There is very good news. We have a second vaccine! Both Pfizer and Moderna’s vaccines have now shown 94%+ effectiveness in treating Covid-19. Not to be outdone by Moderna’s report of 94.5% effectiveness, Pfizer has its final results and they are excellent. 95% effective, 94% effective in people over 65, no major safety concerns. They plan to apply for emergency use authorization within days and likely already have done so by the time you read this. Estimates are that we can vaccinate 20 million people by year’s end.

There is also very bad news. We have more positive test results than ever before, more hospitalizations than ever before, and a substantially increased positive test rate. Deaths lag tests, but are on track to rise proportionally to the rise in tests. 

Within that bad news there is good news. There are signs that the Midwest in particular, where things are maximally bad right now, is getting ready to peak. My estimate for how bad things will get before they get better, and how long it will be before that happens, have both tentatively gone down somewhat. Bad news can still be good news if you expected even worse news.

There is also an important note before we get to business.

Last week in my weekly post, I said some things about the election that did not relate to Covid-19. Doing so broke the norms of my personal blog and the norms of LessWrong, where these updates are reposted. I felt it was the least bad option under the circumstances. I could have executed it better, but I stand by my decision, and accept the consequences. 

I do not intend to let it happen again. 

One of the consequences of breaking this norm is that it is even more important for me to strongly reassert the norm going forward. Tempted as I may become, I will be extra careful not to discuss politics except as it directly relates to Covid-19 or requires us to take precautions for our own safety. I also plan to enforce the rule strongly in the comments. 

If you feel I have crossed that line in this or a future post in this series, once you have verified no one else has yet called me out, please do not hesitate to call me out on that.

If you feel I have crossed that line in a post not in this series, and I didn’t explicitly acknowledge that I was doing a politics-necessary thing like a prediction market post, call me out on that too.

Thank you. Let’s run the numbers.

The Numbers


Sep 17-Sep 2310168932695399
Sep 24-Sep 309349902619360
Oct 1-Oct 779711032308400
Oct 8-Oct 1478212172366436
Oct 15-Oct 2180415912370523
Oct 22-Oct 2889517012208612
Oct 29-Nov 495619772309613
Nov 5-Nov 11108927122535870
Nov 12-Nov 181255293428181127

Deaths are rising everywhere. That’s baked in for at least the next few weeks. Deaths generally continue to reliably follow cases proportionally after a 14-21 day delay, and case counts have yet to stabilize. The Midwest number here is good news after last week’s giant jump, but mostly there are no surprises here.

Positive Tests

Sep 17-Sep 23540258538112773223342
Sep 24-Sep 30554969293210630027214
Oct 1-Oct 7567429724311017034042
Oct 8-Oct 146828412574411799538918
Oct 15-Oct 217557114985113323843325
Oct 22-Oct 289498318188115812357420
Oct 29-Nov 411268425291716709870166
Nov 5-Nov 11157495387071206380108581
Nov 12-Nov 18211222452265255637150724

Things are still getting worse everywhere, but the rate at which the Midwest in particular is getting worse is slowing down. Hopefully people there are starting to get the message. It’s also likely that herd immunity effects are having an impact. With positive test rates in the double digits, there’s little doubt a large majority of cases are being missed, and parts of the Midwest now have positive test counts imply the actual case counts should be very high, in some cases high enough that herd immunity should be rapidly approaching. If anything, there is the worry that the Dakotas are providing evidence that the herd immunity threshold is on the higher end of its potential range, because we are already at or near the lower end. 

Positive Test Percentages

9/10 to 9/162.41%5.99%11.35%4.49%
9/17 to 9/232.20%5.96%7.13%4.11%
9/24 to 9/302.60%6.17%6.18%4.27%
10/1 to 10/72.61%6.05%6.74%4.23%
10/8 to 10/142.57%8.14%7.09%4.75%
10/15 to 10/222.95%8.70%7.85%5.36%
10/22 to 10/283.68%9.87%8.58%6.46%
10/29 to 11/44.28%12.79%8.86%7.04%
11/5 to 11/115.56%17.51%9.89%8.31%
11/12 to 11/186.99%18.90%11.64%10.66%

Percentages and test counts tell the same story. A small relative increase in the Midwest that bodes relatively well, signs that aren’t great in the other regions. 

Test Counts

DateUSA testsPositive %NY testsPositive %Cumulative Positives
Sep 10-Sep 164,636,1405.8%559,4630.9%2.00%
Sep 17-Sep 235,737,9195.2%610,8020.9%2.09%
Sep 24-Sep 305,833,7575.1%618,3781.1%2.18%
Oct 1-Oct 76,009,8455.2%763,9351.3%2.28%
Oct 8-Oct 146,322,8655.7%850,2231.1%2.39%
Oct 15-Oct 216,439,7816.5%865,8901.2%2.52%
Oct 22-Oct 286,933,1567.5%890,1851.4%2.67%
Oct 29-Nov 47,245,6008.6%973,7771.6%2.86%
Nov 5-Nov 118,285,49510.6%1,059,5592.4%3.13%
Nov 12-Nov 188,924,33812.3%1,155,6702.9%3.47%

My prediction last week was 12.9% positive rate on 9 million tests. We got a 12.3% positive rate on 8.9 million tests. Which is still way worse than last week’s rate of 10.8%, but better than my expectations due to things stabilizing somewhat the last few days of this week.

