Prediction last week: 14.3% positive rate on 9.7 million tests, and an average of 2,500 deaths, again with wide error bars.
Results: 16.4% positive rate on 9.3 million tests, and an average of 2,657 deaths.
The phase shift on 12⁄30, in the wake of Christmas, seems to have been real, giving us the clear holiday bump we did not see from previous holidays.
That was both the very bad outcome for infections I was worried about, and also not high on my list of things to be concerned or furious about this week.
For deaths, my estimate was lower than it should have been and I should have assumed a full reversion, so on reflection it’s my mistake rather than especially bad news.
The new strain is not yet prevalent enough to be noticeably impacting the numbers.
Prediction: 17.0% positive rate on 9.5 million tests, and an average of 2,800 deaths. The holidays are over, there will be some fallout, with things getting slightly worse, but with the main boost in deaths from Christmas mostly coming later.
Deaths
Date
WEST
MIDWEST
SOUTH
NORTHEAST
Oct 29-Nov 4
956
1977
2309
613
Nov 5-Nov 11
1089
2712
2535
870
Nov 12-Nov 18
1255
2934
2818
1127
Nov 19-Nov 25
1761
4169
3396
1714
Nov 26-Dec 2
1628
3814
2742
1939
Dec 3-Dec 9
2437
5508
4286
2744
Dec 10-Dec 16
3278
5324
4376
3541
Dec 17-Dec 23
3826
5158
5131
3772
Dec 24-Dec 30
3363
3668
4171
3640
Dec 31-Jan 6
5320
5036
6072
4986
This is the one place it’s not as bad as it looks. The data source for these numbers is Wikipedia, which shows a relatively large amount of shifting of deaths from last week into this week.
If we assume that a lot of this week’s deaths were actually last week, that explains much of the increase. It’s not good news or anything, it’s definitely bad news, but it is not full on terrifying like it would be if we didn’t know about Christmas.
We still should expect further increases in the next few weeks before the tide likely temporarily turns.
Positive Test Percentages
Percentages
Northeast
Midwest
South
West
11⁄5 to 11⁄11
5.56%
17.51%
9.89%
8.31%
11⁄12 to 11⁄18
6.99%
18.90%
11.64%
10.66%
11⁄19 to 11⁄25
7.00%
16.62%
10.41%
11.75%
11⁄26 to 12⁄2
8.38%
17.90%
12.45%
12.79%
12⁄3 to 12⁄9
10.47%
17.94%
13.70%
12.76%
12⁄10 to 12⁄16
10.15%
15.63%
15.91%
13.65%
12⁄17 to 12⁄23
9.88%
14.65%
15.78%
13.82%
12⁄24 to 12⁄30
10.65%
14.54%
17.07%
12.90%
12⁄31 to 1⁄6
12.18%
17.03%
19.69%
15.94%
No way to sugar coat this one. Everything is pointed in the wrong direction in a clear phase shift. The good news is that it could be a one-time move from Christmas, and thus we can still plausibly expect things to remain stable from this point forward without waiting for further control systems to kick in. I hope to start seeing slow improvements soon, but I must admit that the overall graph is not as encouraging of that as I thought a few weeks ago, especially outside of the Midwest, and the turning point may be farther off than we’d like. The Midwest is still probably past its peak for this wave, even if everyone else has to wait a bit.
Positive Tests
Date
WEST
MIDWEST
SOUTH
NORTHEAST
Nov 12-Nov 18
211,222
452,265
255,637
150,724
Nov 19-Nov 25
269,230
435,688
294,230
170,595
Nov 26-Dec 2
256,629
357,102
294,734
185,087
Dec 3-Dec 9
354,397
379,823
368,596
263,886
Dec 10-Dec 16
415,220
315,304
406,353
260,863
Dec 17-Dec 23
439,493
271,825
419,230
236,264
Dec 24-Dec 30
372,095
206,671
373,086
225,476
Dec 31-Jan 6
428,407
251,443
494,090
267,350
Test Counts
Date
USA tests
Positive %
NY tests
Positive %
Cumulative Positives
Nov 5-Nov 11
8,290,417
10.8%
1,059,559
2.4%
3.16%
Nov 12-Nov 18
9,040,426
12.4%
1,155,670
2.9%
3.50%
Nov 19-Nov 25
10,419,059
11.8%
1,373,751
2.9%
3.88%
Nov 26-Dec 2
9,747,026
11.8%
1,287,010
4.0%
4.23%
Dec 3-Dec 9
10,465,254
13.9%
1,411,142
4.9%
4.67%
Dec 10-Dec 16
10,701,134
13.9%
1,444,725
4.9%
5.12%
Dec 17-Dec 23
10,716,189
13.7%
1,440,770
5.1%
5.57%
Dec 24-Dec 30
9,082,257
13.9%
1,303,286
6.0%
5.95%
Dec 31-Jan 6
9,317,985
16.4%
1,365,473
7.3%
6.42%
The failure of test counts to recover here is maddening. We need to be adding more capacity, not taking multiple weeks off. The positive test percentages speak for themselves.
Covid Machine Learning Project
The code doesn’t know about Thanksgiving so it has a dip in infections in late November that clearly is not real, and now it is back to thinking R0 is over 1.1. Meanwhile, it had total infections on December 23 at 21.7%, versus a belief of 20.3% on December 16 last week, which it now thinks was 20.4% back then. Last week’s belief that things were steadily improving has been fixed by additional data to smooth out Thanksgiving.
Vaccinations
5.48 million doses so far, after almost a month. Pathetic. Scandalous. Insane. Shots. In. Arms. Now.
Europe
One of these lines is not like the others. The control system in the United Kingdom is failing.
The English Strain: Are We F***ed? Is it Over?
I found this comment very helpful, so I’ll quote it in full, it’s the best way I know of to estimate exactly how far along we are:
This is not a comforting picture. In my very quick and dirty first highly abstract model, with 50% more infectiousness in the new strain than the old, this would have us be 0.22% of the newstrain during the year’s first cycle, 2% by the end of January, 16% by the end of February, 68% by the end of March. In the simplest possible model that you should not in any way take seriously, the end of May would be the peak at roughly 225% of the infections per day we have now, July 1 we’d be back to where we are now, then a very rapid decline after that.
What evidence do we have about the infectiousness of the new strain?
Denmark sequences the new strain, as a reason not to worry, a play in three acts:
Asking “What evidencefiltered evidence?” in situations like the new strain is important. I do my best to include all the evidence that’s worth including, but the degree to which I look for it might vary, and I could plausibly be doing things like not mentioning places where there is an absence of evidence that constitutes at least some evidence of absence. Please do keep calling me out on such mistakes by pointing out what’s missing, with or without the principle of charity.
(I hadn’t posted the Denmark data last week because I hadn’t seen anything that led me to update substantially in either direction, sample sizes and magnitudes seemed too low.)
So that looks… exactly like what you would see with a 65% more infectious strain, which doubles about once per week, while other evidence points me closer to about 50% more infectious than 65%. I’m essentially assuming at this point that the new strain is at least substantially (35%+) more infectious.
I don’t know what the likelihood ratio on this was, but it seems pretty damn high. And that’s the one country I had heard was doing a bunch of sequencing other than the United Kingdom, on one of only two strains we are worried about. So it seems not very selected. But I could easily be missing awareness of others, if so please do share.
Did I mention most European schools are still open? Well, here’s some news:
If he’s right about the current prevalence being ~1%, that’s not comforting. Quite the opposite. That’s two weeks ahead of my previous estimated schedule, and the one I used for the spreadsheet modeling. Which gives us two less weeks to vaccinate, and for temperatures to improve, before we get hit hard. The good news is that the naive model doesn’t think this does that much additional damage, mostly moving infections forward in time, although it does make the last peak a little bigger and the total infection count a little higher.
This is the kind of thing you see all the time from such sources, and such sources are capable of being a lot louder than this on such things even on very false alarms, so I don’t put too much weight on them yet.
The problem is, I asked my father, who taught immunology at Columbia University and had a business doing due diligence on pharmaceutical companies, to look into the situation, and what he told me was very much not encouraging.
Here is what he came back with:
“A variant isolate of SARS-CoV2 from South Africa is causing concern, because convalescent sera from some patients recovered from infection with the predominant strain, while effective in neutralizing the same strain, were not effective in neutralizing the South African variant. Sera from other such patients did neutralize the variant effectively.
The concern is whether the current mRNA vaccines (coding for the spike protein from the predominant strain) will be effective against the variant. Unfortunately, the authors did not test sera from vaccinated individuals against the variant, so we don’t know for sure. It may be widely effective, effective in some individuals but not in others, or widely ineffective in provoking synthesis of neutralizing antibody to the variant.
