Model the Pandemic, Model the World: Which World? Whose World? – Fleur Johns

The global COVID-19 pandemic has been the subject of extensive regulatory activity, primarily by national governments but also by international organisations. This post draws attention to one regulatory technique that has been central to this activity: modelling. Rather than canvassing the merits and demerits of models of different kinds, it argues that at least as much public attention should be devoted to the world-making effects of models as is customarily directed at multilateral treaties or at national legislation with international impacts.

“We need new models” has been a regular refrain amid the COVID-19 pandemic. New models of vaccine delivery; new models of disease spread and mortality; new models of aged care; new economic models; new models of governance and public engagement: all these and more have been called for of late. In these various calls, the term “model” means slightly different things, yet, the term is nonetheless a recurrent point of reference. It is now ubiquitous among efforts to navigate and regulate the complexities of contemporary life, especially so in a global pandemic. In Australia, modelling has been central to government communication with the public around key decisions, including decisions affecting international legal relations. Models have been used to explain and justify border control and quarantine measures, travel restrictions, social distancing requirements and economic stimulus measures. In Australia and elsewhere, the assumptions and outputs of certain types of model have become embedded in law, policy and official guidance, thereby promoting particular understandings of social and economic life. Global requirements to maintain between one and two metres’ social distance between persons, for instance, under penalty of fines or even jail terms, are based on models of respiratory disease transmission dating back to the 1930s. Much as people remain attentive to legislation or treaties coursing through our parliaments, it is important that we attend to these models’ centres and peripheries, foregrounds and backgrounds, hierarchies and priorities, preoccupations and blind spots.

Models and modelling

Scientific modelling; mathematical modelling; financial and economic modelling; and social modelling: across these different areas of work, the word “model” denotes a representation of, or proxy for, some target about which knowledge is sought, whether that target be actual or ideal. A model describes a structure or puts forward an archetype or set of archetypes. A model is also an analogue in the sense that it typically posits relations of similarity and difference to some worldly phenomena, or to a theoretical description of some worldly phenomena. Models’ analogical status does, however, vary in degree. Some models are designed to be positive analogues of the “real world”. An example would be models used in species distribution modelling, to try to predict the distribution of an extant species over space and time. Others are “working pictures” developed for instrumental purposes and then dispensed with – at most, only ever formally analogous to something in the world. One example of the latter would be English chemist John Dalton’s early nineteenth-century modelling of the atom on a hard, wooden ball like a billiard ball. The idea that the model necessarily represents something other than itself may also be strained. Some models produce data in their own right about their own rendering of non-existent phenomena, as when Daisyworld computer simulation models produce data about planetary scenarios such as life never having existed on Earth. Models have different relationships to the world, but they invariably have things to say about the world – about what it is, and what it isn’t.

Modelling the pandemic

SARS-COV-2, the novel coronavirus that causes COVID-19, and the associated pandemic, have been the focus of extensive scientific and mathematical modelling, through computer simulation especially. Certain model-derived numbers have loomed especially large in popular consciousness and governmental communication about the pandemic, two examples being the basic reproduction number (R0) and effective reproduction number (R). Likewise, particular instances of epidemiological modelling appear to have been particularly influential. The impact of the work of mathematical epidemiologist Neil Ferguson and his team at Imperial College London is a noteworthy example. Projections from their models reportedly prompted changes in UK governmental policy.

Financial and economic modelling of the actual and projected impacts of COVID-19 has likewise been widespread. OECD economists have, for instance, modelled base-case, best-case and downside scenarios of the pandemic’s global economic effects using the NiGEM Model: a quarterly econometric model, maintained by the National Institute of Economic and Social Research in Britain, that is based on real economic data from 46 countries (28 from Europe, including the UK; eleven from Asia and Australasia, six from the Americas, and only one from Africa) and some 19 regions. These modellers’ focus has been on extraordinary disruptions produced by the pandemic – interruptions in supply, declines in demand, and loss of confidence – rather than pre-existing, structural features of the economy bearing upon COVID-19 outcomes, such as inequality, urbanisation, or the uneven distribution of access to healthcare. In economic modelling, the pandemic’s economic repercussions are regularly cast as “fallout”, a “hit” or a “shock”, as though analysing a military attack, industrial accident or natural disaster, thereby highlighting the influential story-telling dimensions of modelling practice.