For next week my best guess is 13.4% positive rate on 9.5 million tests. That’s still headed in the wrong direction, but at a slower pace. One must constantly redefine what counts as good news.

The basic facts are unchanged and worth repeating. It is bad out there and is continuing to get worse. Cases will likely increase for a while longer, deaths for a month longer than that. You need to decide whether you are willing to hold out until the vaccines are available. 

If you do not wish to get Covid-19, and I do not think you want to get Covid-19, then unless you have already had it, now is the time to be even more cautious than ever before (with the exception of the greater New York region in March/April). That means among other things: mask up, do everything outdoors whenever possible and especially don’t spend 15+ minute periods together with others indoors unless they’re in your pod, socially distance, avoid being in the direct path of anyone talking let alone yelling or singing, supplement Vitamin D. Most of all don’t do stupid things, like large indoor gatherings without masks such as a traditional Thanksgiving. Encourage others to do the same. 

Machine Learning Project

They have relaunched (hat tip: Nate Silver) after a hiatus, and in this new iteration the site is focused entirely on now-casting, to answer the question of how many people are currently infected or have been infected. 

Here is their best guess.

For the United States, their answer comes back 631k infected yesterday and 560k infected five days ago (versus 164k positive test results yesterday). That is less cases than I would have expected, and unfortunately if true it implies both that herd immunity is not developing slower than I was guessing, and that the infection fatality rate is higher and remains around 0.5%. I still think the IFR is lower than that at this point, but this updates me somewhat towards less unnoticed infections. It’s plausible we are now better at knowing who to test, and we have more tests, so even though the positive rate is going up, the percentage of cases detected might not be going down. 

Their guess for total infections so far is 16.9%, with numbers in the high 30s for the Dakotas and 22.3% for New York. That also seems on the low end of plausible to me.   


Not pictured for usual reasons is Belgium, among others. If you look at Belgium, you see both a giant peak and now a giant rapid decline. What they are doing is working, and working fast. Positive test percentages are still troubling everywhere else, so there is worry that late reporting is making things look better than they actually are.

Deaths are climbing, but that was already baked in before lockdowns started.

This tweet alerted me to this website that gives excellent visualizations of Covid-19 in Europe. The effects of lockdowns are unmistakable. Again, we need to worry about late reporting, but the borders tell the story. 

Sweden is no longer the control group. An odd time to end such a valuable scientific experiment, but that’s how it goes these days. I suppose there wasn’t that much more to learn.

Go Away Or I Will Taunt You a Second Time

States enact more Covid-19 rules as infections are on the rise. 

The problem is that these new rules are both looking mostly in the wrong places and beyond toothless. 

Much of Europe went into strict lockdown. I was and am still skeptical that they were right to keep schools open, but it was a real attempt that clearly was capable of working, and it seems to be working.

The new American restrictions are not a real attempt, and have no chance of working. They presumably will make a small impact, what trying looks like. Contrast this with Europe, where trying is taking place. When the United States turns this wave around, it will not primarily be due to imposed restrictions, but instead due to some combination of voluntary behavior changes and people becoming immune.

When one sees someone claiming to address a problem, one can ask how many levels of ‘try’ and ‘pretend’ are involved. 

In this case, I believe that the attempts are pretending to pretend to try to try to solve the problem. They are not trying to try, but they are pretending that they are doing something that could plausibly pretend to be trying to try. Their hope is that this symbolically translates as ‘doing something’ and prevents the creation of common knowledge that there is no attempt at all. Ideally, it will allow those who do pretend to pretend to try to try to mark those who do not do this as blameworthy when things go badly, or even allow the meta-pretenders to claim credit when things improve. Ah, the joys of living in a high level simulacrum

The one exception might be California, which may be trying to try or even outright trying. That would be consistent with their approach so far. 

If I don’t write a post within a few months breaking down various meta levels of trying and pretending, please remind me to do so. It’s definitely worth doing.

In the meantime, be under no illusions that we are doing anything as a nation other than giving up and letting people fend for themselves, while hoping that enough people will be successful enough that the hospital systems will remain intact. 

And again, I am not convinced that this is the wrong decision. I am especially not convinced it is the wrong decision given the public choice problems involved. We lack the capacity to do enough to solve the problem before the vaccine arrives. So what is the alternative?  

That article from CNN also introduced me to a great new line.

Our New Motto: Don’t Share Your Air and Don’t Do Stupid Things


That’s a motto being used by Los Angeles Mayor Eric Garcetti, and I for one am here for it. I’ve been saying ‘don’t do stupid stuff’ but I am not good at marketing and stupid things is clearly better. Share your air also rolls nicely off the tongue and summarizes the most important secondary point (even if it is technically logically unnecessary, since it is a stupid thing to do). 

This correctly boils things down to their essence, in the style of You Have About Five Words, provided people can connect ‘don’t share your air’ to not doing things indoors. If they can also connect it to ‘wear a mask’ that might be a reach, but it is both at least plausible and even better. 

If there was more bandwidth available, next up in my priority queue would be telling people to supplement Vitamin D, but the American people’s available bandwidth is, shall we say, not high.

I cannot emphasize enough that people doing stupid things are most of the problem. 

