Robby Bensinger wrote this post and shared it in a private Facebook group, but it seemed good to also have it on LessWrong. Note that I have substantial disagreements with the current content of this document, since it currently seems to advise far too risk-averse of a strategy and makes claims of the type "if you don't fully lockdown you are basically giving up on not getting COVID" which strike me as very wrong. See thread on this comment for details.
Disclaimer: This document was made by non-experts. You may want to spot-check the cited sources to decide for yourself whether you think the reasoning makes sense. This document was created January 6, 2021, and may be out of date on some points.
Q: What's up with the new COVID-19 strain from southern England?
The strain, called VOC-202012/01 (or B.1.1.7 in cov-lineages.org nomenclature, 20B/501Y.V1 in nextstrain.org nomenclature), appears to be much more infectious than other COVID-19 strains. As of Dec. 31, Zvi Mowshowitz thought it was 80% likely the new strain is >50% more transmissible; as of Dec. 27, superforecaster Juan Gambeiro thought this was 65% likely. As of Dec. 31, infectious disease expert Trevor Bedford expected the new strain to be about 50% more transmissible. I expect we'll get increasingly good estimates over the next few weeks.
The new strain doesn't appear to cause worse symptoms (update Jan. 22: it may indeed cause worse systems), but as Zvi Mowshowitz noted on Dec. 24, if transmissibility is as high as it looks and vaccine rollout doesn't speed up dramatically, we should expect a massive fourth wave of infections in the US "likely cresting between March and May, that could be sufficiently powerful to substantially overshoot herd immunity".
"Overshooting herd immunity" means we achieve herd immunity in the space of a few weeks, with perhaps 60+% of all Americans getting sick; and then (because the total number of infectious people is so high) a large portion of the rest of the population gets infected too even though the virus's effective reproduction number R is much lower now.
Daniel Speyer’s description of overshooting herd immunity in a short period of time:
"More people are infected than it would take to drive r<1. More than 2/3 of the population, using the uniform r=3 model. This still leaves individuals the opportunity to go full-lockdown until vaccinated and avoid infection, but it may mean a few months of never leaving the house without a positive-pressure suit."
[...] The baseline scenario remains, in my mind, that the variant takes over some time in early spring, the control system [i.e., people's tendency to take more precautions when things look more dangerous] kicks in as a function of hospitalizations and deaths so with several weeks lag, and likely it runs out of power before it stabilizes things at all, and we blow past herd immunity relatively quickly combining that with our vaccination efforts.
[...] It seems likely we are looking at hundreds or low thousands [of people currently in the US with the UK strain], perhaps mid thousands, which puts the most likely start of the endgame some time in March if current trends continue and infectiousness is the full amount we suspect.
I think the evidence this past week has strongly favored the new strain being more infectious. What we lack is the knock-down info that would differentiate 50% more infectious from 65% more infectious[.]
Q: What should I do?
After discussing this with someone who’s spent a lot of time studying COVID-19, we agreed that this is the decision point and it’s basically all or nothing. Everyone needs to choose essentially one of three orientations:
- The Protected - You protect yourself now and during the crisis, whatever it takes.
- The Switchers - You don’t protect yourself as much now, but you choose a threshold (e.g., a certain day of the calendar year, or a certain prevalence level of the new strain) at which you will shift to being in the first group if you haven’t caught it.
- The Unable/Unwilling - You accept you’re probably going to catch it (and in all likelihood at least soft prefer catching it now versus later).
Regardless of which group you’re in, everyone who can should build up supplies again and start getting ready for an extended lockdown, possibly with supply line issues.
I think you should pick your basic strategy and make your plans and preparations now, or by mid-January at the latest. The downside of reacting too early is far smaller than the downside of reacting too late.
Given that vaccines are near at hand and (e.g.) the long-term health effects of COVID-19 aren’t well-understood even for young people, I think locking down hard for 2-3 months beginning in late February or early March is probably the best option for most people who can take it.
I understand “locking down hard” to include things like “no interacting with people outside your household,” “exclusively working from home,” “no using public transit or ubers,” and “no going to stores.” You may even want to avoid going anywhere near passers-by outdoors, though I think the risk from walking past someone outdoors in a low-population area is normally (e.g., now in December) pretty negligible.
If you want to keep your risk especially low, Zvi notes:
[T]hose positive pressure suits are still for sale. I think they just work? As in, if you use the $2,000 stuff, which could easily be a lot easier and cheaper than relocation [from the city to the countryside], you can [go] outside freely at almost no risk and all you have to worry about is the air circulation in your apartment.
