[This is an advance copy of an article that will appear in print in September 2020 as part of the KIEJ’s special double issue on Ethics, Pandemics, and COVID-19.]
ABSTRACT. Are lockdown measures ethically justified? This paper outlines some of the key issues relevant to answering that question, paying particular attention to how decisions are framed. Section 1 argues that ethical reasoning about lockdown ought to be guided by a distinction between prudential and ethical reasons, grounded in a concern to respect the separateness of persons, but also that—as public health messaging implies—it can be unclear whether measures are in individuals’ prudential interests or not. Section 2 suggests that a similar set of problems affect attempts to adopt alternative cost-benefit-analysis frameworks for assessing lockdown. Section 3 suggests an answer to these shared problems: we need a process for determining when wellbeing claims and systems of categorization are ethically apt. Section 4 argues that settling the question of aptness is our key ethical task in assessing lockdown.
In many countries and regions across the world, the initial response to the massive health risks posed by COVID-19 has been the institution of lockdown measures. Although they vary from place to place, these measures all involve trade-offs between ethical goods and imperatives, imposing significant restrictions on central human capabilities—including citizens’ ability to work, socialize, exercise democratic rights, and access education—in the name of protecting population health. As such, it seems imperative for philosophers to ask whether lockdown measures are ethical.
This is a difficult question. We are deeply uncertain about the possible effects of COVID-19 and of lockdown policies, raising questions about ethical decision-making under uncertainty. We are reliant on models that themselves embed value judgments (Schliesser and Winsberg 2020). Debates over lockdown are invariably highly politicized, with “lockdown sceptics” lumped in with Trump and Bolsonaro, and framed in philosophically unhelpful terms of “following the science” (Dupré 2020). It is hard to know which ethical frameworks to use to guide our assessments: should we think in terms of rights (and if so, which) or in terms of consequences (and if so, how)? Is this the kind of “emergency” situation, where rights must give way to consequentialist concerns (Sorell 2013)? There have been excellent rapid ethical analyses of COVID-19-related issues, such as privacy concerns about tracking apps (Schaefer and Ballantyne 2020), or principles for vaccine research (London and Kimmelman 2020). However, there is a mismatch between the topics we are equipped to discuss and the topic that most affects most of us. I have seen rich discussions of how to distribute scarce ventilators (Liddell et al. forthcoming), but very little on how to distribute the costs of lockdown. My aim in this paper, then, is to explore some core issues in assessing the ethics of lockdown.
Before going on, a note on my general argumentative strategy may be helpful. There are, of course, multiple frameworks for assessing the ethics of public health interventions: for example, Nancy Kass’s (2001) six-stage analytical tool for assessing interventions; Childress et al.’s (2002) nine moral considerations for assessing interventions; the UK’s Nuffield Council on Bioethics “intervention ladder” (Nuffield Council on Bioethics 2007); and so on. One obvious strategy for addressing my question would be to use one of these frameworks to list and assess the ethical challenges of lockdown. That is not my strategy. Rather, I want to point to a concern that arises before we use some framework. We apply frameworks to descriptions of the world, but how we describe the effects of lockdown itself involves making ethical decisions. That is to say, the inputs to our ethical reasoning are already partially “value-laden,” and so we need an account of when the descriptive inputs to reasoning are themselves ethically proper. To make this point as clearly as possible, then, I will largely leave conventional public health ethics frameworks to one side until Section 4, and, instead, sketch problems in the ethics of lockdown in general terms. This has some disadvantages, in that my arguments do not obviously map onto normal frameworks, but, I hope, the great advantage of clarifying some of the fundamental ways in which COVID-19 requires us to rethink public health ethics.
Specifically, I focus on the following problem: on the one hand, it seems that the ethical analysis of lockdown measures must take account of how their costs and benefits are unequally distributed across the population; on the other hand, the pandemic destabilizes familiar notion of “costs” and “benefits,” and familiar ways of dividing up the population to measure inequality. In Section 1, I set up this problem through a study of an apparent disconnect between the ethics of lockdown and how lockdown measures have been justified. In Section 2, I consider a way through this impasse: by thinking about the ethics of lockdown in terms of cost-benefit analysis. Unfortunately, however, consequentialist approaches face problems strikingly analogous to those explored in Section 1. In Section 3, I draw out a lesson from Sections 1 and 2: that the “inputs” to our ethical reasoning must, themselves, be ethically justified, and suggest a way of approaching that task. Section 4 sketches the implications of my arguments for thinking about both the ethics of lockdown and its communication.
