Covid 5/13: Moving On

post by Zvi · 2021-05-13T14:30:01.849Z · LW · GW · 13 comments

Contents

  The Numbers
  Predictions
  Deaths
  Cases
  India
  Maybe It’s All a Coincidence
  The Worst Possible Thing You Could Do
  In Other News
  Not Directly Covid News
None
14 comments

For over a year, Covid-19 has been the central fact of life. 

The goal now is to make that no longer true. 

If you’re reading this, chances are very high you are vaccinated. If you’re not reading this, but you live in the United States, chances are still pretty good you’re vaccinated. 

The question everyone is asking is now, can life return fully to normal? 

Yes.

Well, almost. We’ll never be psychologically quite the same. We’ll never unlearn the lessons we’ve learned over the past year – nor would we want to – about the way our civilization and its institutions work, or about what matters in life. And at least for now, we’ll need to continue to worry about how others will interact with us and how to navigate people’s concerns and various governmental restrictions, as well as the Covid-19 conversations that they’ll doubtless want to have for a long time. If others don’t return to normal, there’s no fully normal to return to yourself.

And that doesn’t mean quite fully normal quite yet, in the sense that there’s an amount of indoor crowding I’d still be inclined to avoid for the next few weeks or months. 

And of course, things may be going well in America and most other highly vaccinated places, but the worldwide pandemic is far from over. Things in many other places remain quite bad, and will be quite bad for some time.

But… mostly? Yes. For those who are fully vaccinated, life can safely return to normal. 

This column can also, events willing, start winding down or transitioning to other matters. If things go as planned, there will steadily be less Covid-19 news to talk about each week, and I can shift my blogging time into other, longer-term pursuits once more.

For now, let’s run the numbers.

The Numbers

Predictions

Prediction from last week: Positivity rate of 3.5% (down 0.4%) and deaths decline by 7%.

Result:

Things improved faster than I expected, which is great. The fall in deaths makes perfect sense, as I was adjusting for strangely small drops in deaths from previous weeks. The fall in cases is mostly in line with expectations. 

Prediction for Next Week: Positivity rate of 3.0% (down 0.4%) and deaths decline by 10%. 

This looks like the endgame. The control system is the potential threat to that, as we’re likely not near full herd immunity if everyone went fully back to the old normal, but I do not expect people to return to the full old normal. We will still pick a lot of the truly ‘low hanging fruit’ and I think that will go far enough to bring us over the top. 

Deaths

Date	WEST	MIDWEST	SOUTH	NORTHEAST	TOTAL
Apr 1-Apr 7	1098	867	1789	1160	4914
Apr 8-Apr 14	1070	1037	1621	1145	4873
Apr 15-Apr 21	883	987	1747	1168	4785
Apr 22-Apr 28	752	1173	1609	1110	4644
Apr 29-May 5	943	1220	1440	971	4574
May 6-May 12	826	1069	1392	855	4142

We saw a steady decline across all regions, and I see no reason not to expect this to continue.

Cases

Date	WEST	MIDWEST	SOUTH	NORTHEAST
Mar 25-Mar 31	49,669	93,690	102,134	145,933
Apr 1-Apr 7	52,891	112,848	98,390	140,739
Apr 8-Apr 14	60,693	124,161	110,995	137,213
Apr 15-Apr 21	54,778	107,700	110,160	119,542
Apr 22-Apr 28	54,887	88,973	97,482	78,442
Apr 29-May 5	52,984	78,778	85,641	68,299
May 6-May 12	46,045	59,945	70,740	46,782

That’s a dramatic acceleration of improvement, especially in the northeast. Progress is much slower in the West, but even 10% a week adds up quickly and there’s still a lot more vaccinations to bring online. This is what the endgame looks like. 

India

This looks about as good as one could have hoped. The control system is indeed powerful, and a few things like the elections have finished up which is presumably helping. There’s some worry about the numbers reflecting limits on the Indian state’s capacity to test and measure, rather than a real limit, but the test percentage is static so I mostly believe that things are no longer getting worse. That doesn’t mean they’re getting substantially better yet, and the whole situation remains quite horrible, but things could have been much, much worse and it looks like those scenarios are going to be avoided. 