A wrinkle: The current vaccines may offer protection from serious illness even in individuals in whom it fails to provide neutralizing antibody, but unlikely to prevent infection.
One of the advantages of the mRNA vaccines is that their manufacture can be modified quickly to accommodate changes. One would hope that the new vaccines could be granted emergency approval quickly, since they would be very similar to the approved ones.”
In conversation, he put the probability that the new variant would interfere with the neutralizing agent of the antibodies generated by those who get the mRNA vaccines at 80%. This is something we can find out quickly through experimentation, and it’s a sign of how bad we are at running experiments that the news has not yet (to my knowledge) reached us. There’s also the chance that other forms of protection could help out, or that the neutralizing agent could still have substantial effects sufficient to prevent illness or reduce severity. But if it comes back that the worst case scenario is indeed the case, and this mutation allows mRNA-vaccinated people to become sick, and the strain is indeed otherwise more infectious than the old one, then we have a very big problem.
So the good news is that mRNA vaccines can quickly be modified to handle the new strain. And if we had a benevolent well-functioning government, we could recognize that it is the same vaccine still, and not require a new round of trials. I have very little hope that we would do this, and thus very little hope we would be able to make the necessary adjustments in a timely manner. I imagine the FDA and company holding up approvals of the new versions for several months, and then the exact same botched roll-out to happen again, and by then for it to be very much too late, even if we don’t face the mutation problem again.
The other good news is that it seems highly unlikely the new strain can do much reinfection of previously infected individuals, and that even if the trend of noticing more ‘reinfections’ is real, it isn’t going to move the bottom lines much. And also that it is possible we get actively lucky with all this, and the folks who are vaccinated can get infected but don’t get sick, which results in them having much more robust protection against future infection or more mutations than they got from the vaccine alone.
Anyone who knows more about this, please do chime in.
Now that we know we are fucked, and for the moment assuming for purposes of analysis that we’re not that fucked, the question then shifts to: Is it over?
This “be more scared” is both people’s natural message to others, and also people’s natural reaction for themselves, for overlapping but also different reasons. When shown evidence that it’s going to get worse later, people ramp up their internal “this is bad” meter, and act more scared now. I know of one non-rationalist person who was shown my post on the new strain, and decided on the spot to head home, and that they were going to start staying home now and not see friends anymore.
I keep pointing out that on a personal level, in terms of your personal incentives, this is wrong.
If things are going to get much worse later, including a possible hospital system collapse, the new strain makes protecting yourself now less valuable to you, not more valuable. It’s now going to cost you much more to stay safe the whole way, and action later is going to be more risky than action now rather than less risky. So if you have risks you need to take to stay sane and/or stay solvent, or things that need attention like other medical concerns, better to take care of that now rather than later.
If you are thinking on a society-wide level, it’s complicated. Slowing spread also slows B.1.1.1.7, giving us a little more time to vaccinate before we need to fall back on our control systems. We might not entirely waste that time. It also means that when the control systems start to notice the new strain (which will happen at the same time regardless, because as I understand the math the strain takes over equally quickly regardless) it will be from a lower baseline level, so depending on exactly how the control systems function, we could get substantially better outcomes when the crisis hits.
On the other hand, you’re burning your remaining ammunition in terms of getting people to listen, lock down and play it safe. You’re also doing so in time for control systems to undo a lot of your work before the crisis hits. And if the hospitals are going to be overwhelmed in the future, to a much worse degree than they are now, then extra infections now could plausibly make the peak less bad, and thus reduce both deaths and the amount of overshoot.
None of which are questions anyone involved is thinking about. It’s one knob. Danger rising, sound alarms for danger, get people more scared, hope they act safer.
What Should You Do Personally?
What should you actually do on a personal level? I have been trying to get a better handle on that. I have a post from yesterday with some toy modeling attempts. It’s not my strong suit and it’s not going to be right, but so far I’ve seen essentially nothing, so giving it a go should hopefully inspire something better to follow.
What is clear is that there is a big difference between 50% and 65% more infectious. If we are looking at only 50%, the new strain will have a big impact, but if we can do a reasonable job (relative to current expectations, mind you) with vaccinations I’m growing more hopeful the final stage may not be too bad, and a range of outcomes seem plausible. If it’s 65%, I don’t see a way out of at least a brief period with a lot more infections than now, and our hope for keeping deaths down and hospitals intact is based on getting enough of the vulnerable vaccinated by then.
Therefore, keep a close eye on which scenario we are looking at, and how things are playing out otherwise. It really does change what behaviors make sense.
If the cost of protecting yourself for long enough is high, that makes the value of staying safe now lower. Getting infected now is not good, the hospitals are already not in a great place, but things could get much worse.
I won’t go over in detail what ‘stay safe’ means, but it means all the things you’d think: Socially distance, wear masks, take Vitamin D, do things and meet people outdoors, check for symptoms, check air ventilation. I continue to think surfaces are a trivial worry beyond washing one’s hands, although if one is trying to be actively super paranoid (e.g. in a hospital collapse) precautionary principle is reasonable.
As a reminder, death rates from Covid-19 for young healthy people are quite low, but there are potential long-term consequences and medium-term consequences that we have large uncertainty about with regard to both frequency and magnitude (because experimentation and data gathering are illegal) and those consequences can be very not fun. Losing your sense of smell for months is much worse than it might sound. You do not want to get Covid-19 if you can avoid it.
Once again, these are the core things to know when forming your plan:
If we are in one of the very bad scenarios, getting the virus during the later peak will be very, very bad, and also rather likely for anyone not either immune or taking lots of precautions.
If you can work from home and otherwise take ‘reasonably careful’ precautions, you will probably not get Covid-19 even in those bad scenarios. Even in the worst case where 75% or more of the population ends up infected. But you need to prepare to take those precautions and sustain them for months, and prepare to pay that price.
It is worth securing your supplies in case the supply chain becomes an issue. The earlier you do that, the more measured you can be doing it, and the better it is for everyone.
Unless you are immune, at some point, whether or not you decide that point is already here, there is a high chance you will want to start taking extreme precautions. Be ready. If you have to choose, better to take your necessary risks now, rather than later.
I’ve said what I feel that I can. I will note that the incentives involved strongly restrict the things I can say here, whether or not I believe them – so you’ll need to work that out for yourselves.
As always, I encourage you to form your own model of the world, think for yourself, and do what you believe is right.
Two Dose, One Dose, Who Knows, You Knows
This week’s crisis is getting shots into arms before the shots spoil, but soon the limiting factor will (I hope) be not having shots to put into arms. Luckily, there are several good ways to help with that, the tricky part is getting those in charge to listen.
The British approved the Pfizer vaccine, they approved the AstraZeneca vaccine, they moved to first doses first and now they are allowing (not yet encouraging they are running a trial) mix and match. Under the present circumstances, the British focus on doing what it takes to save lives is smart, admirable, and impressive.
Mix and matching has two potentially good properties. First, mix and matching could make the immune system response stronger than either vaccine alone because different vaccines stimulate the immune system in different ways. Second, it could help with distribution. It’s going to be easier to scale up the AZ vaccine than the mRNA vaccines, so if we can use both widely we can get more bang for our shot.
(I included the last line because I wanted to note that this is 3,809 deaths per day and the average had not at the time been over 2,700 for any 7-day period so far, that that projection was made on December 31, so that seems like a rather bold prediction if it’s meant to reflect a physical reality rather than being a scare tactic. I can only conclude that it isn’t.)
I do think there is a legitimate worry with people’s faith in the vaccine process if you start mixing and matching, even though I expect mixing and matching to work at least as well as not doing so, and probably better. If you can pull it off without that issue, wonderful, definitely do that. Hell, at this point, anything that is clearly oriented to life saving based on modeling the physical world is something I would welcome even if I think it’s wrong on the merits (up to a point, if it’s sufficiently wrong then no that’s bad and please stop). It’s rough out there.
First doses first went from crazy two weeks ago, to the British are doing it a week ago, to now first doses first plus mix and matching.
Quite right. Changing your mind and publicly admitting you are wrong, opening yourself up to blame from all sides, is hard, and should be applauded, and also there should totally be lots of screwups and reporting errors if we are moving fast enough. Good calls all around.
This is the most recent Marginal Revolution argument for first doses first. A key insight worth repeating is that there have been zero known expected value calculations in support of not doing second doses. There’s a lot of haphazard arguments and fear, uncertainty and doubt, but nothing I find remotely compelling, or that comes with any plausibly compelling numbers attached.