Social modelling of a less technical kind has also been apparent in analyses of the COVID-19 pandemic. In scholarly, clinical and public discussion of the pandemic, certain national archetypes of COVID-19 policy response have been popularised and compared. Those advancing particular policy recommendations have frequently done so with reference to one or other national model – the “Singapore model”, for example. The “Swedish model” – a “relaxed strategy” premised on the build-up of herd immunity within a national population – has been a particular target of scrutiny and debate. Social modelling supports the idea that collective social conduct and legal relationships are best organised, grasped, and evaluated by recourse to a pre-existing array of archetypes, assigned to national containers. It fosters a tendency to take something off the shelf, as it were, rather than approach social analysis and policy-making ab initio.

Models as arguments about the world

The modelled world of the COVID-19 pandemic is, in all these versions, presumed highly governable. For all the challenges posed by the pandemic, the efficacy of human governance in the face of it is more or less presumed. Borders may be closed. Bodies may be rearranged in space and time and contained within categories (nations and genders, for instance). Modelled worlds are amenable to varying degrees of human mastery (depending on their stochastic dimensions and error rates), but most tend to have humans at both their centres and their helms. The systems that they represent are largely anthropocentric, even though the precipitant for these models’ creation was viral zoonosis – the transmission of disease from nonhumans to humans (as in the case of SARS-COV-2). The modelled world of COVID-19 is, on the whole, a remarkably compliant, human-centred world, relatively inattentive to the permeability and contestability of all manner of social, political, and biological boundaries.

Because of its orderliness, the modelled world of the COVID-19 pandemic is necessarily selective. Those people for whom the governance measures canvassed in models, such as social distancing measures, are likely to be problematic, do not tend to feature prominently. Examples include homeless persons or those otherwise living under conditions not amenable to social distancing; disabled persons requiring intimate care; sex-workers and other non-health-related personnel whose livelihoods depend on physical intimacy. Such persons may only register in the unexplained negative spaces of a model: perhaps as a percentage of the population presumed non-compliant. Those negative spaces may be read to invite governmental intervention, or they could be interpreted as too intractable to be worthy of attention. Either way, those upon whom prevailing governance techniques are more likely to have clear, visible purchase are consistently in the foreground of COVID-19 models. They are first in line as objects of analysis and care. Meanwhile, those deemed harder cases (the itinerant, the dependent, the unhoused and so on) tend to recede into the background of modelled worlds.

In these and other ways, COVID-19 modelling has been a mode of argument as well as an analytical practice. To model is to give shape, to craft, or to fashion. Models assemble certain elements and entities and offer them to people’s experience already linked together or kept apart. When models feature humans, or human proxies, they confer upon those figures certain characteristics, functions, needs, and desires, and strip away other properties. Modelling entails determining precisely what will suffice to approximate that which is modelled. In so doing, modelling involves carving out cores and dispensing with inessential aspects of phenomena represented. These features – and the models in which they are embedded – often travel and persist. 

Models as artefacts with politics meriting questioning

Much of the critical commentary on public decision-making by reference to models has revolved around models’ potential to mislead. Of particular concern has been models’ propensity to generate an illusion of truth, integrity and predictive capacity while exhibiting any number of known limitations and potential weaknesses. These may include: poor or biased input data; empirically incorrect assumptions; highly sensitive estimates; thin historical analysis with inattention to prior model-based outcomes; and lack of transparency. Yet models merit questioning on a broader range of grounds than these and regardless of the degree to which one or other model may be proven “right”.

Even the strongest and soundest of models offer up particular, selective renderings of the pandemic and the world it has afflicted – renderings that may be legitimately open to question and controversy. Models are what Langdon Winner has called “artifacts with politics”. They champion and normalise certain configurations of relation, power, capacity and authority over others. Some of these configurations become embedded in law and policy. In this light, the public – including the Australian public – need to scrutinise those models through which they are asked to understand and address the COVID-19 pandemic with the same care and circumspection as they do treaties or bills considered in parliament. Like treaties and legislative texts, models are often complex. Navigating them may require the support of qualified intermediaries and interpreters. We do not shy away from these challenges in the course of exposing conventional (e.g. legislative or treaty-based) instruments of national and international governance to public debate. We ought not to do so when scrutinising and debating models.

Fleur Johns is Professor of Law and Australian Research Council Future Fellow at UNSW Sydney working in international law, legal theory and law and technology. This is a snapshot of a chapter forthcoming in a book ‘Economy and Society in the Time of Covid-19’ edited by Didier Fassin and Marion Fourcade.