My old colleague Ross Rheingans-Yoo asks how much demographics should lower India’s death rate from Covid, concludes about a 61% effect but warns that lack of medical care and abundance of air pollution could undo this.  

Vaccinations

We have now approved the Pfizer vaccine for children 12-15. Surveys say that there will be a lot of reluctance to vaccinate children, and public service approaches like ‘first thing talk to your child, they are sure to have an opinion’ do not seem likely to be helpful in solving this problem. As with vaccination in general, there will be a core of people desperate to get the vaccine for their children one day sooner to feel safe and let their lives move closer to normal, then varying degrees of reluctance and skepticism. A lot of people are using the ‘wait and see how it goes’ option in surveys, despite this attitude making very little physical sense. We are not going to learn anything here.

This is excellent news. We had a steady decline in doses, and that decline has at least temporarily stalled out despite less people needing their second dose and despite more people having already been vaccinated, and before the 12-15 year old crowd started getting their shots. That both can make us more optimistic that there’s still a lot of people not yet vaccinated who are going to end up vaccinated, and it also is more evidence that the J&J pause was a large part of the problem, which is hopefully now receding somewhat in people’s minds. That’s because the alternate hypothesis was that we were running out of remaining willing arms in which to put shots and this was a natural peak, but if it was a natural peak the decline seems like it should have continued. It’s not definitive, but it is definitely suggestive.

Maybe It’s All a Coincidence

The following series of events occurred a few weeks ago:

1. The J&J vaccine was suspended for no (good) reason.
2. A lot less people suddenly got vaccinated in America.
3. Number of people vaccinated kept going down rapidly.
4. I and many others concluded #1 contributed a lot to #2 and #3.

As we all know, correlation does not prove causation, and first doses had already peaked before this. There’s a not-completely-impossible case one could thus make that the J&J pause didn’t contribute meaningfully to the decline in vaccinations, and things are not substantially different from the counterfactual other than the slowdown specifically in J&J doses.

There’s survey data out in support of this thesis:

Note that this is ‘no evidence’ versus there never being a blood clot concern at all rather than being versus the alternative world where the (non)-issue still existed but they didn’t suspend the vaccine. 

I’m going to go ahead and propose this:

Law of No Evidence: Any claim that there is “no evidence” of something is evidence of bullshit.

No evidence should be fully up there with “government denial” or “I didn’t do it, no one saw me do it, there’s no way they can prove anything.” If there was indeed no evidence, there’d be no need to claim there was no evidence, and this is usually a move to categorize the evidence as illegitimate and irrelevant because it doesn’t fit today’s preferred form of scientism. 

You can certainly say that the survey data is an important challenge to the claim that the pause mattered!

But also, you gotta look at this graph and think…

…maybe not exactly this…

But still something like… okay, sure, that’s at least some evidence. There’s no way you can look at that chart and think that the week after the line matches the previous trend. 

Or, alternatively, you could look at all the specific people I know about who expressed uncertainty surrounding the pause, or all the work I’ve been forced to do about the pause, and the professional time I’ve spent advising on how to deal with questions about the pause and how it is a substantial fraction of any draft of a vaccine FAQ at this point. 

I also don’t want to think about the smoothing effects it took to get the hesitancy graph above to be all straight lines. But let’s give the graph the benefit of the maximum benefit of the doubt, and assume that people really did give similar responses before and after the pause. What’s going on with that?

The graph presumes we can divide people into four categories: Already vaccinated, enthusiastic, hesitant and resistant. 

The first thing to note is that the top line, ‘already vaccinated,’ doesn’t look right, because it looks like a smooth increase rather than an increase that is dramatically slowing down over time in the later part of the graph, but let’s assume that this is a resolution issue and the slowdown is present.

The enthusiastic line seems to be continuing to decline slightly, as there are more enthusiastic people getting vaccinated than there are others moving up to enthusiastic. It’s suspiciously flat. That could be a kind of coincidence, but if there’s a mostly static proportion of people who are ‘enthusiastic’ about vaccination, and a mostly static proportion of ‘hesitant’ as well (it’s declining, but much much slower than vaccination rates), what’s causing such a dramatic drop in vaccination rates? 