Our failure to contain this pandemic is not about people making careful informed trade-offs and choosing slightly too much risk slightly too often. Nor is our problem an insufficiently detailed understanding of exactly how to go from somewhat safe to absolutely safe, or exactly how many feet apart to be or which direction to face or how long to interact, or wearing the wrong kind of mask, or any other such details.

Those details matter! They especially matter to you personally if you want to get the most out of life while taking the least risk. The difference between getting those details right or wrong can easily be an order of magnitude of risk. 

But here’s the thing. If you’re even thinking about those questions, you’re not taking that much of the risk. You have already cut out most of your risk compared to your pre-pandemic behaviors. Cutting your risk further could be worthwhile for you but that’s not how we win. It is like worrying about whether you are going to contribute to climate change because you used a paper bag that was slightly too large while looking up at a coal power plant. You are not the issue here.

And let me tell you. In what might be the ultimate evergreen statement, people are doing a lot of stupid things.

There’s a whole section called Thanks for the Hypocrisy later in this post about Thanksgiving plans, where 40% plan to have a gathering of 10 or more people. 

One nurse’s tale from South Dakota, treating patients that don’t believe Covid-19 is real. A huge portion of the public continues to treat Covid-19 as not being real, or no worse than the flu.

It seems Magic players in Oklahoma are trying to gather together for large tournaments on the same day their state ran out of ICU beds? Come on, everyone, we’re better than this.

First cruise ship to set sail since the pandemic has five people test positive for Covid-19. I am Jack’s utter lack of surprise.

Masks continue to be seen as a political statement, and are slightly annoying to wear, so huge percentages of people refuse to wear them.

Indoor dining continues in most of the country because of ‘the economy’ and because we don’t want restaurants to go out of business nor can we agree to give those restaurants money to not go out of business. So this week, Maryland did something about its indoor dining… and went from 75% to 50% capacity.

The New York Times reported on a wedding that had 200 guests and the fact that they had 200 people at an indoor wedding was both barely mentioned and also excused because the 200 people were ‘socially distanced.’ 

Others think that small groups and daily routines don’t count, so they felt like they were ‘doing everything right (WSJ).’ So they gather without masks outside of their home pods like it is nothing.

Most people think that certain types of activities are ‘safe’ and ‘don’t count’ in some important sense. That’s how people instinctively think and it is also the language authorities are using. One can follow all the guidelines and still have a two-household ten-person Thanksgiving, and in most places dine indoors with others several times a week, and so forth. 

The biggest category of ‘stupid stuff’ has been identified, and authorities seem to be converging on informal gatherings of friends of family, indoors, in their homes. This is the new thing to blame, which is the wrong framing to think about almost anything, but is also probably the big reason things are out of control. People don’t think of their friends or family as risky, they let down their guard, and then they’re wrong. 

The new stupid thing is doing intertemporal substitution exactly wrong. Lots of people see that the vaccine is on the horizon, that the end is near, and they start taking more risks rather than less. 

This is of course exactly backwards. An end to the pandemic raises the value of staying safe, and it lowers the cost of staying safe. So you should be safer and take less risk. But people’s minds largely don’t work like that. I’m not sure exactly what they do think instead. One possibility is that they implicitly have an idea for how much risk they are willing to take, and now that they won’t have to ‘spend out of their budget’ for that much longer, they are free to take more risk now. Another is that they get the message ‘things are better’ and act like things are better, without processing any implications at all, and that seems closer to the central thing going on to me. But I don’t understand and insights here are appreciated. 

All I Want For Christmas are a Covid Vaccine And a PS5 But They Underpriced Them And Now They’re All Sold Out

Moderna’s vaccine is 94.5% effective in preliminary results, with 90 symptomatic cases in the control group versus 5 in the treatment group. Consensus seems to think the results are ideal and very good news.  Here is the official press release. Moderna’s vaccine is similar to Pfizer’s, but has the advantage of not requiring storage to be in super cold freezers. Hence the building of giant freezers already underway. We could be doing much better on vaccine logistics and timelines, but we also are doing some things right. 

I have not seen it pointed out explicitly by anyone, but we have far more efficacy and safety data on both vaccines than we think that we do, because they are very similar vaccines. Many did realize this enough to note that they expected the Moderna vaccine to be effective due to the results we got from Pfizer, but it goes further than that. To a large extent, we can use the results from each vaccine trial as additional evidence about the other, and gain even more confidence that both vaccines work. Remember, we haven’t had a few failed attempts and two successes. Nothing has been put in a file drawer. We have two similar success stories and zero failures. 

The reason I am not making a bigger deal about ‘distributing these vaccines yesterday’ is that my understanding is that distribution and production have distinct bottlenecks, and production is already at the maximum regulations and our limited willingness to pay extra (along with any other bottlenecks) will allow. Thus, we are moving November vaccinations into December and January, but April vaccinations stay in April.

There are concerns that states do not have enough money for vaccine distribution (Washington Post). Obviously this is insane and the federal government should give states far more money than they actually need to avoid even slight delays in distribution. But it also shouldn’t matter. I know states are in financial trouble, but the vaccine is the way to make that stop, so states should spend the money if they have to, no matter what it takes, whether or not the federal government picks up the check later. Money is fungible.

One odd thing is that projected timelines did not seem to move at all when we went from one vaccine to two. That could mean that the supply chains both have the same bottleneck, or it could mean that everyone was assuming Moderna’s vaccine would work and it was already scheduled in. Or it could mean that we are scaling up by orders of magnitude, so one doubling now actually does not change the timeline much. 

