ve been open for two weeks.

So let’s presume these statistics come from the past two weeks, and see how it compares to the population.

During that period, the state had 11,008 positive tests out of a population of 2.976 million, which is 0.37%. Our source for estimates suggests that fully 3% of Mississippi caught the virus during that time.

Thus, as is usually the case in such things, the 0.76% rate would be scary if teachers are being tested as often as others in the state, but not scary at all if they are being tested much more aggressively. Given how important it is to detect such cases, both in reality and in terms of perception, I would be very surprised if more testing did not explain this entire difference.

I’m still not optimistic about opening up the schools, but whenever I get data like this, it seems to make me less worried rather than more worried. We’ll see what happens after a few weeks. I’ll keep doing similar calculations in the future when I see clean data to work with.

If you have a good source, especially a good source that actually shows danger, please share in the comments. As a reminder, comments are open on the Google doc as well, as an experiment.

New Biggest Idiot

You may think you know who the biggest idiot prize should go to, but I believe that this week we have found a new champion. Behold, Sweden’s Anders Tegnell:

“It is very dangerous to believe face masks would change the game when it comes to COVID-19,” said Tengell, who is considered the country’s equivalent of Dr. Anthony Fauci from the White House COVID-19 task force.

He noted that countries with widespread mask compliance, such as Belgium and Spain, were still seeing rising virus rates.

“Face masks can be a complement to other things when other things are safely in place,” he said. “But to start with having face masks and then think you can crowd your buses or your shopping malls — that’s definitely a mistake.”

He completely brushed off the prospect of wearing masks last month, saying, “With numbers diminishing very quickly in Sweden, we see no point in wearing a face mask in Sweden, not even on public transport.”

…

Tegnell has argued that evidence about the effectiveness of face mask use was “astonishingly weak.”

Sweden has often been called the control group. There has been great value in seeing how things played out there without a lockdown. It’s still not entirely obvious to me it was the wrong call.

But this is ridiculous.

This is telling us it is dangerous and wrong to tell anyone to wear a seat belt, because they might think they then can’t be hurt in an accident.

No, that’s not quite right.

This is telling us it is dangerous and wrong to tell anyone to wear a shirt, because they might not realize they also have to wear pants, and we wouldn’t want anyone walking around without any pants.

No, that’s not quite right either.

Ah. I got it.

He’s saying that we have no evidence parachutes prevent impact trauma related to gravitational challenge. We shouldn’t tell people to wear parachutes when jumping out of planes, because we don’t have evidence that the parachutes help, and telling them to wear a parachute encourages more people to jump out of planes.

So congratulations to Anders Tegnell, top health official of Sweden, winner and new champion of the Biggest Idiot award for Covid-19.

Purdue!

Mitch Daniels wins this week’s award for actually thinking about the physical world and doing things that might help, via the procedures implemented at Purdue University. Via the excellent PoliMath on Twitter, I learned that they are identifying the most vulnerable students and giving them roommates that have already had Covid-19. Then, after Thanksgiving break, they’re going remote for the rest of the semester because everyone going home to their families is going to create outbreaks, so they need to give people time to ride out Thanksgiving and Christmas.

If we could all think like this, by which I mean actually think, and then act upon it, we could beat this thing.

So bravo, Mitch. Bravo.