What Type of Governmentality is This? Or, how do we govern unknowns

This article is part of the following series:

In Le Temps, a French newspaper, anthropologist Julie Billaud wrote that the governance of COVID-19 represents “a move towards a biopolitical mode of governance that aims to manage human collectives through statistics, indicators, and other measurement tools.”[i] Others are less convinced. Social theorist Joshua Clover has written in Critical Inquiry, a French online journal,that COVID-19 represents not a push towards biopolitics, but an abandonment of it; suggesting that instead we have embraced a post-biopolitical era where we have ironically returned to “capitalism [as] sovereign.”[ii]

However, I believe these attempts do not go far enough in recognizing the novel subjects of governance at stake within the type of governmentality which is on display today. In this piece, I take a step back from these two positions to think through empirical epidemiological evidence to understand the technologies of governance at play within Western populist governments, namely the United States and the United Kingdom. First, is the governance of COVID-19 really a biopolitics? And if so, what form is it taking? I suggest that we are not governing people as much as we are governing the unknowns of the virus. When the subject of governance is the virus’s unknowns, the state becomes a privileged site of negotiating ontology, specifically that of the human-virus relation. I conclude by opening a discussion for what is at stake for these governments to not know SARS-CoV-2.

Biopolitics, revisited

Foucault’s seminal lectures on biopolitics reveal how the state’s claim to authority is in the power to make live and let die. Reflecting on late 19th– and early 20th-century state practices, Foucault discusses how the management of populations is central to the state’s ability to perform this power.[iii] The state produced categories, both biomedical and otherwise, to perform ubiquitous calculations of the population.

Armstrong suggests there was a change in the biomedical observation of the population with the introduction of the British dispensary in the early 20th century.[iv] For Armstrong, the dispensary represented taking the management of populations into the community. This emergence coincided with the rise of miasmatic theories of public health. Miasmatic theory flipped the long-held assumption that citizens were victims when afflicted with disease, but instead citizens were producing the unhygienic environments where diseases could flourish. The dispensary became responsible not just for tracking disease, but for tracking births, deaths, school attendance, and other population statistics. Some may attempt to suggest that this is exactly what is being seen today: the close monitoring of the potential patient which demands social distancing, working from home, and unprecedented border closures.

Latour identifies the figure of the hygienist as one which not only tracks how the characteristics of populations create environments, but how environments and their microbes are part of the social life of populations.[v] The difference from Armstrong is that — as principal actors in the co-constitution of social life — microbes must also be measured: this is what Paxson terms microbiopolitics. The “microbiopolitics [at stake between microbe and human acts] as a productive force, with calculation, classification, and cultivation” of microbes.[vi] Again, if we are not careful, perhaps we may conclude that this is exactly what is being seen today: calculation of the risk of SARS-CoV-2 upon its interaction with a human.

Key to the concept of biopolitics is measurement, and principally measurement of the population and the non-human when it is co-constitutive of the population. Within measurement there is the assumption that data is 1) comprehensive and 2) accurate. However, this assumption is not always true within biopolitics; part of the state’s production of power can be intentionally foregoing measuring populations or measuring inaccurately. Biehl has discussed how AIDS populations in Brazil are excluded from measurement and are left to die.[vii] Gupta has shown how approximating census numbers in India creates below poverty line statistics which are unrepresentative of social realities.[viii]

COVID-19 encompasses both: data is neither comprehensive nor accurate. The measurement which is at stake is not necessarily about populations, but the virus. The genealogy of biopolitics I have briefly outlined only strays from measuring human populations to understand how the non-human affects human populations. The current way of knowing involves inaccurate and uncomprehensive data about populations and the virus during COVID-19 and asks us to question what connection this type of governmentality is making between the virus and the human to be able to better understand the potential of a biopolitics to be at play.

Technologies of Governance in SARS-COVID-19:

If COVID-19 was a biopolitics, one would expect to see some form of data production of populations. The majority of nations have foregone the classic 19th-century public health strategy of test-trace-isolate which can produce this data. Perhaps unsurprisingly, countries with strong public health funding and which have implemented test-trace-isolate strategies have often the lowest case-fatality rates, one such example being South Korea.

Other nations have not done the same. Countries have knowingly foregone testing despite a number of health professionals strongly advocating for this strategy: among these including the Chair of Global Health at the University of Edinburgh, the Editor of The Lancet, and the Director General of the World Health Organization.

Beginning in late February, 2020, governments have relied upon mathematical-epidemiological models with exponential estimates and all have included the preface that these models are based upon assumptions which may not be true. None of these models have estimated the social, health, or economic damages of any of the measures they recommend (i.e. social distancing’s impact on one lost day of work productivity). These models have included suggestions that half of the UK population is infected with COVID-19[ix], that nearly a million people will die in the UK if social distancing policies are not implemented[x], or that there may be up to two million people hospitalized because of COVID-19 in India.[xi] At one point, a model in the United States adjusted for the confounding factor that people would not trust the model.[xii]

I want to highlight three units of measurement to understand the technologies of governance of COVID-19: i) incidence, ii) case-fatality rate, and iii) the basic reproduction number.