1. JUSTIFIABILITY, JUSTIFICATIONS AND THE SEPARATENESS OF PERSONS
During the COVID-19 pandemic, my children are at home from school and nursery, isolated from their friends, complaining that they can’t go to the park. During the COVID-19 pandemic, my parents are also at home, missing out on long-planned holidays and social events. Social distancing is hard for both the young and the old. But there’s an important difference. The data suggests that, were my children to get infected, they would very probably suffer little more than a sniffle; their symptoms would be no worse than those of the common cold they pick up in the playground. By contrast, were my parents to get infected, they could end up very ill indeed (Spiegelhalter 2020). Social distancing has costs for both my parents and my children, but benefits my parents in a way in which it doesn’t benefit my children. In the first part of this section, I argue that this difference is ethically relevant, with apparent implications for how we think about current public health messaging. In the second half, however, I suggest a twist: claims about what is in my children’s or my parents’ interests are malleable.
The fact that my children don’t seem to benefit in a prudential sense from lockdown doesn’t necessarily imply that schools and nurseries should remain open. Children can spread illness throughout the community, risking others’ health. In turn, there are familiar ethical arguments for restricting individuals’ rights when exercise of those rights threatens to harm or infringe the rights of others (Wilkinson 2007). What it does mean, however, is that my children and my parents have different kinds of reasons to adhere to lockdown measures. My parents have reasons grounded in their own health; my children have reasons grounded in others’ health. I will now suggest this distinction should be central to the ethics of lockdown.
To explain why, let me take a step back. Imagine that you wake up after a mystery illness. The doctor at your bedside tells you that, while you were unconscious, she removed one of your kidneys, but, she explains, this was the only way in which to save your life. Imagine, now, a twist: you wake up to discover one of your kidneys has been removed, but this time the doctor tells you that this was the only way of saving a stranger’s life. In both hypothetical cases, you have suffered a harm—non-consensual removal of an organ—which has resulted in a valuable good—a life has been saved. However, from an ethical perspective, the cases seem very different: there’s a difference between harming you to save your life and harming you to save a stranger’s life. Philosophers often capture this difference by distinguishing between intra-personal justifications—which appeal only to yourbest interests—and inter-personal justifications—which balance your interests against others’ interests (Thompson 2018). The intra-/inter-personal distinction is central for understanding the limits on broadly utilitarian approaches to ethical reasoning. As is well-known, such approaches can lead to deeply counter-intuitive results; for example, that there is nothing wrong with killing one person to redistribute her organs to save five. In a memorable phrase, John Rawls claimed that these counter-intuitive results stemmed from the fact that Utilitarianism denies a deep normative fact, “the separateness of persons” (Rawls 1971). Otsuka and Voorhoeve (2009, 179) provide a more prosaic, but helpful way of understanding the underlying problem here:
A single person has a unity that renders it permissible to balance (expected) benefits and burdens against each other that might accrue to her. A group of people, by contrast, does not possess such unity. As a consequence, some forms of balancing benefits and burdens that are permitted when these accrue to a single person are impermissible in cases where these benefits and burdens accrue to different people.
Not only is the distinction between intra- and inter-personal justifications central to large-scale debates in normative ethics, but it is also important to the practical task of thinking about public health ethics (Thompson 2018). Although medical interventions and public health interventions both aim at improving health, they differ in important respects. One is that, typically, an individual can choose not to have some medical procedure, but, typically, individuals lack any choice as to whether they are affected by some public health program. Normally, you can refuse an operation, but you cannot opt-out of clean air or health-and-safety regulations. As such, notions of choice and consent cannot play the kinds of justificatory role they typically play in medical contexts (Wilson 2011; Verweij and Dawson 2013). What, then, can justify public health interventions? One tempting way to think about this problem is to say that the core question for assessing public health policy is whether individuals would or should consent to interventions; that is to say, whether our interventions are justifiable to each.
I claim, then, that we can understand the concept of justifiability as foundational to many debates in public health ethics. In turn, there seems to be an important distinction between two kinds of interventions: those which are intra-personally justifiable to each; and those which are intra-personally justifiable to some, but not to others (Thompson 2018). Plausibly, fluoridating the water supply might fall into the first category. While fluoridation imposes a small risk on each of dental fluorosis, each also gains from a substantial decrease in the risk of dental cavies. Each “gains” more from fluoridation than she “loses.” By contrast, redistributing wealth to lower health inequalities falls into the latter category; diverting resources from richer to poorer areas may improve the expected health outcomes of the poor, but worsen those of the rich. Colloquially, some policies have only winners, whereas some policies have both winners and losers.