The response presumably would be that this is select effects based on logistics. People can say they are ‘enthusiastic’ all they like, but if you’re enthusiastic now without having had a dose, that increasingly means you’re running into some sort of logistical issue stopping you from getting a shot. You can’t navigate a computer, or you can’t get time off work, or something. Over time, that gets worse. And that’s then the effect causing the slowdown.

A question for all such graphs is what shape the peak looks like, and a common mistake is to presume a straight line up then a phase shift into a straight line down. Here, we have a very slowly increasing line where we were (by all reports) mostly supply constrained the whole way, with a supply-related dip in the middle, and then suddenly a steady decline. 

So the story is that there are two limiting factors, supply and demand. We were supply limited until the peak, then suddenly demand limited, and now we’re seeing a linear decline in effective demand due to logistical concerns.

My obvious first question on that would be, there’s a dramatic logistical easing the moment supply is not a binding constraint. Not only did we open almost all appointments to everyone, we also made walk-ins available, including at local pharmacies. If logistics are a major issue, then there have to be a lot of people who couldn’t afford to make an appointment, but I’d like to think almost everyone can find time at some point to show up at their local CVS, even if it takes a few tries due to lines, and as demand falls those lines will get shorter and rarer quickly. 

I do get that in disadvantaged communities, getting the information across can be hard, so presumably this isn’t entirely lizardman constant or social desirability bias. 

We then get to a slowly rising number of ‘resistant’ people, while the number of ‘hesitant’ people slowly goes down, which makes sense. Hesitant people can and often are persuaded, whereas fully resistant ones rarely change their minds, and occasionally hesitancy becomes resistance. If we extrapolate these lines, we’d expect to stabilize at something like 70% vaccinated, at which point we’d only make progress by coercing the resistant into compliance (or somehow persuading them, which seems harder and less likely). 

That’s a consistent, reasonable model, except for the problem that it pretends that the pause and blood clot issue didn’t matter when I have direct observations that say that it did matter and the observed number of vaccinations doesn’t match the survey data. Bubkas, it says, whereas my eyes and ears and the other data points say no, not bubkas. 

Can we potentially reconcile these data points without saying the survey is wrong?

Not quite entirely, but if we can fix the number of vaccinations to match the data then I think we… mostly… can? The way we do that is to notice that there are essentially three categories of attitude here, and all that’s being tracked is the transition between those categories. Which means that you only get ‘noticed’ here if you move from one to the other. 

Thus, if the pause issue was mostly a within-category move, it might not show up much here. If you were enthusiastic before, you’ll say the same now, despite likely being less enthusiastic. If you were hesitant, this makes you more hesitant, but doesn’t really transition that many people into ‘resistant.’ And if you’re already resistant, then obviously nothing changes. 

You’d still expect some flow between categories, but it could reasonably be quite small, or be countered by other effects, such as there being ‘enthusiastic’ people who were going to get J&J, couldn’t get it, and thus stay enthusiastic, maybe? Feels like a reach. 

The other explanation is that there’s at least some disconnect between what people say on a survey and what they do in practice. I can imagine a world in which this disconnect is substantial. People don’t want to say they’re less enthusiastic, because the whole issue is clearly dumb and doesn’t impact Pfizer/Moderna, or something similar, but it still leads to a bunch of FUD that causes them not to actually make the vaccination happen. Or they still want to do it, but figure they’ll wait until this is sorted out, and then keep waiting.

The problem with that hypothesis is that it feels like it should also have shown up in the observations me and those around me made about individuals who are expressing new hesitancy? That’s the problem I can’t explain.

I wouldn’t have predicted the survey results to not show any impact, and was surprised. There’s at least some chance that this reflects something I didn’t understand, but there’s also a chance the survey was flawed, or the curve-fitting used was flawed, or both. I’m not sure.

The Worst Possible Thing You Could Do

I don’t have much I want to add to my case from last week that normalizing the expropriation of intellectual property for exactly the kind of intellectual property we most want to exist in the future, and the resulting additional normalization of the expropriation of exactly the remaining profits that are creating incentive to do the most good, might be a really bad idea if you wanted you, your children and those around you to live a long and healthy life.

These are important questions to understand and get right, but they’re also inherently political and economic questions. Continuing along these lines would be outside where I want these posts to focus. 