Ordering packages from Amazon, paying others to go grocery shopping, etc. are under-used by the general public, I think, as good ways to reduce infection risk and cut down the overall spread of COVID-19. I heard a statement from a source in mid-2020 that grocery stores were the largest source of new infections in one part of the US -- many transmission chains are between store employees.
I think COVID-19 is to a very large extent transmitted indoors, and almost entirely via talking to someone face-to-face, or being in the same room as someone who coughs, shouts, sings, or sneezes. Almost all transmission happens via small droplets that hang in the air (a.k.a. aerosols) and large droplets that quickly arc to the ground. I think people should put relatively little effort or attention into preventing surface transmission (a.k.a. fomites), which seems real but much rarer.
(The new strain seems to rely on the same vectors as other strains; its increased infectiousness seems to stem from increased viral load.)
Q: How should I decide whether it makes sense for me personally to lock down hard?
Review the best current estimates of COVID's health risks for your demographic, review your personal preferences and constraints (e.g., whether you’re able to work remotely), and do an expected value calculation to figure out what makes sense for you personally. Your estimate should include a factor for the risk that if you get sick, you may infect others.
Q: What if I want to lock down hard, but also want to see friends who haven’t been similarly locked down?
I think that if both parties lock down hard for two weeks, it’s basically safe to meet up.
On the other hand, I think getting tested and then meeting once everyone gets a negative test result back basically doesn’t work as a high-confidence way to keep transmission low. Viral COVID-19 tests have a pretty high false negative rate, common US tests tend to have a slow turn-around (so you may get infected after getting tested), and tests can miss infections if you take the test slightly too early or too late. (Though it’s “you took your test too early” that’s the main risk, since COVID-19 is most infectious early on.)
More specifically, Paul Bleicher and Marc Lipsitch recommend getting tested 5-6 days after your last exposure, rather than (e.g.) 2-3 days. But Feretti et al. think people are most infectious 2-8 days after they were infected, with some transmission occurring as early as day 0 or as late as day 12; and they treat 5.5 days as the mean time between infection and symptom onset, for people who develop symptoms at all.
This is not to say that tests are useless, just that they're pretty mediocre evidence for this specific purpose.
Relying on symptoms alone to decide a meeting is safe is an even worse heuristic, since COVID-19 is at its most infectious shortly before symptoms start (and some people who spread the virus never show clear symptoms at all).
Note that this "wait two weeks" advice is specific to people who want to lock down hard in preparation for an imminent large fourth wave. I am definitely not advising that all people adopt a policy this strict at all times and places (e.g., when COVID rates are low).
Q: If most people in the US get infected all at once, will it really be possible to keep my own COVID risk low?
If you’re able to work from home, and you’re informed and conscientious, then yes. I once asked an acquaintance of mine who has spent a lot of time studying COVID-19 whether I should be paranoid about things like washing my hands, given how many people in many parts of the US were getting infected week-by-week. They replied:
My guess is you're too optimistic about how cautious typical people are being. If you think typical people are basically isolating, then current prevalence numbers look like "yikes! you can get sick even if you're mostly isolating!". I think that's the wrong model. There's some significant fraction of ordinary people doing [approximately] nothing to avoid infection. There are literal frat parties going on in Berkeley. Literally. People are going to work, and though they're masking while working at the cash register, they're taking their mask/gloves off in the break room with their colleagues and just doing whatever.
[...] I think you should model out an average Berkeley resident as spending multiple hours per week in public spaces [on public transit, at the grocery store, etc.]
In other words, for a shut-in like me who isn’t interacting with people outside my quaranteam at all, I need to drastically adjust downward my intuitive estimate of how risky any given activity is, because COVID-19 is only infecting a small percentage of people every month even though large chunks of the population are doing plenty of in-person socializing and interacting. There’s nothing virtuous about being over-cautious and burning value.
The new strain looks substantially more infectious, and relatively small increases in transmissibility can have a very large impact on exponential growth. But as Zvi notes:
One issue is that even if there's a lot of risk, it's still going to be highly concentrated in people not taking any real degree of precautions. It's still not going to be that easy to get this from walking down a stairwell or through a lobby or building vent - not a 0% chance, but I think that 'reasonable' precautions make one a solid favorite to get out clean even in the 85% infected scenario, because so few people will do even that, but it does mean not seeing anyone and all that.
