The COVID-19 pandemic has been called “a once-in-a-century evidence fiasco” (Ioannides, 2020), while the editor of the Lancet has declared that “the handling of the COVID-19 crisis in the UK is the most serious science policy failure in a generation” (Horton, 2020). At the heart of the science policy is mathematical modelling, a scientific activity once reserved for mathematicians, epidemiologists and economists, and now widely discussed by politicians, journalists and the wider public. The term “flattening the curve” has gone viral, with DJs, actors and other celebrities exhorting others to do their bit to achieve epidemiological victory. On March 17th, Tom Hanks, one of the first high-profile celebrities to be diagnosed with COVID-19, posted an update to his fans on Instagram, which has had over 1.6 million likes:
“Hey folks. Good News: One week after testing Positive, in self-isolation, the symptoms are much the same. No fever but the blahs. Folding the laundry and doing the dishes leads to a nap on the couch. Bad news: My wife @ritawilson has won 6 straight hands of Gin Rummy and leads by 201 points. But I have learned not to spread my Vegemite so thick. I travelled here with a typewriter, one I used to love. We are all in this together. Flatten the curve. Hanx”.
The day before, The Evening Standard ran an editorial headlined “We must all do our bit to flatten the curve”. In it, Julia Hobsbawm wrote:
“A week is a long time in this new Covid-19 era. Seven days on, I have cancelled all face-to-face events in my networking business for the foreseeable future, including my own book launch next week. It ought to have felt complicated and difficult. It wasn’t. It felt very straightforward and simple. Why? Because I want to flatten the curve. Last week that phrase would have meant ‘get a flat stomach’”. (Hobsbawm, 2020)
Notable in these communications is the entwining of modelling-speak with the everyday of the social – playing cards, eating vegemite, doing the laundry, flattening the curve. As Rhodes and colleagues have observed, “#FlattenTheCurve entangles science into social practices, calculations into materialisations, abstracts into affects, and models into society” (Rhodes et al., 2020).
Beyond simple mantras like “flatten the curve”, whole armies of armchair epidemiologists have emerged in print and social media, producing their own graphs and graphics. In the age of citizen science, popular news articles encourage people to understand the maths and invite them to engage with interactive graphs and graphics. This ‘democratisation’ of expertise has been criticised by some professional scientists, such as Gregg Gonsalves, an Assistant Professor in Epidemiology at Yale, whose research focuses on the use of quantitative models for improving the response to epidemic diseases. In a widely liked and re-tweeted thread on Twitter he wrote, “An epidemic of armchair epidemiology is happening @NYTimes, first @DrDavidKatz, now @tomfriedman decide to opine on the dynamics of epidemics and their control, when neither of them (nor John Ioannidis) work on these topics” (Gonsalves, 2020). Scott Berry, a respected statistician, was similarly disillusioned with the popularisation of epidemiological terms, tweeting: “I’m saddened by the lack of understanding what “exponential growth” means. It’s an adjective without meaning. Maybe it becomes the next #literally? A word with a very precise meaning that is lost… “Hey, dude, that car’s speed is exponential”…” (Berry, 2020).
Historically, epidemiological modelling has been a niche field. First attempts in formulating mathematical expressions to explain the waxing and waning of epidemic phenomena reach as far back as William Farr in the mid-nineteenth century. Even when Ronald Ross, who was credited with a Nobel Prize for the discovery of mosquitos as the vector of malaria, suggested the development of a mathematical “theory of happenings” to explain the dynamics of epidemics (Ross 1916), epidemiologists as well as policy makers saw little use in these novel tools. At the time, epidemics were mostly understood to be the result of the invasion of germs into populations, often associated with global trade, immigration or war efforts. Some considered the constitutions of host populations significant, others favoured explanations based on environmental drivers. The dynamics of smallpox, cholera, measles, tuberculosis and the plague aligned quite well with such mono-causal and straight-forward explanations. It took until after the Spanish Flu in 1918 for the understanding of epidemics to become much more complex and for epidemiologists to develop appropriate ways of thinking. Models could convincingly accommodate the interdependence of pathogen-host-environment, and most importantly, allowed for the development of a new scientific discourse about epidemics beyond simplistic cause-and-effect schemata. But even in the mid-1920s, when Lowell Reed (biometrics) and Wade Hampton Frost (field epidemiologist) designed one of the most influential models of epidemic distribution at Johns Hopkins University, their “simple scheme” remained a tool of academic illustration, a facet of “epidemic theory.” Mechanical models with balls and ramps were built to demonstrate typical dynamics of epidemics to students of epidemiology but were not meant to guide the epidemiologist’s fieldwork, nor were they entrusted with any role in the guidance of policy.
In a TV programme in which Lowell Reed presented their modelling work to the American public, he described mathematical theories of epidemic distribution and models of epidemic dynamics as the “workbench of the epidemiologist’s laboratory,” and “as a critical instrument for experimental epidemiological science”.
In a field which was fundamentally constrained to rely on the (retrospective) observation of epidemics, modelling allowed epidemiologists to emulate the experimental traditions of physiology and bacteriology. However, while modelling remained for Reed and Frost an academic exercise, confined to the “epidemiologist’s laboratory,” without impact on public health advice or the public itself, much has changed since.