In Other News

People think that being virtuous reduces COVID risk. When presented with identically risky scenarios, subjects judged risk to be higher when motives for action were bad.   

I can steelman this difference a bit. Consider the vignette where Joe is headed out either to help an elderly neighbor or to buy drugs. These are not identical distributions of Joes. The Joes headed out to buy drugs likely live in poorer neighborhoods with higher Covid risk. The Joes helping elderly neighbors likely live in relatively richer and safer areas. Thus, given we see a small effect, that small effect could easily exist in reality. 

You could extend that to arguments that those you see at grocery stores or doctor’s offices are relatively low-risk, whereas the people at a beach are relatively high-risk, even if the beach itself isn’t risky, because they’re the people who are ignoring the (blatantly terrible, misleading, dishonest!) public health advice. What other rules are they breaking?

Here’s the chart she linked to:

The idea that grocery shopping is in the low-moderate group with ‘eat outdoors at a restaurant,’ ‘going for a walk’ or ‘playing golf’ and a step safer than ‘going to a beach’ is obviously not reflective of real risk. The first time I went to a grocery store during the pandemic, it was clear that it was more risky than everything else I’d done for weeks, combined. Rating it as low-moderate risk is reflective of the fact that it’s necessary for most people, because they can’t afford not to do it, so authorities want them to think it’s low risk so they will not start thinking ‘well, I already do this at least moderately risky thing already every week’ so what’s the harm in a little something else? And they don’t want people freaking out over buying groceries for their families because what are you gonna do if you can’t afford instacart? Then many of those same people get home and wash all those groceries, some to this day.

Meanwhile, even with the CDC and WHO finally acknowledging how Covid actually spreads, it’s going to be tough to break old habits.

Meanwhile to that, in this word of ‘figuring things out and physically modeling the world’ we have to remind ourselves that these facts were not fully in evidence (and there’s a thread here about the history of how that happened for those interested):

I was aware they weren’t acknowledging Covid was airborne, but it somehow didn’t feel important because who would be listening to those jokers and taking their explicit physical world models seriously? Either you believed in physical world models, in which case these people were obvious lying liars whose models didn’t make sense and should get ignored, or you didn’t, in which case all that mattered was some elite consensus on how to act that didn’t care what the WHO’s or CDC’s physical model might be, and at most cared about their ‘recommendations’ which of course were a horrible mish-mash of nonsense even if their claimed physical models were accurate. 

The test of whether my failure to notice this was correct is whether this change does anything. Congratulations, you’ve finally admitted that Covid spreads in exactly the way we’ve known for a year that it does, what’ll you do next? Go to Disney World and hang out at the beach? Or will you change nothing? So far as I can tell, there hasn’t exactly been a rush to update the safety recommendations, let alone any sign anyone is actually changing behavior much. 

Also important is that the regulations surrounding airborne diseases are different than those for non-airborne diseases, so these fact claims may have been more about what regulations the CDC wanted to offer than about what was physically happening.

Or, basically, this, which solicited this response from the CDC head:

At any given time, 2% of people infected are 90% of the infectiousness, says a new paper, because people are typically only super infectious for a short time. I haven’t looked in detail, but seems plausible. Doesn’t mean you can know which 2% it is easily.

Bryan Caplan compares what happened to his expectations. His analysis for individuals (which he gives a 2nd percentile grade)  is based on the assumption that paranoia about Covid makes no sense and that anyone (such as myself) who spent a year avoiding Covid was acting grade-A stupid. Part of this is that he doesn’t think the risk is serious and thinks that this assessment should be obvious. Part of that is that he implicitly saying that people should ignore social effects of their actions – without the actions he’s condemning, there would have been completely uncontrolled spread and either the government would have clamped down super hard or we’d have had a hospital collapse, and he doesn’t even mention such things. And part of this is that he thinks everyone should accept that the key to and ‘most important source of’ happiness is in-person human interaction, which… well, sure, it’s helpful when done well and I’m happy to get it back, but let’s not go nuts. He also doesn’t think about whether our prioritization of containment strategies, and our coping with the situation, were better or worse than expected, and relative to expectations under these conditions I think people mostly did well. His other assessments seem reasonable to me. We agree business did even better than we expected, and that governments in most places did even worse than expected, and I put individuals close to businesses. But then I’m one of the individuals he’s giving a low grade to! 