Note that the point of distinguishing between “only winners” policies and “winners and losers” policies is not to imply that only the former kind of intervention is justified. (For what it’s worth, I think that radical forms of redistribution may be ethically justified, even if they have losers.) Rather, the point of the distinction is to keep track of the kinds of arguments and justifications that will be required for justifying interventions. In cases that have only winners, we can say that policies are justified to each in solely prudential terms. In cases where there are both winners and losers, we need to identify ethical reasons why these policies are justifiable to the losers; for example, we might need to justify redistributing resources on grounds of equity or justice.
Before going on, it may help the reader to take a step back. One core debate in public health ethics concerns the government’s ethical warrant to pursue and promote public health policies: is it, for example, because promoting health promotes aggregate wellbeing (Bayer and Fairchild 2016) or because public health policies promote justice by securing individual health, an important pre-condition for equality of opportunity (Daniels 2007)? My concerns above imply an obvious worry about the first sort of justification for public health policies (which I explore more fully in Section 2 below): that promoting aggregate health assumes that the population has a kind of unity. Nonetheless, strictly, my concern with justifiability in this section is orthogonal to these debates. My claim is that, regardless of the proper ends of public health, it is important to ask of specific health policies whether they can be justified to the individuals whose capabilities they affect. (By analogy, regardless of why a physician should operate, it is important that the patient gives consent.) In thinking about this question, I suggest that we should distinguish between intra- and inter-personal justifications. Different accounts of the proper grounds for public health policy can be seen as, in part, different accounts of the contents of inter-personal justification; for example, we might seek to justify lockdown measures inter-personally by saying that they are necessary to secure equal opportunity for all. (Or, indeed, we might justify these constraints without any appeal to “public health” at all, but as justified by basic ethical obligations not to risk harming others.) Strictly speaking, then, my claims about “justifiability” are separate from debates about the ends of public health policy; rather, they are a reminder that any policy, however noble, must always be justifiable to those it affects.
Let me return, now, to lockdown measures. I suggested at the start of this section that lockdown measures seem to be in the prudential interests of some—the elderly, those with underlying health conditions, and so on—but not others—the young. Therefore, we cannot use the same justifications for the young as for the elderly. However, I now want to suggest that matters are even more complex. To do so, I will turn to a related question: how do governments actually justify lockdown?
In the UK, we spent all of April 2020 constantly told that it was of the utmost importance that we “stay home, protect the NHS, save lives” (Hope 2020). A striking feature of this messaging was that, while there was explicit acknowledgement that these measures were costly to individuals, and required great sacrifice, there was very little acknowledgment that different sectors of society might have very different reasons to follow this advice. Rather, COVID-19 was presented as a threat to each of us and lockdown as the response to this shared threat; while we each had to pay a lot, we were paying for something that was, on balance, good for each of us. Perhaps unsurprisingly, then, it seems that many members of the public grew highly fearful of their own risks of suffering from COVID-19, far beyond any objective estimate (Spiegelhalter 2020). Of course, there are many determinants of public attitudes and beliefs, and some aspects of policy did acknowledge risk-differentials, but it seems that creating a sense that lockdown was in the prudential interests of each was a deliberate communications strategy. Recent press reports have suggested that senior politicians are now concerned that their messaging was “too effective,” because many people are now overly concerned about their risk of returning to study or work (Hope 2020). In short, there is an apparent gap between the ethics of lockdown—a policy with both “winners” and “losers” —and how we communicate lockdown, as having only “winners.”
Of course, public health messaging simply cannot always provide everyone with the most accurate account of their prospects. For example, an anti-smoking campaign can hardly take account of all of the slightly different ways in which each might respond to cigarette tar. Still, it seems deeply dubitable for a public health campaign to treat a policy that must be justified to some in inter-personal terms as if it is justified to each in intra-personal terms. The problem here is not just that messaging ignored some differences between individuals, but that it masked an incredibly ethically significant distinction; it would be like removing your kidney to save a stranger’s life while telling you the operation was necessary to save your own.
In turn, this concern seems separable from the question of whether lockdown is, in fact, justifiable. Assume that lockdown is justified. Given this assumption, there are clear reasons why the government might wish to paper over issues such as the inter-generational trade-offs involved, as a way of motivating compliance. (Arguably, these concerns might be particularly strong in the UK, suffering from inter-generational conflicts over Brexit.) Maybe blurred messaging was the only effective means for a good end. Still, even making these assumptions, it seems problematic to present the policy as if it only has winners. Justifying such presentation in terms of its consequences seems uncomfortably close to what Williams (1995) called Government House Utilitarianism where the state treats its citizens as if they were to be tricked into acting properly.