My take-away here isn’t “you can be relaxed and expect to avoid infection during a fourth wave.” Rather, I take this as a warning against fatalism: it’s not that hard to be one of the 20% (or for that matter 5%) safest people, if you’re able to e.g. work from home.
Q: Is there any way to stop a large fourth wave from happening in the US?
There are three things I could imagine preventing a fourth wave, or sharply reducing its size:
1. The new strain could turn out to be less infectious than it currently seems. Zvi writes:
"I remain confident that we can probably mostly muddle through if there is small increased infectiousness (up to about 35%), at 50% I am highly skeptical we can stop this, and at 65% it’s (probably) already over."
Early estimates of the strain’s increased infectiousness tend to be significantly higher than 35% at present (as of Jan. 6), so I think this is possible but pretty unlikely.
2. Zvi’s “control system”: people could voluntarily respond to the increased risk by taking more precautions, thereby reducing the overall spread of COVID.
I think it’s quite likely that this will be too delayed and too weak an effect. There has definitely seemed to be a control system like this in place in the US, keeping death rates from climbing too high. The problem is that:
- People aren’t likely to sharply change their behavior until they see large increases in hospitalizations and deaths.
- Hospitalizations and deaths lag behind infections by many days.
- The new strain is highly infectious, so multiple doublings may occur between “when a large subset of the population becomes infected” and “when we see a large enough spike in hospitalizations and deaths to change people’s behavior”.
- People's pandemic fatigue may result in a slower and/or weaker response than we would have seen a year ago.
- We already saw last January, February, and March that public health authorities, journalists, etc. are surprisingly bad at understanding exponential growth. Even if their credibility is still high enough to convince the public, it’s very unlikely they’ll understand and communicate the magnitude of the risk early enough to make a large difference.
- Unlike in Jan/Feb/March, we face the added difficulty that the dominant COVID-19 variant can hide the new variant’s spread. Overall COVID rates won't provide a good warning. Instead, you’ll want to pay attention to data about how many COVID cases (in the US and in your area) are the old strain versus the new strain; and you’ll need to complete all your preparations long before overall COVID rates start rising.
3. Widespread vaccination; extensive testing; an unprecedentedly large contact tracing effort and unprecedentedly heavy-handed quarantining and travel restrictions in affected parts of the US.
None of these currently seem likely to me -- not because any of this is a foregone conclusion or because we don’t know how to solve this, but because the incompetence of the US’ COVID response so far makes me very pessimistic that the obvious interventions will be carried out (at all, or in a reasonable or coordinated way).
The only option that I could imagine the US pulling off at this point is widespread rapid vaccination. But so far, almost every part of the US’ vaccination plan has been terribly misconceived and very poorly implemented. In decreasing order of importance, I want to see changes like:
- giving emergency use authorization to the AstraZeneca vaccine (and accelerate the evaluation of other promising vaccines in the pipeline);
- giving everyone one dose of the vaccine before anyone gets a second dose, since this will do much more to slow the spread of the new strain;
- pouring as much money as possible into buying far more doses than needed and encouraging their speedy manufacture; and
- accelerating the vaccination process by removing red tape and focusing on speed above all else.
Patrick McKenzie has an excellent list of suggestions for accelerating vaccination. Matt Kilcoyne’s plan was developed for the UK, but I expect it to have transferable ideas.
Q: Should I expect our current vaccines to be effective against the new UK strain?
Yes! Mutations that reduce vaccine effectiveness are rare in most viruses. (Influenza is an unusual exception.)
Q: Are these vaccines safe?
Yes, assuming you’ve never had a strong anaphylactic reaction to a previous vaccine.
I plan to get vaccinated ASAP once rollout reaches my demographic, and I don’t think of this as a risky action.
Q: When will I be able to get vaccinated?
(Added Jan. 18)
If you’re 65+ years old or a health care worker, likely now or very soon. Check rollout plans for your state. Note that as of Jan. 18, people in a number of states have been having success getting vaccinated early, and rescuing vaccines from being thrown out in the process, by showing up half an hour before vaccination sites finish for the day.
Zvi Mowshowitz predicted on Jan. 6 "about 38 million vaccinated by April 1, 81 million by July 1, 131 million by October 1". As of Jan. 14, the aggregate forecast on the Hypermind prediction market was 65 million by April 1, 128 million by July 1, 184 million by October 1.