I always find it strange that people say things like “we don’t know whether a single dose vaccine provides protection from severe Covid” in spots like this. I don’t see a plausible physical way for it not to.
My conclusion (at least for now) is that the half dose is probably all you need if you are not elderly and get the second shot afterwards, with close to full effectiveness, and is more efficient per vaccine dose if given as only one dose but less effective than one full dose. There is some danger that the dropoff in effectiveness is bigger than we think, but not so big that we could plausibly come out behind on the deal. If one shot is already 80% effective, I find it highly implausible we wouldn’t do better than that with the double half-dose, given the data here.
That all works for me. Yes, we could do so much better if everyone did all the right things and grabbed all the multiplicative low-hanging fruit, but even grabbing some of the fruit, even different fruit, is a huge win. I’ll take it.
Yes, We Can Agree Andrew Cuomo Is The Worst
That’s less true now than when I wrote this, “because of reasons,” but still. Somehow he has topped himself. I’m not saying History’s Greatest Villain, or even clearly this week’s greatest villain – the competition’s been stiff lately – but it is not through lack of trying.
On January 4, doctors outside of hospitals became eligible. On January 11, home care workers will become eligible. At some unannounced point after that, maybe we’ll get to phase 1B, which includes those over 75 years old. One hopes.
But, Cuomo says, I also threatened our health care providers to speed things up, doesn’t that fix everything?
Did you know that the more you yell and threaten, the more everyone does what you want?
Second-worst person New York Mayor DeBlasio is not happy about the threats, calling them ‘just arrogance’ because he can only think about status implications rather than physical consequences.
Oh, it’s so much worse than that. What happens when the central planner combines threats to those who don’t distribute all the vaccine doses they get, with other threats to those who let someone ‘jump the line’? Care to solve for the equilibrium?
Things we will totally do in the future, to avoid the risk that someone somewhere is using our existing private infrastructure, but we’ll worry about the logistics for this when the time comes:
Note that Cuomo is not using the present tense. He is not saying “We are recruiting” such people, or that we “are establishing” such locations.
Avoiding blame in the next two weeks, you see, does not extend to problems we will have three weeks from now. Worry about tomorrow, tomorrow.
Even if you pull it off and thread the needle, it does not seem that ‘use 99% of your allocation while we have more doses than we can use despite obeying all the restrictions’ leads to the next step of then maybe you should send those places some more:
From the same thread, I’m trying to figure out the logic behind this one if it isn’t purely sacrificial, and it’s a thinker:
You have to be sympathetic, though. These logistics are hard. It’s not like there was any warning that such problems were coming. Yes, we did know vaccine shipments were likely, but it’s not like there was any time or way to plan for mass vaccinations in advance.
So let me get this straight. New York State has a detailed mass vaccination plan, developed well in advance for exactly this situation. It’s ready to go. There are distribution sites, officials in charge in each locality, every county is standing by. The plans are submitted. And we’re just going to… not use it, because we’d rather be sure that the doses all get distributed in exactly the right order.
Even if you think there’s rhyme or reason to the decisions being made, they’re not only not transparent, they’re not being told to anyone.
No one has any idea what is going on, who has priority over who or when anyone can get vaccinated, or how and where and when one might go about doing that if you aren’t living or working at a location doing the vaccinations. No one can plan.
This is all hitting on a personal level, on multiple fronts.
As a concrete example, my wife is a psychiatrist. On Monday, she was informed (with no warning) via various roundabout methods, that she was eligible for vaccination.
The announcements said nothing about how to go about getting vaccinated, by appointment or otherwise. So we thought about what to do and she started making phone calls. A lot of phone calls. Officials directed her to other officials. Officials said they learned about the new eligibility that day the same way we had, and they were scrambling and would have to call us back. No one could provide any clarity.
After all the phone calls, she’s on one list where they will maybe call her back and offer her an appointment somewhere at some point with some amount of notice, none of which can be specified, and then she’ll need to drop everything and take that opportunity, since there might not be another. Then there are the other potential locations, which haven’t even gotten that far, and pretty much everyone has a limited supply available and also no idea how to use it.
So despite being eligible for vaccination, and despite there being lots of unused doses, and despite our willingness to get into a car and drive quite far and make a lot of phone calls, still no vaccination and no timeline for getting one.
Meanwhile, I talked to my father, who is both very old and has comorbidities. He is therefore in “Group 1b” which includes 2.5 million people in the city of New York, behind a huge pool of people in 1a who are mostly not at high risk and that are only half done, and to which groups are being added haphazardly – which makes sense, since Cuomo won’t let anyone from 1b get vaccinated, so it’s vital to expand 1a as much as possible so vaccine doesn’t spoil. When we go to group 1b, he fears slash assumes it’s going to get prioritized to go to ‘harder hit’ zip codes and areas, and that he’ll effectively be behind a quarter of the city and forced to wait, and that nothing he can do will change that. I don’t expect this and told him he was quite wrong, that the system is more than sufficiently a giant mess that being willing to call around and go where you need to go will be quite sufficient to be early in 1b.
But at this point he expects allocation by Politics and Power to do its best to kill him.
Contrast this with the system effectively being used in Florida, featuring such horrible blameworthy actions as finding ways to do things, and to vaccinate seniors who actively want to get vaccinated:
So it seems like we’re debating whether prisoners should be in the same tier, with effectively higher priority, than my very old, very high risk father who has effectively been under house arrest since March, and who would happily give up quite a lot to get it. Are they more deserving? Is he not at much higher risk? What the hell is this?
Except we’re denying we’re debating that because one could be blamed for either decision, and the need to decide is more than two weeks in the future, so we’re not going to decide (or at least not tell people the decision) until the last possible moment.
Do you have any idea how enraging this is?
And of course I’m way behind all those people and who knows when I’ll get a chance.
To his credit, thanks to his motivation to shift blame to the Federal Government, Cuomo is at least yelling at those responsible for getting us more supplies to get us more supplies, and to approve more sources of supply, all of which are fine things to call for, but also cheap talk:
And DeBlasio does have that minor issue that the city that never sleeps does almost no weekend vaccinations, making excuses seem rather weak
Not Cuomo, but to cover bases mentioned above, here’s the quote from Newsome, who is trying but can’t hope to compete with the champion:
In other New York news, I’d note that policies like this might have made sense back when New York’s infection and death rates were unusually low, but make very little sense now:
To be fair, it does seem Washington, DC managed to get on that page, at least once, to go with the previous incident in Kentucky. This taking place in the city that isn’t part of a state seems like it might not be a coincidence, although it is also where there is a lot of media.
What happened was, they lost their freezer, so they had 830 doses and two hours to distribute them. They made the (obviously overdetermined and correct) decision to vaccinate as many as possible regardless of guidelines.
If the doses were spoiling, could they be ‘returned’ to the county or not? If they could, why didn’t they return all the doses? If they couldn’t, how did they return 200? Was there exactly that much space in their freezer? What storage facilities were available in Ukiah?
They also didn’t skip over the issue of proper consent or a ‘safe protocol’, and prioritized on that basis:
The fire department figured out how to handle a true emergency, as one might expect:
In the end, success!
Of course, then the conspiracy theories started flowing, because the freezer breaking let those with local power direct doses to wherever they want. I’m confident that didn’t happen, but I understand the instinct to think that way.
Scott Gotleib, former head of the FDA, making multiple modest proposals, not only suggesting pharmacies administer vaccinations, but also that we administer more vaccinations:
It’s deeply sad that we need to use the language of ‘high-risk Americans have access’ to suggest using our actual existing distribution system for the purpose of distributing the things they normally distribute.
Yes, if the first shot doesn’t do anywhere near what we think it does, and we face a sudden and total supply collapse, then we are slightly more screwed with first doses first.
Options that exist in theory but somehow not in practice, an ongoing mystery:
Look, I totally get that nursing home residents need to get medical histories and such before you can clear them for the vaccine, but the health care workers are literally at exactly the places you give the shots and the nursing home residents are kind of there because they can’t ever be anywhere else, so who exactly are the unchallenging groups? And if it’s too difficult can we then open up the criteria a bit? No?
And yeah, how about you start scrambling and thinking about that wall that you’re going to hit BEFORE you run into the wall?
Given, Dr. Jha, that you outright said you’re going to hit that wall.
This is not mainly about Bill Gates being better than the government at logistics. This is because if the vaccine was being given with bad intent, then there would be no danger of motive ambiguity, and those with power could safely make better choices rather than worse choices.