Yes, full non-emergency authorization matters, do it now, New York edition:

In addition to the schools and the army edition:

Two unrelated matters:

Paper on Texas school reopenings and their impact on spread. Conclusion was that opening schools made adults more mobile, thus accelerating spread. That suggests opening schools was right, since using schools to force people to stay home seems like a very non-optimal (and highly regressive and miserable) way to impose a de facto lockdown, with a possible exception if driving people to/from school was directly causing other interactions, but likely not even then. It’s theoretically possible this is a second-best (third-best?) solution anyway if there’s no political way to impose alternative restrictions?   

College students sent home with full loss of tuition because they took a picture outside without masks.

Paper about Covid skeptic community finds that they highly value… actual scientific inquiry and analysis of data, and thinking for yourself, and are happy to help others think for themselves too. Then laments this ‘refusal to accept the science as settled’ and compares them to those who attacked the capital on January 6. You see, if you get the Wrong Answer, you’re illegitimate, so any use of the proper legitimate techniques is doubly illegitimate and terrible. You might give people the idea that these are tools people use to analyze data themselves and come to conclusions about how to model the physical world. Can’t have that.

What happened to the Novavax vaccine? A combination of the Defense Production Act blocking raw materials and a general raw materials shortage. Which still doesn’t explain why it hasn’t applied for FDA approval yet.  

Analysis of the NHS Covid app and its effects estimates it prevented hundreds of thousands of infections in the UK. 

Uber and Lyft are offering, free of charge and with no government payments involved, free trips to and from vaccination appointments. Kudos to them. So if you’re claiming the logistics don’t work, there are free taxis on standby and walk-ins available. The excuses are getting rather thin. 

Not as good as free Mets tickets but nothing wrong with a good old fashioned twenty:

I do worry about some people getting too smart for their own good and holding out for more.

So how about One Million Dollars:

This is absolutely correct policy and amazingly great and I’d love to see other states and ideally the federal government follow suit. All the incentives align. 

Vitalik gives epic crypto donations, including 1,000 ETH to MIRI, another 1,000 ETH (plus his ELON) to the Methuselah Foundation, more than 13k ETH to GiveWell, and one billion dollars (yes, billion with a B) in SHIBA at its pre-donation market cap to the Indian Covid Relief Fund. That’s 10% of the entire SHIBA blockchain. Looking at the replies to the relief fund’s announcement of the donation and also the current graph does not exactly inspire confidence that a billion dollars of fiat are going to result from this, and it does not bode well for other holders of this second-rate version of a meme coin whose very thesis is war upon the very concept of sensemaking and logic, and I am totally, totally fine with every last bit of all of that. 

There were always bound to be casualties. 

There’s an obvious trade here as I write this. Don’t do it.

Hopefully this encourages rather than discourages future creators of cryptocurrencies considering giving tons of surplus coins to Vitalik. Seems like that’s a great system!

Not Directly Covid News

Not directly Covid, but relevant given we want to understand how the media chooses to cover things. You see, there’s a gasoline problem, in the sense that if you go to the gas station to try and purchase gasoline, chances are high they won’t have any and that’s going to be a problem. You would think this would be a really big story, and yet Robin’s experience matches my own…

The Washington Post website had actual zero stories on its main page about gasoline when I started writing this section, although it added a story later, and the next day had one about the political blame game. Neither did the Wall Street Journal or Bloomberg. CNN had one, and then it went directly into telling people everything is fine and not to ‘panic’ by buying the thing there’s a shortage of but which is legally not allowed to rise in price (and an implicit framing that this is something that happened to the administration, which I am sure would not have been the framing if this had happened last year):

Check out these headlines for those posts:

What in holy hell? This is not a demand spike! This is a supply shortage! There’s no supply, and people are buying the not-price-adjusting supply that remains because it’s going to run out, and they’re framing the issue as a supply shortage. Why is everyone withdrawing their money from this bank after someone robbed them of half their money? You people are yelling at us for no reason, you all need to calm down. 

Fox news mentions it, but the biggest mention is as a metaphor:

Then later they have a literal story about it. Those are all the news sites I checked, there’s no selection going on here. 