“Vaccine chief Moncef Slaoui says there will be enough vaccines to immunize 20 million Americans in December.” My understanding is now that this is from a combination of both vaccines. 

An important part of my model has been that when people become infected, they provide much more effective immunity than you would expect at random, because people aren’t infected at random. 

With the vaccine, it depends on who gets the vaccine. If we base our decisions on who is most vulnerable and/or who is most eager to get the vaccine, immunity from vaccination is going to be much less effective at providing herd immunity than one would expect from random vaccinations. Those are exactly the people who are already being careful, so them being immune won’t make as much difference as we might like. The pandemic won’t end as quickly.

If we go for those who have jobs that put them at highest risk, the opposite happens. Every time we immunize an essential worker who has not already caught Covid-19, or someone who is otherwise taking a lot of risk, we are bringing things closer to a conclusion faster. 

It looks like we are going to do a combination of these approaches. Some essential workers will be at the front of the line, and our most vulnerable will be there as well. Those who want the vaccine the most and value it the most will mostly have to wait, but will likely comprise the ‘second wave’ that I will join. 

When will things be normal again? Some experts say surprisingly quickly: Fauci cautions ‘gradual return’ to normalcy by ‘second, third’ quarter 2021. To me, that’s surprisingly quick. As an individual, I might be ready for normal life in May, but that doesn’t mean normal life is ready for me. If we are less than a year from things seeming mostly normal, I will absolutely take that result. One could plausibly respond that the average person will hear ‘vaccine available’ as ‘pandemic over’ so any realistic timeline will sound pessimistic. It is also possible that this is a case of giving an optimistic prediction, but framing it as pessimistic because Very Serious People only allow pessimistic predictions. One must constantly be cautioning.

Immunity to Covid-19 For Some, Miniature American Flags for Others

I talked about this last week, and now have found some hard data. 

A new Gallup poll finds that 58% of Americans would take a Covid-19 vaccine, up from 50% previously but lower than before vaccine safety turned into a partisan issue. That seems about right for ‘would get the vaccine if it was offered incidentally during a check-up.’ And it is far worse in other places, such as in Spain where only 24% of people want to get the vaccine right away.  My guess is that ‘willing to go out of one’s way to get it’ is lower than that, and ‘willing to put in effort to secure part of a limited supply, maybe even wait on a line’ is lower still, and that it will be even lower once everyone starts talking about the side effects. Which are mild, but still mean there’s a substantial chance you spend the next day in bed feeling bad. A lot of people aren’t down for that.

Many sources, even well-intentioned ones, are not helping matters. Joe Rogan, on the nation’s most popular podcast, openly questioned whether one day feeling bad to get the vaccine wasn’t worse than getting Covid-19 if you are healthy, because his assistant and others he knows have had mild cases they barely noticed, and he noticed that Trump is out of shape, got Covid-19 and didn’t die. 

I even see how he got to that place, and it shows how hard it is to get a good model even when one is trying, as I believe that he is. Joe Rogan’s podcast is so popular for many reasons, but in part it is because he actually tries to use reason and his experiences to build up a full-of-gears model of physical reality in a way regular people can relate to. People are starved for that. I hope he is helping inspire others to think for themselves. He clearly has some important pieces of the puzzle that the people I know mostly lack, and he clearly lacks many important pieces that the people I know have mastered. I strongly disagree with him a lot. He gets a lot of things importantly wrong, but at least he is wrong, and I have no doubt I am frequently wrong too. An exchange of knowledge would be highly valuable. 

Despite all the problems convincing people to take the vaccine, as I noted last week, I do not expect this to impact distribution of the vaccine any time soon. Half the people are plenty of people until half the people are vaccinated, at which point we will know it is safe and we will pick up another 10%, at which point we will be at or near full herd immunity and we will have had plenty of time to use other methods to convince the remaining people, including requiring vaccination to participate in various activities if that proves necessary. 

In the meantime, this only makes it easier for those of us who actively want the vaccine yesterday.

Thanks for the Hypocrisy

Government officials rise as one to urge us to limit or cancel our Thanksgiving plans. They are, of course, correct. By that time things will probably be substantially more dangerous than they are even now. Dining and talking indoors for much of the day in large groups is one of the riskiest activities in terms of Covid-19 infection. Combining it with students returning from college will make it even worse, and many gatherings include our most vulnerable. It’s not a good idea. 

If you have thanksgiving plans outside of your pod and do not want your family to be infected with Covid-19, either everyone coming needs to quarantine for two weeks beforehand, or you need to cancel your thanksgiving plans.  

The people do not seem to be listening. “Nearly 40% of US residents plan to participate in gatherings of 10 or more people this holiday season despite concerns over the spread of COVID-19”. Here are some other numbers:

However, 73% of respondents said they would practice social distancing during the holidays and 79% suggested that they would celebrate or gather only with people with whom they live, the data showed.

Just over 80% indicated that they would ask family and friends invited to events not to come if they had symptoms of COVID-19.

I notice I am confused and a bit boggled. That’s a lot of people who claim to be living in groups of ten or more. 

It’s also almost 20% of people not asking those with symptoms of Covid-19 not to attend. 