So far, so bad for public health messaging. Unfortunately, my arguments face a challenge: is it clear that lockdown is not in the interests of each? Consider, again, the claim that staying at home is not in my children’s best interests, and, so, not intra-personally justified. This claim consists of two elements. First, claims about which sorts of outcomes (going to school, getting sick) are good or bad for my children (what I will call “wellbeing”). Second, claims about the probability of those outcomes. There is always some possibility that my children might contract some deadly illness at school; whether going to school is in their best interests depends on how likely that outcome is. Both claims about wellbeing and claims about probability are, though, contestable.
First, consider the mantra “stay at home, protect the NHS, save lives.” Does “protecting the NHS” promote my children’s wellbeing? Having a functioning NHS provides them with an assurance of care if they fall sick, but, also, they like the idea that others won’t suffer unnecessarily. The first concern seems clearly to promote their prudential wellbeing, but what about the second? On the one hand, their concern for others’ health is clearly not a concern for their own health. On the other, it’s not obvious that we should restrict our account of prudential interests to the concerns of sociopaths. Indeed, one might think that one of the few positives of a pandemic is precisely that, in making us aware of our inter-dependency, it fosters a broadening of our sense of what is good for us, beyond our “interests” narrowly conceived. If people feel miserable at the thought of others’ suffering, why not count the avoidance of that misery as promoting their wellbeing? That’s a rhetorical question, not an argument; there are countervailing considerations here (as far back as Sen 1970). My point isn’t to settle on the one true account of wellbeing. Rather, it is to show that it is not wildly implausible to expand an account of my children’s interests to include a concern for others.
Perhaps surprisingly, something similar is true of claims about the second element of “best interests” judgments: probability claims. All of our data about the probability that individuals will suffer some illness, such as COVID-19, is, ultimately, indexed to their membership of some population for which we have collected statistics. “Jim is at low risk of COVID-19” is, in effect, shorthand for a longer, more complicated claim, along the lines that “Jim is under 20, and the proportion of cases among under-20s is relatively low.” Unfortunately, any individual falls into many different sub-populations. For example, Jim’s risk of COVD-19 may be higher if we categorize him as a man, even higher again if we categorize him as someone who is immuno-suppressed, and so on. Which of these ways of describing Jim is “correct”? Unfortunately, there isn’t necessarily a single figure, “Jim’s true risk” out there, akin to, say, “Jim’s true blood pressure.” Rather, to a large extent, the answer is “it depends on what you’re doing”; for some purposes, it may be best to calculate Jim’s risk as a man, but for other purposes, it may be best to calculate his risk as someone who is immuno-suppressed (Hajek 2007; Fuller and Flores 2015).
Before going on, it is worth noting a tricky issue here: the problem above is a version of what philosophers of statistics call the “reference class problem.” Reichenbach famously set out an influential solution to that problem: that for purposes of calculating the probability of a one-off event, such as Jim contracting COVID-19, we should base our estimate on the narrowest available reference class for which reliable statistics can be compiled (Reichenbach 1949). Even placing to one side various technical worries about Reichenbach’s solution, it isn’t much help in our case. It is implausible to think that government messaging around a pandemic could ever provide each individual with an estimate of her risk relative to the narrowest available class. The government cannot send a letter to each individual giving her a “personalized” risk score (indeed, one of the most fascinating aspects of this pandemic is that it shows the limits to the hype around “personalized medicine”). Rather, the government’s public claims must assume coarse, wide reference classes. If so, it is hard to find a knockdown objection against talking as if everyone’s risk is indexed to the widest available reference class—the UK population—rather, than, say distinguishing the “young” and the “old.”
Here, then, is a general worry: claims about what is in an individual’s “best interests” —and, hence, the distinction between intra- and inter-personal justification—seem to rely on claims about wellbeing and claims about risk. However, both kinds of claim are contestable, malleable, and indelibly indexed to our interests. As such, we might be able to justify treating lockdown measures as in the interests of each by adopting capacious accounts of wellbeing and using very wide reference classes. Note that this is not to say that claims about interests are entirely malleable, completely unconstrained by the world. Simply saying that my children’s wellbeing is promoted when the rich get richer doesn’t make it true on any plausible account of wellbeing; simply saying that Jim has a high risk, regardless of the statistical data, doesn’t give him a high risk. My claim is that facts about the world, including facts about how we conceptualize wellbeing, underdetermine claims about our interests, rather than that they place no constraints at all.
