How about we don’t all get COVID in London?

Note: It usually takes me quite a while to write and publish anything online, but this seems pretty pressing so I’ve done it in a hurry. Please excuse any omissions or mistakes and feel free to yell at me about it in the comments.


TL;DR: (If not for this post), I predict that the median EAG London attendee will be less COVID-cautious than they would be under ideal epistemic conditions* and under their own values.

*Ideal epistemic conditions means something like “if they were better informed about the current state of affairs vis-à-vis COVID in London and any other relevant descriptive facts, and thought long and hard about how much risk they’d like to incur given the costs and benefits of various interventions like masking, social distancing, and getting a last-minute booster.” It does not mean “if they had same degree of risk tolerance as me or as I’d like others to have.”

Point 1: case rates in London are likely higher than you think

Relevant info from the UK Office for National Statistics

The subtitle just above links to a report that was released two days ago on April 8, 2022.

  • In England, the percentage of people testing positive for coronavirus (COVID-19) remained high in the week ending 2 April 2022; we estimate that 4,141,600 people in England had COVID-19 (95% credible interval: 4,033,600 to 4,249,500), equating to 7.60% of the population or around 1 in 13 people.


  • In the week ending 3 April 2022, the Omicron BA.2 variant remained the dominant variant across all UK countries; the percentage of people with infections compatible with the Omicron BA.2 variant remained high in England and continued to increase in Wales, however the trend was uncertain in Scotland and Northern Ireland.

More concisely, it looks like almost 8% of the English population had COVID as of April 2.

According to Our World in Data, the UK and the US had pretty similar confirmed new case numbers until about March 1, when the UK’s case rate started climbing but the US’s did not.

Anyway, the point is that the UK positive rate looks to be about seven times the U.S. rate. For those of us flying in from the U.S., then, our day-to-day impression of how things are going at home is probably misleadingly benign.

Point 2: people have kinda stopped giving a shit

Until about a week ago, so had I. Then I:

  1. Read that case rates on Georgetown’s campus were increasing a lot

  2. Realized that it would be bad for anyone in the GU group to get COVID and either (a) find out and stay home or (b) not find out, go to London, and risk passing it on to other EAG attendees, and

  3. Assumed the role of the grumpy COVID hawk and ordered our group a bunch of rapid tests on Amazon for just over $7 a pop (or <1% the cost of a DC<->London rountrip flight) which were delivered the next morning.

I think the two takeaways from this story are best conveyed via meme:

It ain’t 2021 no more

More generally, I have the pretty banal impression that (at least among EA’s typical demographics) preventative measures like masking, frequent rapid testing, active ventilation, and further vaccination have largely fallen by the wayside. Given the state of the world (e.g., low case rates, Paxlovid) now, this makes total sense and I’m not condemning it.

But I am trying to point out that this limits the inductive validity of extrapolating from the epidemiological outcomes of previous in-person EA consortia how useful it is to conclude from the success of 2021-early ’22 EA events that everything is going to be fine.

Maybe it will be fine, I don’t know, but we’re out of sample here! Never before has there been such a large EA event in a city with such a high COVID-positive rate with attendees from around the world, many of whom haven’t been doing much to mitigate viral transmission these last few months.

Something about individual or total COVID risk changing dramatically for groups made of individuals whose risk of catching and/​or spreading the virus is correlated with one-another thanks to math

I don’t have the time to figure out how to present this idea accurately and concisely, but I’m pretty sure the following thing is true:

Suppose in world , people each of whom have a probability of having COVID get together in a group in such a way that the probability of transmission from one infected person to a non-infected person is (where depends on things like masking and ventilation) and this results in each participant having a probability of ending up with COVID.

Now suppose in an otherwise-identical world , each person begins with a probability of having COVID (i.e., 1% more likely than in world ). Then at the end of the event each person has a probability of having COVID, with for large .

(It’s not a quote; the formatting just makes it look nice)

If anyone wants to clarify what I’m trying to get at here or link to something good on the matter, I’d be very grateful!

What we can do about it

1) Get another booster (if you decide it would be wise)

I’m guessing the single “best” (i.e., most individually and altruistically protective) thing to do is to get a booster by the (end of the) day after tomorrow, April 11. In the U.S., second boosters aren’t approved by the FDA for most people under 55. But, to be frank, I think it is ethical, easy, and not legally risky to get one even if you’re not in this category. For more about second boosters, check out Zvi’s post.

2) The usual spate of non-pharmaceutical interventions

Hey man, I don’t like wearing N95s either, but they seem good for reducing your risk of transmission and infection! Same with frequent rapid testing, ventilation, social distancing, the whole nine yards. I haven’t thought about it enough or done enough research to make a normative claim here, but I will make the utterly banal descriptive claim that doing this sort of thing will reduce the expected number of COVID cases caught at EAG.

3) Figure out what set of things a reasonable EAG attendee should actually do and write about it on the Forum

One of EA’s unusual advantages is that it has a laughably high concentration of extremely smart and numerate people with a good understanding of epidemiology and/​or quantitative modeling. So, go make this post look like a second grader’s crayon scribble next to one of these hyperrealistic masterpieces by figuring out in more detail how worried we should be and what the best practical risk-reduction options are. Your fellow EAs will thank you!

Final note: I may update this post in the next 1-2 days with additional information. Any substantive changes will be labelled as such :)