Even after everything that has happened, that last one blows my mind. People who are sick should stay home and not attend gatherings. This isn’t a new principle. Even if you don’t think Covid-19 exists, even if the year is let’s say 2018, if you are coughing up a storm and feel terrible, stay home. This is not hard.  

Holding the holiday at all is different, even done indoors, without masks and without social distancing. Yes, that’s crazy risky. You should not do this. 

But I understand. I had to step in to tell someone in my family, coming to my own Thanksgiving that they had to isolate or they couldn’t attend, because it’s a damn hard thing to say to someone. That conversation can go very badly. Even I was tempted to let it happen. No one wants to be the villain who ruined Thanksgiving. It has been a long and lonely year. Thanksgiving is to many the second most important holiday of the year, second only to Christmas which is under similar threat. I totally, totally get it.

Even more than that, I don’t even think the decision is obviously wrong. From a personal perspective, the holidays are valuable to us, and who are we to tell people they aren’t worth the risk? People are choosing that risk, they have skin in the game and they are revealing their preferences to us. 

From a collective action standpoint, it would be better if we were all going to do our part to contain the virus. But it is clear at this point that we are not going to do that. How do you tell people to keep picking Stag when half the country keeps choosing Rabbit, if that’s how you view the payoff matrix? 

It’s damn hard! Here’s Germany’s recent attempt, which is worth a spoiler-free watch. On top of the obvious things, it’s worth noting that the message is to do something to improve the world rather than do something as a signal. Nor did it shame anyone. A sharp contrast with the American ads I saw in New York, and a welcome one. Not good enough to get it done, but good show. 

So you’re the nation’s politicians who want to help, and you don’t know what to do. You can’t do lockdowns or restrictions because the public won’t stand for it or listen to you. That bridge is burned. Physical action like advancing the vaccine is out of your hands. Especially if he has anything to say about it. 

I still have an important suggestion for you: Set a good example and don’t be a giant obvious hypocrite. 

As in, if you want to encourage people to be responsible and encourage everyone not to engage in the one activity we most need to stop doing, indoor dining outside one’s household, it would help to not be a giant hypocrite and hold indoor dinners for incoming congressional members of both parties. People notice. People remember. As they should. They tend to react like this. This account sums up my reaction. Canceling the event after people react to it won’t help much. 

Governor Newsom of California regrets attending a 12 person dinner at the French LaundryMore refusal to set a good example. More utter lack of skin in the game.

You might be a member of the New York City government who bragged about not limiting the size of your indoor gatherings. Which is ‘ok, fair’ but in his defense there are others who are also flaunting the rules who aren’t getting called out as loudly, which isn’t fair. Fairness is important.

At least eight members of congress tested positive this past week. Any individual person can be unlucky, but seven in one week makes it clear that good examples are most certainly not being set.

That’s only the incidents on a personal level that I noticed this week. 

The protests. The haircut. The funeral for John Lewis. The Rose Garden ceremony. The president resumed work in person while infectious. Non-existent contact tracing and notifications at the highest levels. The celebrations after the election. Indoor dining at up to 75% capacity. The list goes on. 

Maybe remember that the ‘little people’ who do the work count as people, and don’t keep saying they’re not there and no one was within six feet at least when we have video or photographic evidence.

While they close playgrounds and schools, and tell regular people not to celebrate holidays. 

This keeps happening. If there is to be any hope of convincing people to do their part, it needs to stop.

New York City Closes Schools

This story is, as usual, the embodiment of New York in the age of Covid-19.

Europe has mostly closed everything else but left schools open. 

New York City has a teacher’s union

Unions are not known for their flexibility. When schools reopened, it was agreed that they would close again if positive test rates in the city hit 3%. 

They hit exactly 3% yesterday, so today the schools are closed.

Hours before this, a reporter asked Cuomo if New York City schools would be closed, which caused him to tell the reporter they were ‘confused’ and generally go on a tirate.

Kids can’t go to school anywhere in the city despite many areas still being at roughly 1% positive rates, but they can go to a movie theater for ‘enrichment’ instead

De Blasio has no plan for how to reopen the schools, because ‘“This day seemed far off, thankfully.” Isn’t it great when your false impressions save you from having to do the work of being mayor? 

These stories always put me in a strange spot, because I am not in general in favor of schools, but closing schools while allowing indoor dining to continue is everywhere and always a stunningly major league screw-up. Schools so far do not seem to be a major source of transmission in the city, and many find them invaluable. Even if you agree with me that school is terrible, remote learning as implemented by schools seems to be far worse, and be designed to ensure kids don’t escape any of the tortures of attending school while not providing what benefits they were getting by being around other kids and occasionally even learning things. It’s almost as if the system acts to punish people rather than solve the problem.

I am confident that closing schools while leaving other things open is going to damage the kids involved, given the current state of remote learning, and enrage the parents especially those put into logistical nightmares, which is a lot of them. They will remember. 

And remember that if remote learning is making the whole class depressed, or otherwise destroying lives, that is a choice made by humans that can be undone by humans. Consider contacting the other parents in the class, standing as one voice, and saying no. Tell them what the new rules are, and dare them to fail the entire class over failing to sit there and be tortured. Or, of course, consider withdrawing from remote learning entirely, and using other methods.