Texas does reasonable thing, and opens up Phase 1B which includes everyone over 65, which should be sufficient to find arms in which to put shots as needed:
In other words, ‘data and science’ that do not exactly conform to standards don’t exist, so until you file the proper forms representing the properly numbered trials, in triplicate, we are not going to schedule a meeting for a few weeks later to consider your proposal to prevent more deaths rather than prevent less deaths via the use of these mysterious things you call “Bayesian evidence” and “logic.”
Any defense beyond that and the response to it are going to fall under “Something Is Wrong On the Internet” so feel free to skip this section, but it seems like it should be written anyway.
This seems like the line of defense for those who aren’t going to fall back on the pure ‘it is forbidden’ and bury their heads in the sand like ostriches until someone does an RCT. In advance I admit I may be unfair here, but this seemed like a relatively thoughtful version to me, while being illustrative of the fallacies involved:
She then links to this study and this other study, which are models that conclude that vaccinating old people before young people results in more infections and less deaths, whereas vaccinating young before old does the opposite. The second, to its credit, points out that there are other considerations than a pure count of deaths, and it’s not obvious the optimization target of public policy should be purely deaths prevented – as one comment pointed out, there’s a reasonable case that QALYs saved is a better health target even if you ignore economics.
Which all makes sense, and I do buy that elderly vaccinations first minimize deaths, but only if you assume that both scenarios vaccinate the same number of people.
So it doesn’t address the question in a useful way. She goes on this way:
The third tweet illustrates the core issue. She is thinking of the alternative as ‘sits on the shelf for a week’ but that all the vaccinations still happen a week later.
If we all had confidence in that, there would (I presume) not be this chorus of panic and outrage. The problem is that we are literally worried the vaccines will expire and be thrown out, and the whole timetable will be thrown off, and it will be a huge disaster on every level. Which I’m presuming she would agree, if it were to happen, would be a huge disaster on every level.
And also that, by saving the extra dose, we would give a nursing home resident a dose when they would otherwise not have gotten one. Which doesn’t even parse for me at this point. When it comes to nursing homes, supply is not even an issue. When you are administering 20% of your doses after several weeks, returning the excess supply does actual nothing. The bottlenecks lie elsewhere.
So while I’m very confident that she’s right that it’s worth waiting a week if it transfers the dose to a nursing home resident that would otherwise have to wait a long time (e.g. a month or more), I don’t see how that has any bearing on our current reality.
The whole thread also illustrates the general assumption, by those who think we can do central planning, that we should treat the ‘baseline scenario’ as being able to allocate the way the planner wants, with no waste, no delays, no additional costs, no regulatory capture, no corruption, no allocation by politics and power that does something different from their policy paper, no perverse incentives causing strange destructive behaviors, and so on, when the actual ‘baseline scenario’ is to assume you get all of that.
No, you in practice mostly can’t usefully return excess doses, and that’s not going to change. There are lots of logistical and storage issues. And more to the point, once you allocate doses to an institution, what the hell did you think was going to happen? That they were going to give it back to be reallocated no one knows when to nowhere they’re associated and nowhere they’d get any credit, denying it to their employees and allies, rather than allocate the doses via politics and power to their employees, allies and associates?
That they’d spend their own resources to find a systematic way to quickly find people who fit into remarkably strict categories, to whom to give, at their own time and expense, completely free shots, taking on all the liability involved, including threats from the government for criminal prosecution if such people turned out not to fit the criteria?
And you also want, in that last tweet, to put (at least some of) the burden on these random institutions to then allocate the vaccine based on who has suffered disproportionately? While also obeying all the official restrictions or else, and also when did that become a priority?
If we lowered the thresholds at least to something reasonably common and easy to verify, such as being Over 65, we could at least give them a chance. Right now, I really, really don’t know what you were expecting.
It seems that nurses’ unions are taking stands to prevent the shots from being mandatory. Which takes away the whole ‘people can safely use the medical system’ advantage of doing health care workers first.
Thus, they can take actions that are motivated by avoiding blame for a lack of vaccinations within the next two weeks or so. They can take actions to avoid blame for vaccines actively going into trash cans. What they cannot do are the things one would do if one wanted to maximize the number of needles that get put into arms.
Thus, restrictions on who can take appointments that are so harsh that a majority of appointment times are unclaimed, rather than open up to the next tier to ensure enough demand, or allow lower tiers to make appointments on shorter notice – e.g. if the simple solution is unthinkable, how about only those in Tier 1A (whatever that locally is) can book two days in advance, then it’s opened to Tier 1B the day before, and Tier 1C same day, and you can be put on their waiting list? Or a line outside the building (at whatever tier of eligibility), and if there’s an empty slot, first come first serve?
Because I’ve certainly learned my lesson and I’m not at all bitter about it or saying I Told You So over and over again in great detail:
They were planning to do all residents and some staff, then lost doses, so they did some residents and some staff. Sounds like some staff (and perhaps some residents) had local political power to steer allocation, and others did not.
And of course, refer to much of the section on Andew Cuomo, as I’ve moved the rest of the New York examples over there.
From a tweet on the source thread that would not like to get broadcast too widely, not verified at all:
I don’t expect this to turn out to be a big deal, but I could easily be wrong and am choosing not to do the research, so if the evidence indicates otherwise, please speak up and explain.
Off-topic but awesome: Money Stuff is back! Matt Levine returns from parental leave with the best daily newsletter I have ever known, and it is not remotely close. This is about the latest happenings of all things money, hence the name, but written in such a way that it appeals both to professional traders and to people who don’t otherwise know or care about money stuff. I’ve been known to read the column out loud to my wife, without having read it first, and it’s reliably both funny and enlightening. The perspective you get here is not the whole picture, but your picture of the world without it will be incomplete. You can fix that. Strong endorsement.
AstraZeneca has vowed not to make a profit on their vaccine ‘until the pandemic is over.’ They have also declared the right to declare the pandemic ‘over’ in July 2021. So due to the fear that someone, somewhere, might be earning a profit, a major corporation makes money if they sell doses after July 2021, but not if they sell them before July 2021. I wonder what they will do. (Original Hat Tip)
Every few weeks, the evidence for immunity for those who have been infected gets one week longer, and by Lindy that means two more weeks of immunity, yet I still get people claiming “we don’t know how long immunity lasts” or saying it fades after a few months, and so on. This leads to insanity like this, where in order to satisfy such folks, a star NBA player is missing games under quarantine for exposure despite having not only already had Covid-19, but having been confirmed to have antibodies:
I don’t blame the NBA either. People respond to incentives. The NBA is very well known to not be an exception to this. Trust the process.
You know who has a lot of supply coming soon? India:
If you live in the greater Washington D.C. area, there was a super-spreader event on January 6, so stay safe. Everyone else, you too.
I’ll see everyone next week. Hopefully with better news.
Covid 1/7: The Fire of a Thousand Suns
Link post
I’ll summarize everything up front.
Might as well get on with it.
Let’s do the numbers.
The Numbers
Predictions
Prediction last week: 14.3% positive rate on 9.7 million tests, and an average of 2,500 deaths, again with wide error bars.
Results: 16.4% positive rate on 9.3 million tests, and an average of 2,657 deaths.
The phase shift on 12⁄30, in the wake of Christmas, seems to have been real, giving us the clear holiday bump we did not see from previous holidays.
That was both the very bad outcome for infections I was worried about, and also not high on my list of things to be concerned or furious about this week.
For deaths, my estimate was lower than it should have been and I should have assumed a full reversion, so on reflection it’s my mistake rather than especially bad news.
The new strain is not yet prevalent enough to be noticeably impacting the numbers.
Prediction: 17.0% positive rate on 9.5 million tests, and an average of 2,800 deaths. The holidays are over, there will be some fallout, with things getting slightly worse, but with the main boost in deaths from Christmas mostly coming later.
Deaths
This is the one place it’s not as bad as it looks. The data source for these numbers is Wikipedia, which shows a relatively large amount of shifting of deaths from last week into this week.
If we assume that a lot of this week’s deaths were actually last week, that explains much of the increase. It’s not good news or anything, it’s definitely bad news, but it is not full on terrifying like it would be if we didn’t know about Christmas.
We still should expect further increases in the next few weeks before the tide likely temporarily turns.
Positive Test Percentages
No way to sugar coat this one. Everything is pointed in the wrong direction in a clear phase shift. The good news is that it could be a one-time move from Christmas, and thus we can still plausibly expect things to remain stable from this point forward without waiting for further control systems to kick in. I hope to start seeing slow improvements soon, but I must admit that the overall graph is not as encouraging of that as I thought a few weeks ago, especially outside of the Midwest, and the turning point may be farther off than we’d like. The Midwest is still probably past its peak for this wave, even if everyone else has to wait a bit.