The obvious harmless explanation for all this is that perhaps the pipeline disruption was minor – the claim is that supplies will mostly be flowing again by the end of the week – and it didn’t cause much supply disruption, and all we’re dealing with is a few areas where people started hoarding gas and things will be annoying for about a week. So in some sense (the sense in which you have zero respect for the people buying gas responding to very clear incentives) this sort of is a demand-side problem and not worthy of much coverage. But also covering it more would make it happen more so it’s easy to understand the desire to downplay it a bit. I’ve similarly downplayed or ignored stories about Covid for similar reasons, for example when someone with a large audience started talking about how they didn’t ‘need’ the vaccine and encouraging people not to get it, I asked myself ‘how would talking about this help anything?’ and didn’t mention the Person Being Wrong On The Internet. 

Still, we have photos of the New York Times saying there are no gas stations without gas at the same time as there were definitely a bunch of stations without gas, and a general (implicit) conspiracy not to talk about what really should be a pretty big issue. Which should update you in favor of the media doing similar things in other situations, in the past and in the future. 

Overall, I’m going to have to go with ‘unless people are flat out flying about physical conditions something that matters a lot is happening here’:

That account seems great if you want to follow what’s happening:

That was not the peak of the trouble, that was Wednesday morning. It will take several days, by all reports, to fully restore service. 

So of course the governor of Georgia responded to this supply shortage by… suspending the state’s gas tax.

Meanwhile, in ‘Zvi is tempted to offer a simple policy solution’ discourse…

I also wonder what would be happening if we still had the previous administration, and it seemed like there was a problem and the public response was literally nothing. Presumably questions would have gotten asked. So that suggests that we should be far more worried this year than last year about problems getting buried. 

This whole episode was of great practical interest to me, because the plan was for my in-laws to be driving with my children down from New York to Florida for a vacation, directly through the area that does not have reliable access to gasoline. This was very important news for us to know, and knowing it allowed us to book flights instead. That’s exactly what you want to happen, with people reallocating consumption away from scarce resources to use other resources, but the media decided to do its best to bury the information. 

I do understand that other people react in other ways that one might reasonably want to minimize. I sympathize. It’s not that there isn’t a positive motivation here… 

There is also the issue of how this whole episode was allowed to happen. Our pipelines, it seems, can be shut down by cyber attacks for substantial periods. It could have been a lot worse, as there wasn’t an environmental catastrophe and things will be back online soon, but we really, really should be treating this as a sign that we need to take such risks a lot more seriously. As opposed to, say, our warning signs about a possible coronavirus pandemic.

Not directly Covid, but if you want to know how disconnected from physical reality politicians can get, this is what happened when candidates for New York City mayor were asked about home prices in Brooklyn. Then Yang got several questions like this exactly right while using strong physical-world explicit reasoning to cover up what was presumably frantic Googling of the answer or previous Googling because the questions had already been leaked. But compared to someone who spent years in charge of ‘affordable housing’ and wants to be mayor being off by a factor of 10 on what housing costs, give me the person who knows what actual physical-world Fermi estimation looks like and uses it to cover up looking up the right answer any day. The best part is when one of the candidates doubles down and says ‘including apartments?’ when told the right answer. Alas, I didn’t get a chance to guess unanchored, and different parts of Brooklyn are sufficiently different that the question is tricky. Just not ‘missing a zero’ tricky.

Not directly Covid, but relevant given how much we wonder about the origins of what looks like incompetence that California is going full Harrison Bergeron and sabotaging advanced math education because knowing too much math leads to inequality. In other news, popular claims that ‘urgency’ or ‘objectivity’ are white supremacist concepts might also be what we call ‘not helping.’

Not directly Covid, but highly relevant to note that a lot of people are optimizing purely for ‘normal’ and will attack anything they see as different from ‘normal,’ even if their ‘normal’ is in the minority. This is distinct from the mode where there is a righteous agenda of some kind, and anyone not onboard with that agenda is history’s greatest monster, except that one of the strategies of such agendas is to code acceptance or fear of them as ‘normal’ to trigger the first group. 

Not directly Covid, but highly relevant is perhaps the greatest Twitter reply of all time.

To which there can only be one reply, which is perfectly understandable:

It’s for the best.

13 comments

Comments sorted by top scores.