In Other News

You know how you get people to give up entirely and let the pandemic run wild? One easy way is you say things like this, and tell people that even after they are personally vaccinated with a 95% effective vaccine, and wait the 14 days after the second shot, they still need to wear masks and socially distance. Not only wouldn’t I blame people for giving up if they thought that was the official Very Serious Person line, I’d think that from their perspective they were absolutely doing the right thing. 

Marginal Revolution reports that NY Times reports a 30-minute at-home Covid test has been approved, although a prescription is required because the cartel must be paid. 

A study came out in Denmark looking at the effectiveness of mask usage. It was underpowered, leaving it unable to confirm its hypothesis that mask use is highly effective, and you can guess how people are interpreting that. It looks like if you don’t use antibody tests, which in context introduce a ton of noise to an already underpowered study, you do find sufficient evidence to conclude the masks worked. This despite low power, highly uneven mask compliance (which is to be expected in the real world of course), and also everyone treating ‘what we can show with p<0.05’ as equal to ‘how effective masks are’ when evaluating the study. Plus no one can make the other person wear a mask, which is where the majority of effectiveness lies. All in all, it seems like even more strong evidence masks work, and full mask compliance would be more than sufficient to end the pandemic quickly on its own.   

Mayo Clinic has had 900 people infected over the past two weeks, with 93% of them getting infected while off the job, and the majority of the rest being infected in the break room while eating. Indoor dining is still super dangerous, and we get another data point that it is possible to take proper precautions when providing health care. If you want to do something safely badly enough, you can do it, and the professionals want it badly enough.

Dolly Parton helped fund the Moderna vaccine. Neat. No idea why anyone needed to do that, but still. Neat.

You remember that thing we heard about two weeks ago, that it seems we can mostly detect who has Covid-19 for zero marginal cost through analysis of an audio recording of a forced cough? And then everyone forgot about it, presumably because regulatory barriers mean it’s useless? It’s still there and still looks much more accurate than it needs to be in order to be the basis of a solution to this whole thing. It would still be a real, real shame if someone were to find a way to make this available for free online to anyone who wants it. A real shame! 

There’s also now a proposal to use smartwatch data, which can happen automatically for existing users. The question the paper does not yet answer is how many false positives the system would find. If that number is sufficiently low, and regulations don’t get in the way, this seems valuable. Until we get that information, nothing to see yet. But it’s more evidence for the theory that ‘Covid-19 causes a bunch of changes to most people and you can detect those changes in a lot of ways.’

Senator Rob Portman (R-OH) has announced he is part of the Janssen-Johnson and Johnson Phase 3 vaccine trial. Good for him!

Elon Musk gets tested four times for Covid-19 in one day, gets two positives and two negatives, shares news with the world with framing that “something very bogus is going on” via Twitter. As anyone who knows Bayes’ rule can figure out, he almost certainly had Covid-19. A few days later, he had to watch his latest SpaceX launch remotely. But also he put astronauts into space so let’s all go easy on him and wish him a speedy recovery. It’s all right that he has useful models of some parts of the world but not others. 

If you had 50%+ positive test rates and death tolls to match in the ‘what it takes to get Governor Doug Burgum of North Dakota to issue a mask mandate’ pool, congratulations, you’re a winner. It can be done.

Last week Cuomo told us he was going to try and stop distribution of the vaccine to New York until Biden was president. In a rare show of bipartisan cooperation, Trump won’t deliver it until Cuomo says he’s ready. Cuomo of course responded by threatening to sue. Biden has announced he is going to appoint a ‘supply commander’ to distribute the vaccine, which could be quite the point of leverage if one wanted to use it.

Remember February? When the incoming Chief of Staff Ronald Klain was one of many telling us things like “If you want to do something useful today, go to Chinatown — buy a meal, go shopping. The virus attacks humans, not people of any ethnicity/race. Fear is hurting Chinese-American owned businesses, baselessly. Let’s fight the disease AND let’s fight prejudice” or “We don’t have a #COVIDー19 epidemic in the US but we are starting to see a fear epidemic. Kudos to @NYCMayor (and others) for standing against that”? It’s important. We need to remember.

Surgeon General backs the ‘pandemic fatigue’ framing of why containment has collapsed. 

Marginal Revolution looks back at Economics and Epidemiology. All points made here seem right to me.

Minnesota Republicans test positive for Covid-19, alert Republicans but do not alert Democrats. SFailure to notify Democrats in the state legislature of Republican positive tests seems to also have happened in Pennsylvania. And Ohio. 

Thread in which a Very Serious Person points out things are bad, but also finds new evidence that antibodies are protective via a case study at a camp. Water is wet, sky blue, experts report. Which is valuable when many Very Serious People are continuously saying they are not sure.

More than 80 percent of prisoners in Carson City prison test positive for Covid-19. Reminder that a large percentage of prisoners have not even been convicted of a crime.

So this is weird, there’s a pattern of people who have had Covid-19 saying Coke no longer tastes good. How many lives will be saved? How many others improved? And presumably this wouldn’t only happen to Coke, so what is the general pattern? It would be great if this was a general awakening that artificial superstimuli were not good.