Positive Tests
Test Counts
The failure of test counts to recover here is maddening. We need to be adding more capacity, not taking multiple weeks off. The positive test percentages speak for themselves.
Covid Machine Learning Project
The code doesn’t know about Thanksgiving so it has a dip in infections in late November that clearly is not real, and now it is back to thinking R0 is over 1.1. Meanwhile, it had total infections on December 23 at 21.7%, versus a belief of 20.3% on December 16 last week, which it now thinks was 20.4% back then. Last week’s belief that things were steadily improving has been fixed by additional data to smooth out Thanksgiving.
Vaccinations
5.48 million doses so far, after almost a month. Pathetic. Scandalous. Insane. Shots. In. Arms. Now.
Europe
One of these lines is not like the others. The control system in the United Kingdom is failing.
The English Strain: Are We F***ed? Is it Over?
I found this comment very helpful, so I’ll quote it in full, it’s the best way I know of to estimate exactly how far along we are:
This is not a comforting picture. In my very quick and dirty first highly abstract model, with 50% more infectiousness in the new strain than the old, this would have us be 0.22% of the newstrain during the year’s first cycle, 2% by the end of January, 16% by the end of February, 68% by the end of March. In the simplest possible model that you should not in any way take seriously, the end of May would be the peak at roughly 225% of the infections per day we have now, July 1 we’d be back to where we are now, then a very rapid decline after that.
What evidence do we have about the infectiousness of the new strain?
Denmark sequences the new strain, as a reason not to worry, a play in three acts:
Act One (posted 12⁄31):
Asking “What evidence filtered evidence?” in situations like the new strain is important. I do my best to include all the evidence that’s worth including, but the degree to which I look for it might vary, and I could plausibly be doing things like not mentioning places where there is an absence of evidence that constitutes at least some evidence of absence. Please do keep calling me out on such mistakes by pointing out what’s missing, with or without the principle of charity.
Act Two (posted 12⁄31 in reply):
(I hadn’t posted the Denmark data last week because I hadn’t seen anything that led me to update substantially in either direction, sample sizes and magnitudes seemed too low.)
Act Three:
So that looks… exactly like what you would see with a 65% more infectious strain, which doubles about once per week, while other evidence points me closer to about 50% more infectious than 65%. I’m essentially assuming at this point that the new strain is at least substantially (35%+) more infectious.
I don’t know what the likelihood ratio on this was, but it seems pretty damn high. And that’s the one country I had heard was doing a bunch of sequencing other than the United Kingdom, on one of only two strains we are worried about. So it seems not very selected. But I could easily be missing awareness of others, if so please do share.
Did I mention most European schools are still open? Well, here’s some news:
The San Diego case of the new strain gets contact traced, the results are in and 32 more cases have been found:
If he’s right about the current prevalence being ~1%, that’s not comforting. Quite the opposite. That’s two weeks ahead of my previous estimated schedule, and the one I used for the spreadsheet modeling. Which gives us two less weeks to vaccinate, and for temperatures to improve, before we get hit hard. The good news is that the naive model doesn’t think this does that much additional damage, mostly moving infections forward in time, although it does make the last peak a little bigger and the total infection count a little higher.
In other evidence (source 1, source 2):
At this point, if this were outside the range of 40%-75% more infectious, I would be very surprised.
The South African Strain: Are We Even More Fucked? Is It Even More Over?
Don’t look now, actually do look, we might actually truly be Malcom Reynolds “It’s worse than you think” level fucked by the South African strain, in which case it might be really, really over:
Here’s another source that seems to be further along in its analysis, linked paper here:
This is the kind of thing you see all the time from such sources, and such sources are capable of being a lot louder than this on such things even on very false alarms, so I don’t put too much weight on them yet.
The problem is, I asked my father, who taught immunology at Columbia University and had a business doing due diligence on pharmaceutical companies, to look into the situation, and what he told me was very much not encouraging.
Here is what he came back with:
In conversation, he put the probability that the new variant would interfere with the neutralizing agent of the antibodies generated by those who get the mRNA vaccines at 80%. This is something we can find out quickly through experimentation, and it’s a sign of how bad we are at running experiments that the news has not yet (to my knowledge) reached us. There’s also the chance that other forms of protection could help out, or that the neutralizing agent could still have substantial effects sufficient to prevent illness or reduce severity. But if it comes back that the worst case scenario is indeed the case, and this mutation allows mRNA-vaccinated people to become sick, and the strain is indeed otherwise more infectious than the old one, then we have a very big problem.
So the good news is that mRNA vaccines can quickly be modified to handle the new strain. And if we had a benevolent well-functioning government, we could recognize that it is the same vaccine still, and not require a new round of trials. I have very little hope that we would do this, and thus very little hope we would be able to make the necessary adjustments in a timely manner. I imagine the FDA and company holding up approvals of the new versions for several months, and then the exact same botched roll-out to happen again, and by then for it to be very much too late, even if we don’t face the mutation problem again.
The other good news is that it seems highly unlikely the new strain can do much reinfection of previously infected individuals, and that even if the trend of noticing more ‘reinfections’ is real, it isn’t going to move the bottom lines much. And also that it is possible we get actively lucky with all this, and the folks who are vaccinated can get infected but don’t get sick, which results in them having much more robust protection against future infection or more mutations than they got from the vaccine alone.
Anyone who knows more about this, please do chime in.
Now that we know we are fucked, and for the moment assuming for purposes of analysis that we’re not that fucked, the question then shifts to: Is it over?
But What Do We Do Now?
The same thing we do every night, Pinky. Tell everyone to Be Afraid. Be Very Afraid.
(That first part about vaccinations (MR) is completely right, of course.)
This “be more scared” is both people’s natural message to others, and also people’s natural reaction for themselves, for overlapping but also different reasons. When shown evidence that it’s going to get worse later, people ramp up their internal “this is bad” meter, and act more scared now. I know of one non-rationalist person who was shown my post on the new strain, and decided on the spot to head home, and that they were going to start staying home now and not see friends anymore.
I keep pointing out that on a personal level, in terms of your personal incentives, this is wrong.
If things are going to get much worse later, including a possible hospital system collapse, the new strain makes protecting yourself now less valuable to you, not more valuable. It’s now going to cost you much more to stay safe the whole way, and action later is going to be more risky than action now rather than less risky. So if you have risks you need to take to stay sane and/or stay solvent, or things that need attention like other medical concerns, better to take care of that now rather than later.
If you are thinking on a society-wide level, it’s complicated. Slowing spread also slows B.1.1.1.7, giving us a little more time to vaccinate before we need to fall back on our control systems. We might not entirely waste that time. It also means that when the control systems start to notice the new strain (which will happen at the same time regardless, because as I understand the math the strain takes over equally quickly regardless) it will be from a lower baseline level, so depending on exactly how the control systems function, we could get substantially better outcomes when the crisis hits.
On the other hand, you’re burning your remaining ammunition in terms of getting people to listen, lock down and play it safe. You’re also doing so in time for control systems to undo a lot of your work before the crisis hits. And if the hospitals are going to be overwhelmed in the future, to a much worse degree than they are now, then extra infections now could plausibly make the peak less bad, and thus reduce both deaths and the amount of overshoot.
None of which are questions anyone involved is thinking about. It’s one knob. Danger rising, sound alarms for danger, get people more scared, hope they act safer.
What Should You Do Personally?
What should you actually do on a personal level? I have been trying to get a better handle on that. I have a post from yesterday with some toy modeling attempts. It’s not my strong suit and it’s not going to be right, but so far I’ve seen essentially nothing, so giving it a go should hopefully inspire something better to follow.
What is clear is that there is a big difference between 50% and 65% more infectious. If we are looking at only 50%, the new strain will have a big impact, but if we can do a reasonable job (relative to current expectations, mind you) with vaccinations I’m growing more hopeful the final stage may not be too bad, and a range of outcomes seem plausible. If it’s 65%, I don’t see a way out of at least a brief period with a lot more infections than now, and our hope for keeping deaths down and hospitals intact is based on getting enough of the vulnerable vaccinated by then.
Therefore, keep a close eye on which scenario we are looking at, and how things are playing out otherwise. It really does change what behaviors make sense.
If the cost of protecting yourself for long enough is high, that makes the value of staying safe now lower. Getting infected now is not good, the hospitals are already not in a great place, but things could get much worse.
I won’t go over in detail what ‘stay safe’ means, but it means all the things you’d think: Socially distance, wear masks, take Vitamin D, do things and meet people outdoors, check for symptoms, check air ventilation. I continue to think surfaces are a trivial worry beyond washing one’s hands, although if one is trying to be actively super paranoid (e.g. in a hospital collapse) precautionary principle is reasonable.