Measuring COVID-19 Incidence:

Caduff has pointed out how most nearly 50% of people do not experience symptoms and will not become confirmed cases[xiii]. The ability of a patient to qualify for a test often varies by country, frequently requiring recent travel or direct contact with a COVID-19 patient, and many are not able to be tested until they are admitted to a hospital. Even if systematic testing were to be done there are i) technical and ii) social causes which continue to create unknowns. Within the technical realm, Street and Kelly have pointed out the unreliability of ‘the test.’ Testing relies upon “supply chains and some form of training” which are highly variable characteristics across different health systems. Further, if a test is successfully administered false negatives are common in the early stages of the pandemic because “there is not enough genetic material for the test to detect.” Street and Kelly describe the technology of the test as a tool which produces ‘unknown unknowns.’[xiv] The lack of testing and the inaccuracy of the test means decision makers don’t know what they don’t know; the data that has been collected is both incomplete and uncertain. There are social drivers of measuring COVID-19. In the United Kingdom, incidence (and mortality) rates rise by nearly 20% every Monday, after reductions on Saturday and Sunday; as put poignantly by a Financial Times columnist, less people work on the weekend, and less testing is done so the incidence is lower.

In light of not having comprehensive data epidemiologists are tasked to produce incidence predictions. The data these models rely upon produces different outcomes based upon the method of analysis applied. On April 6th, 2020, one method predicted cases in the UK are decreasing, another predicted cases in the UK are increasing. Both models were statistically significant. As Street and Kelly have pointed out, when epidemiologists can produce near exact opposite results, we may really not know what we don’t know.

Measuring COVID-19 Case-fatality rate (CFR):

CFR is calculated by dividing the number of deaths by the number of incident cases of COVID-19. However, as detailed above, the number of cases is unknown. As Caduff has discussed, the characteristics of the virus’s 2-week incubation and procedures of testing policy inflate the case-fatality ratio. This drastic uncertainty can be represented by the model released by the Institute of Health Metrics and Evaluation on April 6th, 2020, which suggested 60,000 fewer deaths in the United States would occur than was originally predicted two weeks before.[xv] The embedded uncertainty and changing numbers of CFR in COVID-19 suggest something else is at stake in the state’s perception of risk than life and death; risk is not about the CFR, but hospital bed capacity.

Measuring COVID-19 Basic Reproduction Number (R0):

R0 is a variable which indicates for each person infected with SARS-CoV-2, how many more people will contract the virus. If this number is above 1, the virus is spreading. This number can allow epidemiologists to predict future incidence when there may be little data from testing if there is a base population of incidence; however, to do this a number of assumptions have to be made. An R0 is calculated by i) observing SARS-CoV-2 in a laboratory setting to understand the rate of reproduction; ii) assuming a number of social factors until they can be epidemiologically proven including how many people the average person contacts within a day or personal protection measures taken by an infected individual (i.e. wearing a mask); and iii) assuming a number of biological factors about the virus until they can be epidemiologically proven including the length of exposure required to contract the virus and the length of incubation of the virus. In other words, to calculate R0 accurately, incidence must be measured accurately because R0 invariably varies by social context.

R0 has become relevant for COVID-19 governance because it has worked to justify a number of policy responses. For example, Public Health England spoke on April 7th, 2020, that social distancing is the best strategy because it brings R0 below 1, and because of this contact tracing is not indicated.[xvi] By making this claim, the state reveals a logic of governing through unknowns. The state claims that R0 is truth but denies the opportunity to produce any evidence that would verify R0. Further, the (nearly unsubstantiated) truth that R0 is below 1 is turned back on itself by the state to establish the logic that contact tracing would not be cost-effective. Thus, it is cyclical: the way we have chosen to ‘know’ R0 has ensured that we can never prove R0. Epidemiologists and public health experts have chosen to not know the relationship between virus and human — between rate of transmission and social context — and as such are governing the unknowns.