As one of those who fled New York City due to the pandemic, but who plans to return, I’ve been curious to get a good estimate for how many people left the city. My best guess was that a lot of people left Manhattan, with the majority of some richer areas emptying out, based on various anecdotal stories of empty buildings, but never found any hard data. The New York Post tallies the change of address forms and comes up with 244k such requests from March through July, versus 101k the previous year. So that would be a 143k change of request forms. The Post doesn’t then multiply by household size, but I think this is an oversight and should increase this by at least a factor of 2.4? Given that families have greater incentive to leave than singles, probably more. Rising crime certainly isn’t helping but that didn’t happen until after June and the spike was almost entirely done by then, so that’s more the Post showing its agenda than a major real cause. I do buy that the spike in departures from the Upper West Side zip codes housing new temporary homeless shelters is not a coincidence. 

“It is heartbreaking to see the politicization of sensible precautions. Think how quickly everyone agreed to take off their shoes in airports.” Because of one otherwise failed terrorist attack that resulted in zero casualties, and almost two decades later, in many places we are still doing it. Quite the example that got picked there. Maybe there is a reason people are skeptical when the government tells them what they need to do to stay safe. 


It’s awful out there. Stay safe. The vaccine is coming, but for most of us not until something like April or May.

Google estimates there are 18 million health care workers in the United States. We will have enough vaccine doses in 2020 to give to 20 million people. Adding in police and firemen would make it an even 20. Assuming that is indeed where we start, there won’t be much left for others, even the most vulnerable. There are 49.5 million Americans over the age of 65. 

Meanwhile, people are going to keep acting irresponsibly and doing stupid things. America is not going to do what it takes to get things under control in the next few months. The message here hasn’t changed. Time to buckle down. Will the medical system hold together? My guess is it probably will, but we are about to test it and find out. You don’t want to be part of that test. 

Don’t do stupid things.

New to LessWrong?

New Comment
29 comments, sorted by Click to highlight new comments since: Today at 8:41 AM

Re ML for audio of a forced cough to detect COVID, my prior here is that ~all new impressive results in the ML space fail to work in the real world. Translating from the lab to reality is just really hard in the ML space, especially once you are talking about things like audio which vary a lot depending on the environment and mic. 

All of which is to say someone should definitely build an app, but it's far from a slam dunk that it'll work.

The good news is that according to the paper, they are working with someone already:

"To that end, we have reached an agreement with a Fortune 100 company to demonstrate the value of our tool as part of their COVID-19 management practices."

Most of the value here will be in the third world, I think -- by the time this is ready to go we'll have access to vaccines in first world countries.

From my personal experience reading ML research papers and transferring the technology to consumer applications, the data used to test algorithms in papers tends to be far cleaner than real world data.

...stop distribution of the virus to New York...

I assume you meant to say vaccine here.

Interesting typo there!


The next two vaccines that will get approved (1-2 months after the first ones) are themselves adenoviruses!

Your don't do stupid things was largely me reaction to what some (many?) of the local restaurants seem to be doing. Outside dining started with spaced tables for nice weather. Then some covers to address sun/heat and shelter for light rain. Then as the weather has grown cold the tents are being enclosed. This last steps has always struck me as a WTF are you thinking here? Just because you have people outside in the parking lot, putting up that enclosed tent without any clear ventilation units to ensure the interior air is replaced with fresh external air is just trapping whatever people are exhaling within the confines. 

This strikes me as potentially worse than letting people eat inside where there is actually some form of ventilation and air filtration (though yes, might be something of a coin toss depending on what seat you end up occupying inside)

Outdoor dining in a tent has a lot more ventilation than being inside. But less ventilation than being fully outdoors. So I see your point, but it's about harm reduction. If you make people choose between being outside and unsheltered on a December evening and being inside, most of us will choose inside, so the tent, while probably more risky than fully outdoors, reduces the risk.

"An end to the pandemic raises the value of staying safe, and it lowers the cost of staying safe. So you should be safer and take less risk. But people’s minds largely don’t work like that. I’m not sure exactly what they do think instead."

One possible explanation would be that they don't really think, but they fantasize. About a future where the current inconveniences are over.

I know the research is in a slightly different domain (goal directed behaviour) but this situation reminds me of Oettingen's research about mental contrasting where solely fantasizing about a positive future can seduce a person to mentally enjoy this imagined future right now, decreasing the energy to go after a positive goal.

Much of Europe went into strict lockdown. I was and am still skeptical that they were right to keep schools open, but it was a real attempt that clearly was capable of working, and it seems to be working.

The new American restrictions are not a real attempt, and have no chance of working.

The way I understand it is that 'being effective' is making an efficient choice taking into account asymmetric risk and the value of information, and the long-run trade-offs. This involves things like harsh early lockdowns, throwing endless money at contact tracing, and strict enforcement of isolation. Think Taiwan, South Korea.

Then 'trying' is adopting policies that have a reasonable good chance of working, but not having a plan if they don't work, not erring on the side of caution of taking into account asymmetric risk when you adopt the policies, and not responding to new evidence quickly. The schools thing is a perfect example - closing has costs (makes the lockdown less effective and therefore longer), and it wasn't overwhelmingly clear that schools had to close to turn R under 1, so that was good enough. Partially funding tracing efforts, waiting until there's visibly no other choice and then calling a strict lockdown - that's 'trying'. Think the UK and France.

And then you have 'trying to try', which you explain in detail.

Dolly Parton helped fund the Moderna vaccine. Neat. No idea why anyone needed to do that, but still. Neat.