As a reminder, death rates from Covid-19 for young healthy people are quite low, but there are potential long-term consequences and medium-term consequences that we have large uncertainty about with regard to both frequency and magnitude (because experimentation and data gathering are illegal) and those consequences can be very not fun. Losing your sense of smell for months is much worse than it might sound. You do not want to get Covid-19 if you can avoid it.
Once again, these are the core things to know when forming your plan:
If we are in one of the very bad scenarios, getting the virus during the later peak will be very, very bad, and also rather likely for anyone not either immune or taking lots of precautions.
If you can work from home and otherwise take ‘reasonably careful’ precautions, you will probably not get Covid-19 even in those bad scenarios. Even in the worst case where 75% or more of the population ends up infected. But you need to prepare to take those precautions and sustain them for months, and prepare to pay that price.
It is worth securing your supplies in case the supply chain becomes an issue. The earlier you do that, the more measured you can be doing it, and the better it is for everyone.
Unless you are immune, at some point, whether or not you decide that point is already here, there is a high chance you will want to start taking extreme precautions. Be ready. If you have to choose, better to take your necessary risks now, rather than later.
I’ve said what I feel that I can. I will note that the incentives involved strongly restrict the things I can say here, whether or not I believe them – so you’ll need to work that out for yourselves.
As always, I encourage you to form your own model of the world, think for yourself, and do what you believe is right.
Two Dose, One Dose, Who Knows, You Knows
This week’s crisis is getting shots into arms before the shots spoil, but soon the limiting factor will (I hope) be not having shots to put into arms. Luckily, there are several good ways to help with that, the tricky part is getting those in charge to listen.
From Marginal Revolution:
The British approved the Pfizer vaccine, they approved the AstraZeneca vaccine, they moved to first doses first and now they are allowing (not yet encouraging they are running a trial) mix and match. Under the present circumstances, the British focus on doing what it takes to save lives is smart, admirable, and impressive.
As I wrote on Dec. 10, in Herd Immunity is Herd Immunity:
Mix and matching has two potentially good properties. First, mix and matching could make the immune system response stronger than either vaccine alone because different vaccines stimulate the immune system in different ways. Second, it could help with distribution. It’s going to be easier to scale up the AZ vaccine than the mRNA vaccines, so if we can use both widely we can get more bang for our shot.
Addendum: The CDC is projecting 80,000 COVID deaths in the United States over the next three weeks.
(I included the last line because I wanted to note that this is 3,809 deaths per day and the average had not at the time been over 2,700 for any 7-day period so far, that that projection was made on December 31, so that seems like a rather bold prediction if it’s meant to reflect a physical reality rather than being a scare tactic. I can only conclude that it isn’t.)
I do think there is a legitimate worry with people’s faith in the vaccine process if you start mixing and matching, even though I expect mixing and matching to work at least as well as not doing so, and probably better. If you can pull it off without that issue, wonderful, definitely do that. Hell, at this point, anything that is clearly oriented to life saving based on modeling the physical world is something I would welcome even if I think it’s wrong on the merits (up to a point, if it’s sufficiently wrong then no that’s bad and please stop). It’s rough out there.
First doses first went from crazy two weeks ago, to the British are doing it a week ago, to now first doses first plus mix and matching.
It gives me hope, you love to see it.
I’d also like to point out that the British did a great but hard thing, closing schools exactly one day after opening them even though doing that made them look superficially like the complete and total idiots they are:
Quite right. Changing your mind and publicly admitting you are wrong, opening yourself up to blame from all sides, is hard, and should be applauded, and also there should totally be lots of screwups and reporting errors if we are moving fast enough. Good calls all around.
This is the most recent Marginal Revolution argument for first doses first. A key insight worth repeating is that there have been zero known expected value calculations in support of not doing second doses. There’s a lot of haphazard arguments and fear, uncertainty and doubt, but nothing I find remotely compelling, or that comes with any plausibly compelling numbers attached.
Allow us to present a very clean intuition pump:
Even Operation Warp Speed is feeling enough pressure to get creative. Since there are huge low hanging gains everywhere, there’s room for different countries to find different fruit. For example, did you know that the studies seem to show half doses work as well as full doses for the non-elderly?
So, crazy idea, maybe give them half doses?
This thread breaks down the data available. Key findings:
Here’s another thread, I’ll focus on takeaways.
And then there’s this, which suggests that half doses might already be way more than we need although I haven’t examined the data involved:
I always find it strange that people say things like “we don’t know whether a single dose vaccine provides protection from severe Covid” in spots like this. I don’t see a plausible physical way for it not to.
My conclusion (at least for now) is that the half dose is probably all you need if you are not elderly and get the second shot afterwards, with close to full effectiveness, and is more efficient per vaccine dose if given as only one dose but less effective than one full dose. There is some danger that the dropoff in effectiveness is bigger than we think, but not so big that we could plausibly come out behind on the deal. If one shot is already 80% effective, I find it highly implausible we wouldn’t do better than that with the double half-dose, given the data here.
Meanwhile, on the matter of first doses first, I’d offer to eat my hat as well, except that’s a one sided bet and also I don’t own a hat:
And everyone’s entitled to their opinion:
That all works for me. Yes, we could do so much better if everyone did all the right things and grabbed all the multiplicative low-hanging fruit, but even grabbing some of the fruit, even different fruit, is a huge win. I’ll take it.
Yes, We Can Agree Andrew Cuomo Is The Worst
That’s less true now than when I wrote this, “because of reasons,” but still. Somehow he has topped himself. I’m not saying History’s Greatest Villain, or even clearly this week’s greatest villain – the competition’s been stiff lately – but it is not through lack of trying.
On January 4, doctors outside of hospitals became eligible. On January 11, home care workers will become eligible. At some unannounced point after that, maybe we’ll get to phase 1B, which includes those over 75 years old. One hopes.
But, Cuomo says, I also threatened our health care providers to speed things up, doesn’t that fix everything?
Did you know that the more you yell and threaten, the more everyone does what you want?
Second-worst person New York Mayor DeBlasio is not happy about the threats, calling them ‘just arrogance’ because he can only think about status implications rather than physical consequences.
Oh, it’s so much worse than that. What happens when the central planner combines threats to those who don’t distribute all the vaccine doses they get, with other threats to those who let someone ‘jump the line’? Care to solve for the equilibrium?
Here’s a new framing to help you out:
Eliezer got this one wrong, because actually there is a particular nothing that will not get ruled out.
Here’s the equilibrium:
Things we will totally do in the future, to avoid the risk that someone somewhere is using our existing private infrastructure, but we’ll worry about the logistics for this when the time comes:
Note that Cuomo is not using the present tense. He is not saying “We are recruiting” such people, or that we “are establishing” such locations.
Avoiding blame in the next two weeks, you see, does not extend to problems we will have three weeks from now. Worry about tomorrow, tomorrow.
Even if you pull it off and thread the needle, it does not seem that ‘use 99% of your allocation while we have more doses than we can use despite obeying all the restrictions’ leads to the next step of then maybe you should send those places some more:
From the same thread, I’m trying to figure out the logic behind this one if it isn’t purely sacrificial, and it’s a thinker:
You have to be sympathetic, though. These logistics are hard. It’s not like there was any warning that such problems were coming. Yes, we did know vaccine shipments were likely, but it’s not like there was any time or way to plan for mass vaccinations in advance.
Oh. Wait..(you might want to read the whole thing on this one)
What was that, Mr. Reynolds?
So let me get this straight. New York State has a detailed mass vaccination plan, developed well in advance for exactly this situation. It’s ready to go. There are distribution sites, officials in charge in each locality, every county is standing by. The plans are submitted. And we’re just going to… not use it, because we’d rather be sure that the doses all get distributed in exactly the right order.
Ross Barkan has similar thoughts about the whole mess, and offers some reminders about Cuomo’s past performance.
Even if you think there’s rhyme or reason to the decisions being made, they’re not only not transparent, they’re not being told to anyone.
No one has any idea what is going on, who has priority over who or when anyone can get vaccinated, or how and where and when one might go about doing that if you aren’t living or working at a location doing the vaccinations. No one can plan.
This is all hitting on a personal level, on multiple fronts.
As a concrete example, my wife is a psychiatrist. On Monday, she was informed (with no warning) via various roundabout methods, that she was eligible for vaccination.