So, can this be a biopolitics? Certainly, there is something bio about this. Nearly the whole world has been put on lockdown because of the plausibility that millions will die. But what type of measurement is this? It is not a comprehensive measurement of the population: (most) governments refuse to test the entire population. Nor is this a measurement of the population with some excluded parts left to die as Biehl has suggested: for the most part COVID-19 victims have been dying in hospitals (so far — although the excluded are certainly relevant and worryingly overlooked as pointed out by many others both within and beyond COVID-19). Nor is this an intentionally inaccurate measurement as Gupta suggests: epidemiologists (and the politicians they influence) very much believe their numbers to be highly probable and the phrase ‘we don’t know’ is not to be muttered (especially publicly) within these epistemic circles. Nor is this a measurement of the virus in relation to particular social lives as Latour and Paxson discuss: all data is immediately extrapolated out to attempt to be seen as nationally and globally comprehensive. If life is at stake, what does it indicate that governments have chosen to produce measurements of uncertainty claimed-truth? To understand this, we first must work to deprovincialize life as the principal subject of governance; the anthropocentric lens of biopolitics obscures the agency of the virus.

Theories of Government of the Non-Human:

This is not the first time that governments have chosen to govern with limited and paradoxical data, it is just the first time that data could be collected but is not; because of that effects are taking place at a population wide scale in real-time. In his writings about pandemic preparedness, Lackoff suggests that the notion of unpreparedness is dependent upon “the event whose probability cannot be calculated.”[xvii] To overcome unpreparedness but never become prepared, states become responsible for not investing in their populations but in their vital systems (for example, respirators and hospital beds). Thus, the state generates “knowledge about internal system-vulnerabilities” rather than populations. In one sense the measurement of vital systems is taking place in assessing hospital bed capacity for COVID-19. In another sense, this is very different from what Lackoff describes, because the pandemic is here, and the technologies of governance have chosen to continue to measure vital systems capacity rather than attempt to produce more rigorous epidemiological evidence. Other than temporality, the pandemic in the present allows us to understand how governance will (or not) protect these vital systems. Globally there have been calls for governments to provide hospitals with respirators and health care workers with personal protection equipment; most of these calls have gone unanswered. Instead of funding these vital systems, the responsibility has shifted to the citizen which is expected to stay at home. Even if governments would fund the vital system, unreliable incidence and mortality data is unable to indicate how much to fund. So we can again ask ourselves: how is this neoliberal responsibilization governed?

For Elizabeth Povinelli, the source of power for biopolitical technologies of governance often relies upon the ability of biopolitics to create a “self-evident distinction of Life and Nonlife” whereby the stakes of government are then exclusively between life and death and excludes consideration of Non-life.[xviii] As Povinelli exactly points out, “the division of Life and Nonlife does not define or contain the Virus … the virus copies, duplicates, and lies dormant even as it continually adjusts to, experiments with, and tests its circumstances. It confuses and levels the difference between Life and Nonlife.”

A pandemic produces a potential for confrontation between the imagined human stakes of life vs. death and the virus’s stakes of life vs. nonlife. In the pandemic, the virus can no longer be controlled through measurement as in Paxson’s microbiopolitics. The virus has arrived and is dangerous; but contrary to Paxson, the virus is largely unknowable. However, some aspects of the virus could be known. As I suggested above, there is an active choice not to know the virus. By not implementing community-wide testing policies, properties of the virus including the R0 cannot be confirmed. By choosing not to know the virus, perhaps this evades confronting the stakes of life vs. nonlife.

Caduff attributes this property of the virus’s unknowns to the continuously evolving “cosmology of mutant strains … [which has made] the unknown possible as an ontological given.”[xix] In other words, even if more properties of the virus were known, the virus may always remain unknown because it can change. To Caduff, the unknown becomes an ontological given; in other words, it will never be disputed that the virus cannot be known. 

What I am attempting to suggest here is that there is a lot at stake for governments in not knowing the virus. It is not populations nor is it vital systems that governments are managing; this is a governance of the virus’s unknowns. To govern the unknowns of the virus, governments must maintain the ontological given that the virus can never be known. The COVID-19 pandemic is thus a clash. The virus could be known; if the virus was known, would late-liberalism’s underlying distinction between Life and Non-life be faltered, and if so, what would happen? This is the question governments are trying to avoid. By not testing the population, the distinction between Life and Non-life is denied and the ontology of the virus as unknowable is maintained. As Caduff observes, when “risk assessment is inconclusive … [and] truth is suspended … authorities [are required] to consider the worst imaginable case as the most likely scenario.” In Caduff’s work on bio-preparedness research he observed how considering ‘the worst imaginable case as the most likely scenario’ produced a logic for multi-billion-dollar investments into the biomedical industry. Today, by maintaining the ontological given that the virus is unknowable, the entire government apparatus can ‘consider the worst imaginable case as the most likely scenario.’ Importantly here, by not knowing the virus, the worst imaginable case is exactly that: imaginable. Whatever someone can imagine is possible. By not-knowing, we can now imagine, and in these sites of imagination we are seeing the contemporary forces of biopower being strengthened, as well as new sites being formed. Remarkably, this time the population is nothing but imagined.

Where to?