It's reassuring to know that if the administrative state and the pharmaceutical industry fails, we have Dolly Parton.

Tempted as I may become, I will be extra careful not to discuss politics except as it directly relates to Covid-19 or requires us to take precautions for our own safety.


I don't think an oops is necessary here in this case (beyond just not crossing the norm again), but this is still appreciated. Thank you.

Elon Musk gets tested four times for Covid-19 in one day, gets two positives and two negatives,

I guess I’m confused too. The rate of a false negative is that high?

Different tests have different false negative rates. Elon was using antibody tests which are faster at producing results but have higher inaccuracy. 

Futhermore it makes sense that there an amount of antibodies in the body where 50% of the tests you do pass the tests nearly regardless of the quality of the antibody tests.

I don’t understand the last paragraph. How does that work? Feel free to reply with a link to something if it’s too long to explain.

An antibody test gives a scalar value of how much antibodies there are. If you want to go from a scalar value to a binary value you have to decide on a cutoff that switches from infected to -not-infected. 

Given even a tiny bit of test inaccuracy, when the amount of antibody reaction is right at your cutoff, you will have 50% showing infected and 50% showing not not-infected.

Got it, thank you!


It was a rapid antigen test, which does have a very high rate of false negatives.  Depending on these tests without any barriers or isolation is how you got rolling White House outbreaks.

I would presume that many people here had bought stuff at Silk Way - it is puzzling that the discourse about antigen tests is still about the official approval and not about how you can import reliable tests from abroad. For my personal use I found online pharmacy in Germany that sends to Poland and does not check if I am really a doctor - but they probably don't send to US, so you need to find your own ways.

I agree! Access to cheap informal testing has made a HUGE difference in my social circles, and I don't understand why it isn't being used all around me. Life is so much easier when you can work around different risk tolerances by giving tests rather than breaking off contact.

Are the antigen tests that much harder to get hold of in other countries?

At the end of the article, you wrote "Adding in police and firemen would make it an even 20." Does adding EMS personnel change the estimate at all?

I Googled for "health care workers" and was assuming they would count as part of that 18 million, but didn't look more carefully than that. Some health care workers will refuse the vaccine, at least for a few months, and some have already had the virus, and it's all ballpark estimates. The basic idea is, you can do the consensus-essential people more or less in 2020, but that's your whole budget.

This post motivated me to order Vitamin D supplements, and write a thoughtful email to my family advocating that they do the same. Note: Mayo Clinic advocates around 600 IU/day for young adults and 800 IU/day for adults over 80. Too high a dose can apparently counteract the benefits. Most Vitamin D supplements on Amazon are in the 5,000-10,000 IU range. These ones are 400 IU/day.

The official recommendations are crazy low. Zvi's recommendation here of 5000IU/day is the number I normally hear from smart people who have actually done their research. 

The RCT showing vitamin D to help with covid used quite a bit. This converter from mg to IU suggests that the dose is at least somewhere around 20k on the first day and a total of 40k over the course of the week. The form they used (calcifediol) is also more potent, and if I'm understanding the following comment from the paper correctly, that means the actual number is closer to 200k/400k. (I'm a bit rushed on this, so it's worth double checking here)

In addition, calcifediol is more potent when compared to oral vitamin D3 [43]. In subjects with a deficient state of vitamin D, and administering physiological doses (up to 25 μg or 1000 IU daily, approximately 1 in 3 molecules of vitamin D appears as 25OHD; the efficacy of conversion is lower (about 1 in 10 molecules) when pharmacological doses of vitamin D/25OHD are used. [42]

I've always been confused why the official recommendations for vitamin D are so darn low, but it seems that there might be an answer that is fairly straight forward (and not very flattering to the those coming up with the recommended values). It looks like it might be a simple conflation between "standard error of the mean" and "standard deviation" of the population itself.

Thanks for that information. I'll pass it along.

Pure side note here. It seems that most of the vitamin and supplement producers all tend to go big -- which is generally less good than smaller, more frequent doses.

I've considered starting to directly source the chemical and then make my own doses for personal consumption. Still tossing that idea around in my head but leaning that direction the more and more I come across information as you just mentions.

"Deaths lag tests, but are on track to rise proportionally to the rise in tests".

At the height of the first wave we had 32,787 cases and 2,231 deaths (all 7-day averaged). That's CFR=0.068

Now (11/19/20) we have 164,996 cases and 1,266 deaths (same averaging). That's CFR=0.007.

Notice a whole extra "0" in CFR. Yes, x10 reduction.

Technically - very technically - deaths are indeed rising proportionally. But that's a technicality that is not relevant to anyone.

MA did strict lockdowns. MA is 3rd in the nation - still - on deaths per million. Maybe we should stop confusing how bitter a pill is with how effective the medicine is?


CFR will inevitably be much higher when you don’t have enough tests available.If memory serves NYC had 40-50% positive tests in the early peak. It is now running at 2.9%.


Deaths are highly correlated with cases 21 days earlier.  When cases are rising precipitously, it temporarily decreases the naive CFR.  When you take into account that lag, the apparent CFR of documented cases has been a remarkably flat about 2-3% since the summer.

To clarity I meant in the near term rather than comparing to March - if you want to look at the week over week death change this week look at the case over case change 2-3 weeks before that. Didn't mean death rates didn't fall since April.

If others think that was unclear I will edit to clarify.