From the AP:

South Korea’s capital closed down more than 2,100 bars and other nightspots Saturday because of a new cluster of coronavirus infections, Germany scrambled to contain fresh outbreaks at slaughterhouses, and Italian authorities worried that people were getting too friendly at cocktail hour during the country’s first weekend of eased restrictions. The new outbreaks — and the fears of a second wave of contagion — underscored the dangers authorities face as they try to reopen their economies.

Like all of you, I’ve seen a lot of scorekeeping over the past couple of months. South Korea had the best response! Sweden is paying the price for its easygoing ways! We should do whatever Germany is doing!

I’ve done some of this myself, and it’s certainly worthwhile to try to figure out best practices and emulate them if possible. But keep in mind that it’s still early days in a pandemic that will most likely last a year or two. By the end of 2020 it’s possible that we’ll have a whole new idea of which countries have done the best and which haven’t.

For Sweden in particular I’d keep my powder dry for the moment. Remember, their light lockdown rules weren’t put in place because authorities thought they’d be more effective than a tight lockdown. They acknowledged that their death toll might be high at first. But their goal was to find the right sweet spot: a set of rules tight enough to keep the virus under control but loose enough that people could comply with them over the long term. It will be a long time before we know for sure if this worked.

Here’s the coronavirus death toll through May 9. Nothing special to report, though it looks like countries have a hard time keeping lockdowns in place all the way to the bottom of the curve. That’s no surprise, I suppose, but it means that the path from, say, 4 deaths per million to 1 death per million might be a long one. In the United States, we’ve been plateaued at 5.5 for eight days now. Will we decline any further before the numbers explode again in late May?

The raw data from Johns Hopkins is here. The Public Health Agency of Sweden is here.

Quidel Corporation

A couple of weeks ago Deborah Birx delivered a confusing Sunday morning answer to Chuck Todd about COVID-19 testing. She seemed to say that we were two or three weeks away from a big breakthrough in PCR testing, the current gold standard. But she also said that we had to have a breakthrough in antigen testing, and lots of companies were working on that. So . . . maybe she was actually referring to a breakthrough in antigen testing? Maybe. In any case, it looks like we’re getting it:

The Food and Drug Administration has granted emergency-use authorization to Quidel Corp. for the first antigen test for the Covid-19 virus—a step that could escalate the nation’s ability to test for the disease. It is believed to be generally faster, cheaper and easier to manufacture than most current diagnostics.

….San Diego-based Quidel, which specializes in tests for flu, strep and other infectious diseases, already has placed about 36,000 test-analyzer instruments around the U.S. in places like hospital labs, emergency departments and doctors’ offices. “We are ramping up manufacturing to go from 200,000 tests next week (week of May 11) to more than a million a week within several weeks,” said Douglas Bryant, Quidel’s chief executive.

A million tests a week is still a small fraction of what we need, but if we can ramp up to a million, we can ramp up to ten million. And antigen testing is faster and more convenient than PCR testing: it still requires a nasal swab, but results are returned in less than 30 minutes and don’t require any special expertise to interpret.

So: good news, right? Not so fast. The Quidel test, it turns out, is only about 85-90 percent accurate, and that’s mostly because it returns a lot of false negatives. This means that if you test negative—as the vast majority of people do—you need to follow up with a traditional PCR test to be sure. It’s still a useful test, but until it gets better it’s not all that useful.

Here’s something interesting. Maybe. I’ve been hearing for weeks about the bioluminescence event on the Southern California coastline, but I haven’t taken the time to go check it out. Apparently it’s not going away, though, so last night I headed out to Laguna Beach to see if I could take some pictures of it. First off, here’s Laguna Beach all by itself:

This was taken with a 2-second shutter speed, which gives it kind of a cool impressionist effect.