For comparison, based on a quick googling, it looks like there are ~18 million American health care workers and ~55 million Americans age 65+. And as of Jan. 14, we had vaccinated ~11 million Americans, including ~4.5 million in the last week.
This suggests that it may be difficult for a lot of non-health-care-workers age 65+ to get vaccinated before a likely fourth wave hits in March, in spite of the new directions from the CDC to prioritize a rapid roll-out to everyone 65+. Zvi Mowshowitz comments:
I hope [Hypermind's] predictions are right, or better yet pessimistic. The difference in those guesses is a really big deal. If we take my model’s guess and boost vaccinations an additional 50%, and the additional infectiousness of the new strain is only 50%, we plausibly (mostly) stop the fourth wave, despite the new strain currently being ahead of my previous guess for that. I’ve added knobs in the spreadsheet to vary both numbers.
Under the optimistic scenario, this increased vaccination pace would be the difference between ending with 40% or 50% of the population having had Covid-19 when it’s all over, with about 24% having already been infected. So this would prevent over a third of all future infections. Speed matters more than anything.
There are 209 million Americans over the age of 18. Given how many don’t want to be vaccinated, these predictions above imply that by April 1 you can likely get the vaccine as long as you care about finding it, and by July 1 you can get the vaccine provided you want it at all, as there would be enough vaccinations to cover every adult who wants one at all.
Q: Will it be safe to get vaccinated during a massive spike in infections?
It’s possible to get infected when you leave your lockdown to go get vaccinated. The first dose confers zero protection until a week or two after you get it.
On the other hand, even if there’s a massive ongoing fourth wave of infections at the time, it seems easy to make the risk from getting vaccinated much lower than the benefits, by calling in advance to make sure the vaccination area is uncrowded and following reasonable safety protocols, and by wearing protective equipment such as an N-95 mask and a face shield. Zvi tells me:
It would take a very extreme situation before getting a vaccine was so actively risky you couldn't do it - we're talking (hopefully) about a quick trip to CVS, which isn't nothing, but assuming they understand the issues you should be able to be in+out in a few minutes with generally good protocols, and potentially they can do it all outdoors. Yes, entering+leaving the apartment isn't a free action, but turning down the vaccine because of this seems unlikely to me."
Getting your first dose of the vaccine seems to confer ~80% protection (after a few weeks), while the second dose confers ~95% protection (again, after a few weeks).
Q: Should I try to get infected now?
I can imagine two reasons to want to do this:
(1) You might be worried that hospitals will be overloaded during a large fourth wave. If hospitals in your area aren’t currently overwhelmed (and aren’t likely to be overwhelmed in the next few weeks), then you might receive better medical care now than you would in March/April/May.
(2) You might think the risk of getting infected with a large viral load is higher later. If you get infected with a small viral load now, you’re likely to have milder symptoms.
On the other hand, it’s pretty hard to guarantee you’re getting a small viral load; if you try to get infected, you’ll need to lock down hard (even if you didn’t succeed) to avoid exposing others to extra risk; and the benefits of trying to variolate at this point are much smaller, now that you have the potential to get vaccinated in the next few months.
Guaranteeing you get COVID probably requires that you expose yourself to a large viral load, which is more dangerous. But if you shoot for a small viral load instead, the likeliest outcome is that you need to self-quarantine for weeks as a precautionary measure, only to find that you didn’t end up infected at all.
I don’t want to claim that trying to catch COVID early is definitely a bad idea for everyone, since I’m not an expert and people’s life circumstances vary a lot. The #1 thing I’d advise is to avoid any action like this that might unintentionally infect others.
Q: What should I do, then, if I’m not willing or able to lock down hard?
Stock up on delicious non-perishable food so you can make your grocery runs in March/April shorter and less frequent. (Or avoid grocery shopping altogether, and spend your microcovids on more valuable things. Tool and explainer for thinking about microcovids: https://www.microcovid.org/)
Front-load your activity – if you can lock down slightly harder later by being less cautious now, then this is probably a good trade.
Prioritize activities that are short and have low exposure risk. These are the cheapest activities in general, in terms of microcovids; and as a bonus, if you do get infected in this way, the viral load is likely to be smaller. Additionally, these activities reduce the risk you’ll infect others (or expose them to a higher viral load) if you’re asymptomatically or presymptomatically sick. If you're doing something like this, take periodic home tests to check whether you’re sick, and avoid older or otherwise at-risk people.