On Average

our lives are ruled by numbers, and this is ordinarily a good thing; in times of pandemic, a very good thing. As past efforts to combat epidemics show, social statistics and public health crises have been historically intertwined, from the bills of mortality that counted deaths in plague-stricken London in the seventeenth century to the latest graphs and figures showing the worldwide case-fatality rate of COVID-19 right now. Without reliable numbers about how many people are infected, and when and where, all attempts to contain and ultimately explain the latest plague must stumble in the dark.

But numbers are as various as words, and states of exception highlight the risks of relying too heavily on one kind of number: the ubiquitous average. It is in the nature of data, including quantitative data, to exhibit variability. Already in the eighteenth century, astronomers bothered by differing observations of the positions of the same stars and comets were devising statistical methods—in this case, the method of least squares—to smooth out the scatter of observation points and find the most probable value. In the nineteenth century, statisticians of social phenomena, from births to suicides, averaged numbers annually and marveled at their stability from year to year. In the twentieth century, empirical researchers in fields ranging from physics to psychology to economics invented sophisticated methods for dealing with outliers, those errant data points that diverge wildly from the swarm of other points through which a plausible line can be drawn—the average.

Nor are averages confined to the sciences: thinking in averages, ironing out variability, is the way most of us deal with most of life. We plot our daily routes by average traffic patterns, dress according to averaged weather predictions, calibrate a thousand daily risks from driving to eating out by average experience. Under stable conditions, when nothing much changes from one week to the next, one year to the next, thinking in averages is a perfectly rational strategy.

When, however, novelty erupts in the midst of routine, averages and the infrastructure built upon them fail us. This is not because the novelty is wholly unforeseen: geologists know that sooner or later California will be shaken by a tremendous earthquake; economists know that sooner or later just-in-time supply chains will snap because of monsoons or pirates or politics; airlines know that sooner or later to their efficiency-driven policies will come a cropper in a blizzard; epidemiologists know that sooner or later a dangerous pandemic will ambush the world. The timescale resolution of these disruptions—once in 10 years, or once in 100 years?—is too coarse for short-term planning driven by economic goals and a four-year election cycle. Yet without an elongated planning timescale that builds redundancy (for example, more than the number of hospital beds needed on average) into institutions and infrastructure, we are doomed to disaster when the unpredictable but in the long run inevitable state of emergency hits.

This much would have been true for almost anywhere, anytime in human history. But there is reason to think that in an era of nearinstantaneous global connectivity without global governance and with runaway climate change, even the most privileged inhabitants of the world are leaving the realm of stable, predictable conditions in which thinking in averages makes sense.