But now for the bioluminescence. I don’t know if this was because the blue glow was weak last night or if this is always the way it is, but it turns out you can’t really see it with the naked eye. It showed up fine in my camera viewfinder, but my eyes saw nothing but whitecaps. The other interesting thing is that the blue glow showed up only briefly and in individual spots. Small waves produced nothing. Only the bigger waves churned up enough algae to produce spots of bioluminescence. For example:

You can see five individual spots here. This was taken with a ½-second shutter speed, which might explain the large number of spots. More typical is this one:

This one has a nice long stretch of bioluminescence plus one more bright spot near the middle. It was taken with a 1-second shutter speed. In the end, that ended up being the best compromise exposure time.

As it turns out, the blue glow comes and goes pretty quickly, so the only way to get a picture is to snap the shutter the instant it appears. Most of the time even that’s not fast enough. You just have to take a picture of every wave and then pick out the ones that have plenty of blue. In the end I took 65 pictures before I was told to move along.

Kevin Drum

I’m not just a former management dweeb, I’m also a general purpose dweeb. So naturally I was interested when this question was posed last night on Twitter: which is safer, driving from San Francisco to San Diego or taking a plane?

The driving part was pretty easy. DOT estimates 150 fatal crashes per billion miles driven. It’s about a thousand miles round trip between San Francisco and San Diego, so that means the chance of dying is about 0.015 percent. Roughly 15 in a hundred thousand.

The flying part is trickier: you need to know both the odds of getting infected with coronavirus and the odds of dying from it. The latter is fairly easy. Assuming a normal, healthy, middle-aged adult, the case fatality rate is about 0.2 percent or so.

The odds of being infected in the first place, however, is hard to get a handle on. I read one paper that outlined an absurdly complicated model of virus transmission in airports and airplanes, but the authors were mostly interested in showing off their new algorithm and how it reduced compute time by 59x on a supercomputer. They never actually put a number to anything. However, WHO says that traveling by plane is pretty safe, and given how sparsely filled planes are these days, I can believe it. So let’s be very liberal and figure the odds at 1 percent for the round trip. I don’t think anyone can complain that I’m lowballing this. With these two numbers in hand, the chance of dying is 0.002 percent. Roughly 2 in a hundred thousand.

So flying wins pretty easily. Aside from the usefulness of knowing this, it’s a good demonstration of how the availability heuristic frames our estimates of danger. Getting killed in a car crash is something we live with all the time, so we tend to discount it. Conversely, COVID-19 is brand new and we’re all scared of it, so we tend to overestimate it. But driving is still pretty dangerous!

What’s dangerous and what isn’t? Erin Bromage, a professor of biology at the University of Massachusetts Dartmouth, has a pretty useful roundup here. The main lesson is: it’s all about Volume x Time. That is, it’s dangerous to be in places with a large volume of viral particles and it’s also dangerous to be in places that might have low volumes but require you to stay a long time. As a former management dweeb, I immediately had to put this into BCG chart form:

Supermarkets, Bromage says, are generally fairly low in viral particles; aren’t very crowded; and usually require only 30-60 minutes of your time. They’re pretty safe. Conversely, a public toilet requires only a few minutes of your time but is shockingly high in viral particles if an infected person has been in it recently.

Indoor workplaces mostly have fairly modest volumes of viral particles, with obvious exceptions like meatpacking plants. However, you’re there eight hours a day. If anyone is infected, there’s a decent chance you will be too. And then there are restaurants, which can have quite high viral loads and often take 2-3 hours of your time.

Anyway, read the whole thing. “The main sources for infection,” Bromage says, “are home, workplace, public transport, social gatherings, and restaurants. This accounts for 90% of all transmission events.” The rest of the piece is pretty helpful too.

Here’s the coronavirus death toll through May 8. Spain, France, and Italy show slight worrisome signs of flattening out now that they’re getting close to zero. But it may be nothing. Sweden’s data is a weirdo mess. There’s no telling what’s going on there. And the United States continues to decline slightly. Enjoy it while you can.

The raw data from Johns Hopkins is here. The Public Health Agency of Sweden is here.