In these times of COVID-19, models have left the niche of academic specialism and have assumed substantial validity in the guidance of public health policy and apparently gained some trust in the general public. And while they might have shed associations with the laboratory workbench, they still operate in the realm of experimentation, speculation and simulation. However, now, different camps of epidemiologists and modellers debate models in the mainstream press – for example, Adam Kucharski, an epidemiologist at the London School of Hygiene & Tropical Medicine, wrote a piece in The Guardian titled “Can we trust the Oxford study on Covid-19 infections?” (Kucharski, 2020). In these very public exchanges, models are assumed to be of public interest – and indeed they are, because in the UK, the prime minister has openly stressed that the country’s approach to managing the pandemic is driven by mathematical modelling.
What are we to make of this domestication of mathematical modelling, where to domesticate is literally to bring models into the home? As we have suggested, domestication implies not just representation in the media, but the active appropriation of epidemiological discourse into everyday life – whether through hashtags, celebrity endorsements, or discussions at the breakfast table. During these times of enforced absence from the public sphere, when citizens are largely confined to their homes, what role does the domestication of modelling play in shaping people’s understanding of themselves, their fellow citizens, and their role in shaping the pandemic? As sociologists of science, what questions can we usefully ask, which will help make sense of science-society relations in times of pandemic and ensure that public health policy is scientifically informed and democratically accountable?
The widespread uptake of epidemiological discourse by policy makers and the wider public raises questions about the work that modelling does to create strategies of pandemic response, galvanise public support, and build a well-informed populace. The role of modellers in creating knowledge of COVID-19 and communicating this successfully to different audiences needs to be understood, as does the domestication of models in the public imagination. This domestication is not uncontested, as the open disagreements between modellers and the epistemological gatekeeping in print and social media show. It thus becomes a question not only of how the public assume an epidemiological imagination, but how modellers recruit the public in their work. As Steve Hinchliffe (2020) underscores, it is important to listen for the other voices and forms of knowledge in this pandemic, to pay attention to the specificities and spatialities of local conditions and practices.
‘Epidemiological publics’ might then refer both to the way in which population groups are constituted and represented through modelling and the way in which the production, circulation and use of epidemiological models in policy making and in the media creates particular forms of public participation. In other words, it refers both to the production of knowledge and to that knowledge’s use. It directs us to investigate the moral and political capacities with which mathematical modelling comes to be invested in times of pandemic and the forms of participation it impels. This is important because public health strategies to manage the pandemic rely on people participating in disease control measures such as social distancing, hand-washing, self-isolation and lock-down. Without strong and informed public participation, these measures will fail; without trust and democratic accountability, the long-term future of science-based policy is in doubt.
Catherine Montgomery is a Sociologist of Science, Technology & Medicine at the Centre for Biomedicine, Self & Society at the University of Edinburgh.
Lukas Engelmann is a Chancellor’s Fellow in History and Sociology of Biomedicine at the University of Edinburgh.
Berry, S. 2020. I’m saddened by the lack of understanding what “exponential growth” means. It’s an adjective without meaning. Maybe it becomes the next #literally ? A word with a very precise meaning that is lost… “Hey, dude, that car’s speed is exponential”… In:@STATBERRY (ed.) 3:53pm ed.: Twitter.
Gonsalves, G. 2020. An epidemic of armchair epidemiology is happening @NYTimes, first @DrDavidKatz, now @tomfriedman decide to opine on the dynamics of epidemics and their control, when neither of them (nor John Ioannidis) work on these topics. In:@GREGGGONSALVES (ed.). Twitter.
Hinchliffe, S. 2020. Model Evidence – the COVID-19 case. Somatosphere[Online]. Available from: http://somatosphere.net/forumpost/model-evidence-covid-19/[Accessed 09 April 2020.
Hobsbawm, J. 2020. We must all do our bit to flatten the curve. Evening Standard [Online]. Available: https://www.standard.co.uk/comment/comment/we-must-all-do-our-bit-to-flatten-the-curve-a4388381.html[Accessed 01/04/20].
Horton, R. 2020. The handling of the COVID-19 crisis in the UK is the most serious science policy failure in a generation. Last week, the Deputy CMO said, “there comes a point in a pandemic where that [testing] is not an appropriate intervention.” Now a priority. Public mesage: utter confusion. In:@RICHARDHORTON1 (ed.) 07.13 ed.: Twitter.
Ioannides, J. P. A. 2020. A fiasco in the making? As the coronavirus pandemic takes hold, we are making decisions without reliable data. STAT[Online]. Available: https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-as-the-coronavirus-pandemic-takes-hold-we-are-making-decisions-without-reliable-data/[Accessed 01/04/2020].
Kucharski, A. 2020. Can we trust the Oxford study on Covid-19 infections? The Guardian [Online]. Available: https://www.theguardian.com/commentisfree/2020/mar/26/virus-infection-data-coronavirus-modelling [Accessed 01/04/20].
Montgomery, C. M. & Pool, R. 2017. From ‘trial community’ to ‘experimental publics’: how clinical research shapes public participation. Critical Public Health,27,50-62.
Rhodes, T., Lancaster, K., et al. 2020. A model society: maths, models and expertise in viral outbreaks. Critical Public Health [Online]. Available: https://doi.org/10.1080/09581596.2020.1748310.
We draw here on previous work on ‘experimental publics’ in relation to clinical trials: Montgomery, C. M. & Pool, R. 2017. From ‘trial community’ to ‘experimental publics’: how clinical research shapes public participation. Critical Public Health,27,50-62.