Risk, Decision-making and Covid19
The novel coronavirus (Covid19) has brought the world to a standstill. In just a few months, illness, and the fear of it, have swept the globe up into a collective freeze. At the time of writing, more than half of the world’s population is confined to their homes. Modern life has stopped: commerce, exchange, gatherings, jobs, schools, travel, all have ceased. But there’s a more widespread, if diffuse darkness hovering over hovering over the planet, infiltrating protective masks and gnawing at our sleeping hours, sowing fear and panic, and that is uncertainty.
If asked, few would deny the uncertainty we face. The word is peppered across email subject lines and press releases. Yet at the same time, from the early days of the pandemic, when it was still just an exotic illness, headlines were instructing us on risk assessment, with a puzzling confidence in doling out advice and clarifications. For instance, the imperative to avoid wearing masks, because they were useless, was logic defying, yet that advice was delivered with such certainty that it was picked up and repeated by numerous people, including clever folk. The problem with following advice on Covid19, even now, months into the pandemic, is that the virus is novel – and therefore by definition, remains fraught with unknown. But the unknown comes in different grades: known and unknown. You may chuckle and roll your eyes at the memory of that sound bite but these forms of uncertainty in the field of decision-making sciences are called, respectively, risk, and ambiguity.
In this New York Times article, experts describe our irrationality in the face of Covid19. The tone toes a paternalistic line and can be summed up as: poor dears, can’t compute risk and act accordingly, victims of their biases. But forget the tone, and even the conclusion. The first is immaterial, if tedious coming from the field, and the second is still up for debate. It’s the argument they were wrong on, the very measure, what they deign to call the “math” in the article. We don’t have a reliable measure of Covid19 –related risk to under- or over-weight. All the metrics published so far, from mortality, to prevalence, to symptom frequency, to re-infection and antibody production – are so variable that they carry little information. In other words, we are swimming in a sea of Covid19 ambiguity.
Risk is a metric we compute from data. It allows us to make an informed decision – we can choose how much unknown we are willing to take on. Risk is the known error on your choice, the plus or minus on your outcome. The key to having an accurate measure of risk is repetition of an event. For instance, we have a very good idea of how long the human gestation period is because have accumulated experience on human birth. For most of our everyday decisions, we implicitly project likely outcomes and their associated risk. For bigger decisions, like buying a house or insurance, we might take out a calculator or look up some numbers. What’s nice about risk is that people can decide for themselves how much they are willing to take on – because the bounds of the unknown are known. Another thing about risk? It’s unavoidable. Uncertainty in a system can be minimized by sampling and learning, but not beyond risk. So risk also goes by the name “irreducible uncertainty”. That means, if you bring a new baby into the world today, it won’t change the uncertainty bound of a human gestation period (it may however change your estimate of your own pregnancy durations, but chances are you will never meet the criteria to reliably estimate such a thing1).
Now ambiguity on the other hand is the infamous – and it is truly that – unknown unknown. Ambiguity exists in the new, the novel, the untried. Ambiguity is the sigh you suppress on the first day in your new school, when you have yet to figure out who is going to be your friend, today, and/or the rest of the year. That uncertainty is uninformative. The best you can do is sample, and learn. Sample, and learn. Repeat until you have a measure of risk. It can be a painful process. Or not. But the road traveled from ambiguity to risk will be time-consuming. You could take a short cut by asking an expert – for example, asking a kid that’s been in the same school for many years who’s nice and who isn’t. However, you’d have to have a good measure of risk on that kid – what if he’s a bad judge of character? Or what if she thinks you’re competition and wants to see you fail? Another way to shorten the learning curve is by observation and imitation, a fun strategy that human toddlers thankfully exploit2 (think of the error in the trial and error strategy). In any case, by the end of the year, if you’ve survived, you will have a measure of risk concerning the kids in the class, as well as some friends, enemies and stuff in between. So ambiguity is the reducible uncertainty that can be driven down to valuable, usable, information (risk). Ambiguity can also be called Knightian uncertainty3.
The tide has turned against the viewpoint held in the article, but the tendency towards bold conclusions has not gone away. Models projecting stuff are bandied about online by all and sundry (few with error bars, hmm) with accompanying prophecies. People may hold on to one model, the one that fits their desire, field, emotional tendencies, or viewpoint best, or even still, the one that gets passed around social media the most. All that said, the scary, sobering and boring order of the day is that when it comes to Covid19, we cannot speak of known unknowns, only ambiguity. And few have been brave enough to simply state that deeply unpleasant reality. Ambiguity is a product without a buyer: no one wants it (except, perhaps, the aptly named risk-seekers). But if we don’t recognize these uncertain times for what they are, we will pay for it later, in distrust of the experts, the media, science and statistics4. So it behooves the media to temper headlines and experts to eye their own moment in the sun with deep skepticism.
- https://www.simplypsychology.org/central-limit-theorem.html
- Want, S. C., & Harris, P. L. (2002). How do children ape? Applying concepts from the study of non‐human primates to the developmental study of ‘imitation’ in children. Developmental Science, 5(1), 1-14.
- Knight, F. H. (1921) Risk, Uncertainty, and Profit. Boston, MA: Hart, Schaffner & Marx; Houghton Mifflin Company
- Van Der Bles, A. M., van der Linden, S., Freeman, A. L., & Spiegelhalter, D. J. (2020). The effects of communicating uncertainty on public trust in facts and numbers. Proceedings of the National Academy of Sciences, 117(14), 7672-7683.