Hey look: a new paper using a large sample size that estimates the hazard ratios for all the various pre-existing conditions that affect COVID-19 survival. Let’s see how I stack up:

Hmmm. Multiply all this together and it looks like my chance of dying if I get infected is 108 percent of the baseline, whatever that is. I don’t think the paper tells us. Still, this sounds bad!

But I suppose it’s not kosher to just multiply all the hazard ratios together. There’s probably either some fancy function for doing it, or else you just go with your highest hazard ratio and all the others contribute only a little. Still, I’ve nibbled around the edges of the internet for numbers that seem reliable (not easy), and my best guess is that if I get infected I have about a 10 percent chance of dying. This is consistent with the hazard ratio chart if the baseline level of death for hospitalized patients is 1 percent.

Anyway, just in case you think I’m taking Donald Trump’s idiocy with this stuff a little personally, you’re right. There are a lot of people for whom this is no game, and I’m one of them.

I guess this post is meant to put down a marker of some kind. Those of us who think it’s disastrous to ease up on social distancing this early are working from a mental model that looks like this:

• We started serious countermeasures in mid-March, and deaths peaked about three weeks later. Then we started slowly declining.
• A few days ago some states started to ease up on countermeasures. This will not have a noticeable effect immediately.
• Starting in late May the death toll will start to increase. At first it will be lost in the noise, but by early June it will be obvious that lifting countermeasures was a mistake.
• At that point we’ll panic and put the countermeasures back in place, but it will be too late. For the next two or three weeks we’ll be climbing up a brand new bell curve, which will peak in mid-June.
• If we stay serious about countermeasures, it will then start to decline. But because it’s now at a much higher level, it won’t get down to (near) zero until the start of August.

The chart below is a rough, stylized version of this projection. It’s based on the CDC’s prediction of 3,000 deaths per day by early June, with deaths continuing to increase from there:

This is very approximate, but the important part is that COVID-19 deaths will continue to plateau for another couple of weeks or so. Only after that will they start to seriously climb.

If this is what happens, it means that six weeks of lockdown have been wasted. By late May it will be obvious that we have to do it all over again, and this time the lockdowns will need to be universal and they’ll probably have to last at least two months. Maybe longer. (The decline I show in my chart is just a guess. It could easily be until September or later before we finally get close to zero.)

If the alarmists are wrong, our daily death toll will continue to decline steadily, and we’ll be ready for phase two by mid-June. I doubt this is what will happen, and anyway, phase two is aggressive test-and-trace. We’re not ready for that now, and we won’t be ready by mid-June either, since no one in the federal government seems to care about it. In other words, it’s probably bad news no matter what happens. But it’s way worse news if the CDC/alarmist projection turns out to be right.

One of Mike Pence’s aides tested positive for coronavirus today. President Trump says this is why tests are a bad idea:

This is why the whole concept of tests aren’t necessarily great. The tests are perfect, but something can happen between the test where it’s good and then something happens and all of a sudden— She was tested very recently and tested negative and today I guess for some reason she tested positive.

Trump has said before that he’s not crazy about tests because all they do is make him look bad. Now he’s going further: he seems unsure the tests are even measuring something real. They’re negative one day and then positive the next. Why? “For some reason.”

I wonder how far down this rabbit hole he’s gone? Does Trump even believe the pandemic is real? He can’t see it, after all. Does he figure that if everyone would just stop reporting the numbers then everything would be OK? Or maybe he’s taking a purely transactional view: testing is fine in the abstract, but not so good when it tags someone close by who might be used to humiliate him. There’s no telling. But whatever it is, it’s pretty obvious that the stress of dealing with COVID-19 is accelerating whatever mental deterioration he was undergoing already.

Trump: Katie, she tested very good for a long period of time and then all of the sudden today she tested positive… This is why the whole concept of tests aren’t necessarily great pic.twitter.com/iM9Xo3jddU

— Acyn Torabi (@Acyn) May 8, 2020