The announcements said nothing about how to go about getting vaccinated, by appointment or otherwise. So we thought about what to do and she started making phone calls. A lot of phone calls. Officials directed her to other officials. Officials said they learned about the new eligibility that day the same way we had, and they were scrambling and would have to call us back. No one could provide any clarity.
After all the phone calls, she’s on one list where they will maybe call her back and offer her an appointment somewhere at some point with some amount of notice, none of which can be specified, and then she’ll need to drop everything and take that opportunity, since there might not be another. Then there are the other potential locations, which haven’t even gotten that far, and pretty much everyone has a limited supply available and also no idea how to use it.
So despite being eligible for vaccination, and despite there being lots of unused doses, and despite our willingness to get into a car and drive quite far and make a lot of phone calls, still no vaccination and no timeline for getting one.
Meanwhile, I talked to my father, who is both very old and has comorbidities. He is therefore in “Group 1b” which includes 2.5 million people in the city of New York, behind a huge pool of people in 1a who are mostly not at high risk and that are only half done, and to which groups are being added haphazardly – which makes sense, since Cuomo won’t let anyone from 1b get vaccinated, so it’s vital to expand 1a as much as possible so vaccine doesn’t spoil. When we go to group 1b, he fears slash assumes it’s going to get prioritized to go to ‘harder hit’ zip codes and areas, and that he’ll effectively be behind a quarter of the city and forced to wait, and that nothing he can do will change that. I don’t expect this and told him he was quite wrong, that the system is more than sufficiently a giant mess that being willing to call around and go where you need to go will be quite sufficient to be early in 1b.
But at this point he expects allocation by Politics and Power to do its best to kill him.
Contrast this with the system effectively being used in Florida, featuring such horrible blameworthy actions as finding ways to do things, and to vaccinate seniors who actively want to get vaccinated:
What about teachers? Good question. Who counts as an ‘education worker’ here?
What are we going to do about prisoners? Different people get different messages, so no one knows anything.
Then when confronted about having sent contradictory messages to different people, they denied everything and moved on to Cincinnati:
So it seems like we’re debating whether prisoners should be in the same tier, with effectively higher priority, than my very old, very high risk father who has effectively been under house arrest since March, and who would happily give up quite a lot to get it. Are they more deserving? Is he not at much higher risk? What the hell is this?
Except we’re denying we’re debating that because one could be blamed for either decision, and the need to decide is more than two weeks in the future, so we’re not going to decide (or at least not tell people the decision) until the last possible moment.
Do you have any idea how enraging this is?
And of course I’m way behind all those people and who knows when I’ll get a chance.
Even DeBlasio is declaring his intent to get our collective asses in gear, vowing one million vaccinations in the city in January. The problem, of course, is that he’s not allowed to do that yet, and who knows when he will be allowed (see: Andrew Cuomo is the Worst):
From NY Times via Nate Silver, so no link (Scott Alexander situation remains unresolved):
This could not be more explicit. We cannot vaccine people because there is no one we are allowed to vaccinate.
So of course Cuomo fires back:
Letting the days go by, letting the virus keep you home…
Cuomo’s not going to stand for this kind of undermining. Respect his authoritah:
To his credit, thanks to his motivation to shift blame to the Federal Government, Cuomo is at least yelling at those responsible for getting us more supplies to get us more supplies, and to approve more sources of supply, all of which are fine things to call for, but also cheap talk:
And DeBlasio does have that minor issue that the city that never sleeps does almost no weekend vaccinations, making excuses seem rather weak
Not Cuomo, but to cover bases mentioned above, here’s the quote from Newsome, who is trying but can’t hope to compete with the champion:
Perhaps he was mad about these horrible people?
I’ll leave this here:
Detailed story of the evacuation is here.
In other New York news, I’d note that policies like this might have made sense back when New York’s infection and death rates were unusually low, but make very little sense now:
Although I am glad to see this which seems worthwhile:
Also, did I mention the inevitable yet?
All I Want For Christmas is a Covid Vaccine, But They Somehow Underpriced Them So Much No One’s Even Bothering To Sell Out
Israel’s approach is personified by the Pizza Guy:
Ours, not so much.
To be fair, it does seem Washington, DC managed to get on that page, at least once, to go with the previous incident in Kentucky. This taking place in the city that isn’t part of a state seems like it might not be a coincidence, although it is also where there is a lot of media.
Then there’s the natural experiment we ran in California (source article here).
What happened was, they lost their freezer, so they had 830 doses and two hours to distribute them. They made the (obviously overdetermined and correct) decision to vaccinate as many as possible regardless of guidelines.
They still tried to prioritize:
Also, this just raises further questions:
If the doses were spoiling, could they be ‘returned’ to the county or not? If they could, why didn’t they return all the doses? If they couldn’t, how did they return 200? Was there exactly that much space in their freezer? What storage facilities were available in Ukiah?
They also didn’t skip over the issue of proper consent or a ‘safe protocol’, and prioritized on that basis:
The fire department figured out how to handle a true emergency, as one might expect:
In the end, success!
Of course, then the conspiracy theories started flowing, because the freezer breaking let those with local power direct doses to wherever they want. I’m confident that didn’t happen, but I understand the instinct to think that way.
Regardless, do read the whole thing.
The conclusion is clear as day. If we cared about distribution of the vaccine, the vaccine would be distributed. Period.
The chorus rises up, and speaks as one: Shots. In. Arms. Now.
The most obvious suggestion is to maybe use the pharmacies (WSJ) that already administer the flu shot, and let them do this one in the pharmacy as per normal (MSNBC) rather than only contracting for helping with long term care facilities? Still blows my mind this one is hard. Which, in good news seen later in the week, could be happening sooner rather than later, if nowhere near soon enough.
Scott Gotleib, former head of the FDA, making multiple modest proposals, not only suggesting pharmacies administer vaccinations, but also that we administer more vaccinations:
It’s deeply sad that we need to use the language of ‘high-risk Americans have access’ to suggest using our actual existing distribution system for the purpose of distributing the things they normally distribute.
Even if we do the second shot on schedule, surely it’s madness to hold back the doses for it rather than count on future supply, right?
Yes, if the first shot doesn’t do anywhere near what we think it does, and we face a sudden and total supply collapse, then we are slightly more screwed with first doses first.
Options that exist in theory but somehow not in practice, an ongoing mystery:
Look, I totally get that nursing home residents need to get medical histories and such before you can clear them for the vaccine, but the health care workers are literally at exactly the places you give the shots and the nursing home residents are kind of there because they can’t ever be anywhere else, so who exactly are the unchallenging groups? And if it’s too difficult can we then open up the criteria a bit? No?
And yeah, how about you start scrambling and thinking about that wall that you’re going to hit BEFORE you run into the wall?
Given, Dr. Jha, that you outright said you’re going to hit that wall.
Meanwhile, the wall we’re in now:
What having a chance of being sufficiently cynical looks like:
This is beyond fully endorsed:
This as well:
Take the following tweet far more seriously than you think:
This is not mainly about Bill Gates being better than the government at logistics. This is because if the vaccine was being given with bad intent, then there would be no danger of motive ambiguity, and those with power could safely make better choices rather than worse choices.
Also I have no idea why Mark Primiamo is even half-joking:
Here’s to the courage to admit when one is both correct and isn’t joking and also everybody knows this already:
Texas does reasonable thing, and opens up Phase 1B which includes everyone over 65, which should be sufficient to find arms in which to put shots as needed:
Meanwhile, in Israel, where they seem to care about vaccinating people…
Current general mood about the whole thing:
What have you to say in your defense, “ethicists”?
OK, that’s not fair. The American medical system is not Andrew Cuomo, but what would be fair?
We could… quote the FDA’s statement in full so there can be no misunderstanding?
In other words, ‘data and science’ that do not exactly conform to standards don’t exist, so until you file the proper forms representing the properly numbered trials, in triplicate, we are not going to schedule a meeting for a few weeks later to consider your proposal to prevent more deaths rather than prevent less deaths via the use of these mysterious things you call “Bayesian evidence” and “logic.”
See thread:
Points also awarded here:
And yes, I think that is all entirely fair.
FDA delenda est.
Any defense beyond that and the response to it are going to fall under “Something Is Wrong On the Internet” so feel free to skip this section, but it seems like it should be written anyway.
This seems like the line of defense for those who aren’t going to fall back on the pure ‘it is forbidden’ and bury their heads in the sand like ostriches until someone does an RCT. In advance I admit I may be unfair here, but this seemed like a relatively thoughtful version to me, while being illustrative of the fallacies involved:
She then links to this study and this other study, which are models that conclude that vaccinating old people before young people results in more infections and less deaths, whereas vaccinating young before old does the opposite. The second, to its credit, points out that there are other considerations than a pure count of deaths, and it’s not obvious the optimization target of public policy should be purely deaths prevented – as one comment pointed out, there’s a reasonable case that QALYs saved is a better health target even if you ignore economics.
