Pan demos
Disclaimer: I am not an epidemiologist, don't believe me.
A few weeks ago, a brutal Marine injected some science into the muscle tissues of my left upper arm. I qualified for such an injection through my extraordinary position as a member of the general public who waited long enough.
This enabled me to talk to friends in person for the first time in 15 months. After the prolonged deprivation, I have much more appreciation for such simple pleasures of life. Talking to non-scientists and talking off zoom is really cool, you guys. But a dark cloud lingers over all these conversations, that of the ongoing Covid-19 pandemic. Stories of quarantine hobbies, international border crossings, vaccine searches, first-hand Covid experiences, and struggles to smell and taste food again. And naturally, on the same questions, people disagree.
In March 2020, a large part of my immediate social circle was making up epidemiological models, mostly by recalling what they know about exponential growth and SIR-type models. Physicists are particularly guilty of that, and very sloppy about parameter fitting. After reading a bunch of modeling attempts across media such as Facebook, Medium, and arXiv, I came across this wonderful piece on Reddit that echoed my thoughts (usual internet rules apply: don't read the comments). At that time, I chose a high-level strategy to not work on Covid as an academic subject and keep doing my regular research.
But even if I choose to not contribute to the information noise surrounding Covid (at least in public venues), I can read the original studies without compromising research ethics. Reading research papers is a skill that one acquires in bizarre ways in graduate school over years of practice, and I am still endlessly amused that I can read a paper from somewhere like The New England Journal of Medicine. Reading research papers on Covid is a slightly cleaner way to get the fresh information from primary sources, before it is re-framed by journalists, public health officials, and comic writers. But reading such stuff, especially outside of my immediate field, is hard, and gets me worried about all the pieces that end up misrepresented. I worry about what the "vaccine efficiency" number means (here is a great infographic that helped me). I worry about the confidence intervals on parameter estimations for the disease and the vaccine. I worry about the conflation between the numbers of people who "got 1 shot" and "fully vaccinated". I worry about whether vaccinated individuals can transmit the virus.
Someone I was talking to quoted a study linking vaccination progress with economic recovery. The loose quote was along the lines of "we already vaccinated 50% of people, so soon we would be pretty much done and the economy will come back". It is true that the United States vaccinated about 50% of people in 5 months of active campaign (with some significant accounting assumptions). After the initial ramp-up of opening the vaccination sites, the progress was almost linear, with a small slow-down recently. These claims are only based on real data, what actually was reported by the CDC. But wait, what is that "would be pretty much done"? What is the vaccination projection here and what is the confidence interval? How does it connect with widespread vaccine hesitancy and international travel with countries with much lower vaccination rate?
I made another high-level strategic choice: to not make Covid-related decisions on behalf of anyone outside of my household. But I will say one thing as a complex systems scientist: the pandemic is a collective phenomenon as much as it is a biological phenomenon. The first time I was taught epidemic models in a classroom, the instructor wrote down an equation, where the left hand side described the interaction network determined by social behavior, and the right hand side described the biological properties of a given disease. The epidemic progression in that simple model depends on which side is larger. But of course a few years ago this description was met with hesitant "it-can't-happen-here" stares.
The main collective effect that everyone is buzzing about is the herd immunity. If a large fraction of the population is immunized against a particular disease, then the disease can't spread, and the remaining few percent who can't get a vaccine due to medical or religious exemptions are still indirectly, effectively protected because they are very unlikely to be exposed to carriers. Herd immunity is usually discussed for the diseases that are nearly eradicated in the modern world through lengthy vaccination campaigns - that is, diseases for which we are way above the herd immunity threshold. But humanity hasn't tried before to strangle a novel pathogen within a year or two, this is new territory for everyone. The herd immunity threshold for Covid-19, given today's vaccines, is estimated at something like 70-80%. But you should not take this as some magical, if slightly uncertain, number. The moment we hit a putative 81% vaccination target, Covid will not cease to exist.
The softness and ambiguity of this threshold has at least two reasons: one social, one biological. Both of these boil down to the Covid pandemic not behaving like mean-field models. Mean-field models assume that everyone in the population has an equal chance to meet everyone else, and that the disease spreads continuously. Above the herd immunity threshold, the infected population will start to decline. Below the threshold there is no decline; just above the threshold the decline is very slow. In technical terms, applied to this naive model, herd immunity appears like a continuous phase transition. The only way to reliably get rid of the disease is to make everyone immune, but that will never happen.
But if you start zooming in on the situation, you will notice the social heterogeneity. Some of us reliably escaped Covid infection behind locked doors, while others can't or don't want to lock up. Some of us can go to hospitals for healthcare, others can't. Some of us have access to vaccines, others don't. When you say the threshold is at 80%, that is 80% of whom? Of the population of the world? Country? City? But what if people start crossing regional and national borders? What about the age groups? We know that children can catch and spread Covid, while vaccine safety is not ascertained for them yet. Many models advocate intentionally unequal distribution of vaccine to induce herd immunity using a small number of doses, but those models are not ironclad and the ethics are debatable.
It is not only the social structure of the community that matters, but the biological heterogeneity of the virus itself. In a way, SARS-CoV2 is a "bad" virus since it has a much lower chance to kill the host than its cousin SARS-CoV1. Yet back in my childhood SARS-CoV1 was just a tiny blip in the news cycle in a country far away (people from there would beg to differ). Infection with SARS-CoV2 can lead to a wide variety of outcomes, including asymptomatic spreaders, which allows the virus to spread rapidly and undetected. Add to these the poorly-understood superspreading events, and you see how being a "bad" virus helps SARS-CoV2 in an ecological sense as it builds up a population across many hosts.
But when ecology gives a large, rapidly replicating viral population, evolution kicks in. Every time the virus replicates within a host, it has a tiny chance of copying RNA incorrectly, resulting in mutant offspring with a slightly different genetic code. These minute differences in RNA are mostly harmless, though they serve as unique temporal fingerprints and help scientists keep track of virus spreading patterns. But sometimes the RNA sequence change results in a slightly different protein structure, thus creating a virus variant. The variants that are more infectious hop onto more hosts than the previous ones, and thus replicate more and take over the viral population. This is Darwin's natural selection unfolding in front of our eyes. The chance of a new dangerous variant to emerge per unit time is roughly proportional to the virus population in the world, i.e. the number of active Covid cases. One year ago, when the cumulative number of Covid cases was 25 times smaller, virus mutations were the subject of scary it-can't-happen-here science fiction (here is an excellent short story in Russian). Now they are a reality. Every new variant slowly eats away the protection afforded by first-generation vaccines. So you getting a vaccine does not make you safe forever so long as the pandemic reigns across the world.
Given the vaccine options in my country of residence, I chose to get my shots without hesitation - but this choice is contextual. Getting the shots is by no means the end of the road. The pandemic and your choices with regard to it are not about you. It's about pan demos.