I have tried to open up a discussion of what governmentality, and specifically population measurement, looks like in times of COVID-19. But what avenues can we take with this way of thinking. I want to conclude on two suggestions. Frist, returning to Foucault, Caduff and Rabinow have discussed how the state as a security apparatus takes the object of population as a formed entity and becomes interested in modulating external events of things like the market and the potential (now present) pandemic.[xx] Second, another productive avenue of thinking may be through James Scott’s work in Seeing Like a Sate.[xxi] Scott suggests that states need to see certain things to be able to govern; perhaps, here, we can ask what the state chooses to see and not see. Further, what new practices of state seeing are being required to modulate the security apparatus, and what contradictions will this bring to light of past, present, and future state practices?

Robert D. Smith is an MA candidate in Anthropology and Sociology at the Graduate Institute of Geneva. He is interested in primary health care in the treatment of non-communicable diseases in Delhi, India, and more broadly practices of public health policy prioritization.


I would like to thank Purbasha Mazumdar, Carlo Caduff, Dorota Kozaczuk, Bryan M. Dougan, Anthony Rizk, Aditya Bharadwaj, Pooja Sharma, Mohandas Mallath, Kawahya Tizhe, Daniel Villamarin, and the Critical Theory Reading Group at the Graduate Institute of Geneva for earlier comments on this paper as well as stimulating conversations which have shaped my thinking about what may only be best described as ‘unique’ times. None of these teachers, fellow students, or friends are responsible for the opinions raised in this paper.


[i] Billaud, Julie. 2020. “Coronavirus: Moins D’Humanitaire, Plus De Politique!”. Le Temps.

[ii] Clover, Joshua. 2020. “The Rise And Fall Of Biopolitics: A Response To Bruno Latour”. In The Moment.

[iii]  Foucault, Michel. 1976. The History Of Sexuality.

[iv] Armstrong, David. 1983. Political Anatomy Of The Body. Cambridge: Cambridge University Press.

[v] Latour, Bruno. 1993. The Pasteurization Of France. Cambridge, Mass.: Harvard University Press.

[vi] Paxson, Heather. 2008. “Post-Pasteurian Cultures: The Microbiopolitics Of Raw-Milk Cheese In The United States”. Cultural Anthropology 23 (1): 15-47. doi:10.1111/j.1548-1360.2008.00002.x.

[vii] Biehl, João. 2013. Vita: Life In A Zone Of Social Abandonment. Berkeley, Calif. [u.a.]: Univ. of California Press.

[viii] Gupta, Akhil. 2012. Red Tape. Durham (NC): Duke University.

[ix] Lourenco, Jose, Robert Paton, Mahan Ghafari, Craig Thompson, Peter Simmonds, Paul Klennerman, and Sunetra Gupta. 2020. “Fundamental Principles Of Epidemic Spread Highlight The Immediate Need For Large-Scale Serological Surveys To Assess The Stage Of The SARS-Cov-2 Epideic”. Medrxiv. doi:

[x] Verity, Robert, Lucy C Okell, Ilaria Dorigatti, Peter Winskill, Charles Whittaker, Natsuko Imai, and Gina Cuomo-Dannenburg et al. 2020. “Estimates Of The Severity Of Coronavirus Disease 2019: A Model-Based Analysis”. The Lancet Infectious Diseases. doi:10.1016/s1473-3099(20)30243-7.

[xi]  The Center for Disease Dynamics Economics and Policy, and Johns Hopkins University. 2020. “COVID19 For India Updates.”

[xii] Washington Post. 2020.

[xiii] Caduff, Carlo. 2020. “What Went Wrong: Corona and the World after the Full Stop.”

[xiv] Street, Alice, and Ann Kelly. 2020. “Counting Coronavirus: Delivering Diagnostic Certainty In A Global Emergency”. Somatosphere.

[xv] IHME. 2020. “COVID-19 Estimation Updates”. Institute For Health Metrics And Evaluation.

[xvi] Cosford, Paul. 2020. “RSM Live Webinars | The Royal Society Of Medicine”. Rsm.Ac.Uk.

[xvii] Lakoff, Andrew. 2008. “The Generic Biothreat, Or, How We Became Unprepared”. Cultural Anthropology 23 (3): 399-428. doi:10.1111/j.1548-1360.2008.00013.x.

[xviii] Povinelli, Elizabeth A. 2016. Geontologies. Duke University Press Books.

[xix] Caduff, Carlo. 2014. “Pandemic Prophecy, Or How To Have Faith In Reason”. Current Anthropology 55 (3): 296-315. doi:10.1086/676124.

[xx]  Caduff, Carlo, and Paul Rabinow. 2007. “Security Territory Population”. Anthropology Of The Contemporary Research Laboratory ARC Concept Note (8).

[xxi] Scott, James C. 1999. Seeing Like A State: Yale University Press.