Which all makes sense, and I do buy that elderly vaccinations first minimize deaths, but only if you assume that both scenarios vaccinate the same number of people.
So it doesn’t address the question in a useful way. She goes on this way:
The third tweet illustrates the core issue. She is thinking of the alternative as ‘sits on the shelf for a week’ but that all the vaccinations still happen a week later.
If we all had confidence in that, there would (I presume) not be this chorus of panic and outrage. The problem is that we are literally worried the vaccines will expire and be thrown out, and the whole timetable will be thrown off, and it will be a huge disaster on every level. Which I’m presuming she would agree, if it were to happen, would be a huge disaster on every level.
And also that, by saving the extra dose, we would give a nursing home resident a dose when they would otherwise not have gotten one. Which doesn’t even parse for me at this point. When it comes to nursing homes, supply is not even an issue. When you are administering 20% of your doses after several weeks, returning the excess supply does actual nothing. The bottlenecks lie elsewhere.
So while I’m very confident that she’s right that it’s worth waiting a week if it transfers the dose to a nursing home resident that would otherwise have to wait a long time (e.g. a month or more), I don’t see how that has any bearing on our current reality.
The whole thread also illustrates the general assumption, by those who think we can do central planning, that we should treat the ‘baseline scenario’ as being able to allocate the way the planner wants, with no waste, no delays, no additional costs, no regulatory capture, no corruption, no allocation by politics and power that does something different from their policy paper, no perverse incentives causing strange destructive behaviors, and so on, when the actual ‘baseline scenario’ is to assume you get all of that.
No, you in practice mostly can’t usefully return excess doses, and that’s not going to change. There are lots of logistical and storage issues. And more to the point, once you allocate doses to an institution, what the hell did you think was going to happen? That they were going to give it back to be reallocated no one knows when to nowhere they’re associated and nowhere they’d get any credit, denying it to their employees and allies, rather than allocate the doses via politics and power to their employees, allies and associates?
That they’d spend their own resources to find a systematic way to quickly find people who fit into remarkably strict categories, to whom to give, at their own time and expense, completely free shots, taking on all the liability involved, including threats from the government for criminal prosecution if such people turned out not to fit the criteria?
And you also want, in that last tweet, to put (at least some of) the burden on these random institutions to then allocate the vaccine based on who has suffered disproportionately? While also obeying all the official restrictions or else, and also when did that become a priority?
If we lowered the thresholds at least to something reasonably common and easy to verify, such as being Over 65, we could at least give them a chance. Right now, I really, really don’t know what you were expecting.
Vaccine Allocation By Politics and Power
I keep hearing that nurses are actually unusually likely to refuse the vaccine, with the usual explanation being that they know all the things that can go wrong in the medical system:
It seems that nurses’ unions are taking stands to prevent the shots from being mandatory. Which takes away the whole ‘people can safely use the medical system’ advantage of doing health care workers first.
Luckily, nursing homes, where the most is at stake, seem to have it less bad.
Nursing homes seem to have reassuringly low refusal rates. 90% acceptance is not so bad.
Fluent Cherokee speakers eligible for the vaccine. If we had only known this earlier, would a bunch more people be fluent in Cherokee by now?
That one time in DC when they managed to get expiring vaccines into random people’s arms was good. Any idea why the vaccine was about to expire?
Here’s a hint:
Here’s another hint:
I think that’s a bit harsh there by Shoddy. More precise would be they instinctively avoid being seen as taking actions which could plausibly be interpreted as motivated by getting people vaccinated. If they had a choice, they would totally want people to get vaccinated! But they sense a grave danger if the primacy of that motivation on their part were to become common knowledge.
Thus, they can take actions that are motivated by avoiding blame for a lack of vaccinations within the next two weeks or so. They can take actions to avoid blame for vaccines actively going into trash cans. What they cannot do are the things one would do if one wanted to maximize the number of needles that get put into arms.
Thus, restrictions on who can take appointments that are so harsh that a majority of appointment times are unclaimed, rather than open up to the next tier to ensure enough demand, or allow lower tiers to make appointments on shorter notice – e.g. if the simple solution is unthinkable, how about only those in Tier 1A (whatever that locally is) can book two days in advance, then it’s opened to Tier 1B the day before, and Tier 1C same day, and you can be put on their waiting list? Or a line outside the building (at whatever tier of eligibility), and if there’s an empty slot, first come first serve?
Because I’ve certainly learned my lesson and I’m not at all bitter about it or saying I Told You So over and over again in great detail:
Another problem is that if something goes wrong and some doses get spoiled, which is something that will happen, who knows when you can get more of them?
I found this interesting in particular:
They were planning to do all residents and some staff, then lost doses, so they did some residents and some staff. Sounds like some staff (and perhaps some residents) had local political power to steer allocation, and others did not.
And of course, refer to much of the section on Andew Cuomo, as I’ve moved the rest of the New York examples over there.
How Bad is it Out There Right Now?
Pretty bad (this is from Southern California):
That was a few days ago, now the headlines from LA are getting much worse. We have reached full triage mode:
Hopefully we are very near the peak of how bad things will be this wave, but that is unlikely to be true in at least some places.
A thread on what it’s like to be treated for Covid-19, first tweet below.
Paper on deaths of despair.
Even the paper itself is not safe:
A study out of the Ohio State University from September (I looked for it and couldn’t find a reference in past weeks, but I could have missed it) finds that 30% of student athletes who previously had Covid-19 had heart damage linked to Covid-19. The whole thing is surprising not only because it means that athletes at the Ohio State University had hearts to begin with (Roll Tide on Monday!), but also because it means that they did useful work at Ohio State, it potentially interfered with their athletic department and they still published. It all makes so little sense. My model of the world can at least take comfort in knowing that this in no way actually interfered with anyone’s athletic operations.
From a tweet on the source thread that would not like to get broadcast too widely, not verified at all:
I don’t expect this to turn out to be a big deal, but I could easily be wrong and am choosing not to do the research, so if the evidence indicates otherwise, please speak up and explain.
Hospitals are not collapsed as such, but things are very much not good:
So, I guess this might have happened, with 44 cases at the hospital:
In Other News
Off-topic but awesome: Money Stuff is back! Matt Levine returns from parental leave with the best daily newsletter I have ever known, and it is not remotely close. This is about the latest happenings of all things money, hence the name, but written in such a way that it appeals both to professional traders and to people who don’t otherwise know or care about money stuff. I’ve been known to read the column out loud to my wife, without having read it first, and it’s reliably both funny and enlightening. The perspective you get here is not the whole picture, but your picture of the world without it will be incomplete. You can fix that. Strong endorsement.
Bryan Caplan asks, have you tried… being happy this year?
Your periodic reminder that in 1947, New York City vaccinated ~6.35 million people (80% of their population) for smallpox in less than a month. If you do not think we can do this, what changed to make it impossible?
Your periodic reminder that the FDA’s incentives are bad.
What did China do to stop this virus and get back to normal? What didn’t it do? (Inessential and snarky Twitter thread)
British official report on AstraZeneca vaccine is out. As we expected, less good than Pfizer or Moderna’s in terms of effectiveness, far better than it needs to be to be worthwhile.
There’s an AstraZeneca manufacturing plant in Baltimore. It seems we’re really going to wait until April to approve their vaccine, the interpretation of which I leave to you the reader.
AstraZeneca has vowed not to make a profit on their vaccine ‘until the pandemic is over.’ They have also declared the right to declare the pandemic ‘over’ in July 2021. So due to the fear that someone, somewhere, might be earning a profit, a major corporation makes money if they sell doses after July 2021, but not if they sell them before July 2021. I wonder what they will do. (Original Hat Tip)
Every few weeks, the evidence for immunity for those who have been infected gets one week longer, and by Lindy that means two more weeks of immunity, yet I still get people claiming “we don’t know how long immunity lasts” or saying it fades after a few months, and so on. This leads to insanity like this, where in order to satisfy such folks, a star NBA player is missing games under quarantine for exposure despite having not only already had Covid-19, but having been confirmed to have antibodies:
I don’t blame the NBA either. People respond to incentives. The NBA is very well known to not be an exception to this. Trust the process.
You know who has a lot of supply coming soon? India:
If you live in the greater Washington D.C. area, there was a super-spreader event on January 6, so stay safe. Everyone else, you too.
I’ll see everyone next week. Hopefully with better news.