Still, even with this caveat, the malleability of interests claims suggests a problem for the argument sketched in the first part of this section. I suggested that government messaging was clearly wrong to imply that lockdown measures benefit each of us; such messaging is, at best, a trick to get us to behave as we ought to. However, there may be accounts of our interests and ways of categorizing the population such that lockdown measures are, apparently, in the interests of each. In effect, we can turn what may seem a genuinely knotty ethical problem about justifying an intervention that has both winners and losers into the relatively easier problem of justifying an intervention that has only winners.
We face a problem. It seems that we have excellent reasons why our ethical reasoning in general, and hence our responses to COVID-19 specifically, should respect the separateness of persons. In turn, this concern gives us good reasons to be suspicious of public health messaging that seems to imply that we are all in this together. In a pandemic, however, it is natural to question the distinction between persons, as we all learn to think of ourselves as nodes in a larger framework of infection, and as we recognize the costs of being socially-distanced from friends and family. Pushing these thoughts, we can apparently legitimately redescribe our normative problem, such that the apparent clash of interests between my parents and my children, and the associated worries about public health messaging, simply disappears. Given the apparent instability of claims about interests, how can we even start to go about thinking about the ethics of lockdown?
2. COSTS AND BENEFITS
It has become standard to distinguish two broad ways of thinking about interventions that involve distributing costs and benefits across the population: through a broadly contractualist lens, focused on justifiability to affected individuals, and through a broadly consequentialist lens, focused on the (expected) aggregate costs and benefits of action (Suikkanen 2020). One way of understanding the previous section is as suggesting a serious problem for thinking about public policy—at least, this kind of public policy—in terms of justifiability. Doing so seems to require an unstable distinction between intra- and inter-personal justification. Beyond this simple argument from elimination, there are more positive reasons to adopt a broadly consequentialist approach to assessing lockdown measures (and their communication). By their very nature, public health policies operate at a population level, in ways that makes it next to impossible to identify individual beneficiaries (Cohen, Daniels, and Eyal 2015). For example, we might be able to say that the early and effective lockdown measures instituted in Germany saved many lives, but permanently unable to point to any individual German citizens and say “her life was saved.” Preventive public health policies such as lockdowns always deal in the currency of statistical lives. The core epistemic unit of public health policy is the population; it seems plausible that our ethical reasoning must follow suit.
It should, therefore, be no surprise that most of the discussions of the ethics of lockdown measures have been framed in terms of cost-benefit analyses (CBA). For example, Peter Singer and Michael Plant (2020) have suggested that key to assessing lockdown measures is ensuring that the “cure” is not worse than the “disease.” The key challenge for academic researchers, they suggest, is to calculate lockdown’s effects, “not in terms of wealth or health, but in the ultimate currency, wellbeing”; if the costs to wellbeing outweigh the benefits, then lockdown measures are ethically wrongful. Alex Broadbent (2020) has argued for a more specific claim, that “the cost-benefit analysis of [lockdown] measures yields a different result in Africa than in Europe, North America and large parts of Asia,” and, as such, “lockdown is wrong for Africa.” These moves have great appeal. There are difficult problems in assessing when policies are inter-personally justified; for example, even placing to one side all the complexities of the previous section, it is difficult to assess whether we can justify greatly restricting the capabilities of younger people for the sake of helping older people, given that, strictly, young people pose a risk,rather than a certainty, of transmitting COVID-19. And we can’t just ignore the arguments of the previous section, but must face the problem that best-interests-claims are indeterminate! By assessing policies solely in terms of their aggregate consequences, we seem to cut through this Gordian knot of problems.
Despite their appeal, however, there are well-known problems with assessing public health interventions, including our response to COVID-19, in cost-benefit terms. I will not run through all of them here. Rather, I will point to two problems of interest because of their relationship to the previous section.
The first problem concerns the calculation of “costs” and “benefits.” To make claims about the aggregate costs and benefits of lockdown, we must be able to say how good or bad outcomes are for affected individuals. In Singer and Plant’s terms, we must translate claims about consequences into the currency of wellbeing. Of course, in principle, there are thoroughgoing techniques for calculating the goodness and badness of outcomes, for example on the basis of revealed preference. However, as the previous section noted, these ways of calculating what is good for individuals are contested. To make matters worse, even if we have a resolution of these theoretical disputes, it is unclear how they relate to practice.
For example, in their article, Singer and Plant suggest that we might commensurate concerns about the effects of lockdown on health and on economic goods using a formula where each gain in a Quality-Adjusted-Life-Year (QALY) is valued at c£25,000 on the grounds that it is employed by the NHS in the UK. However, this seemingly simple way of commensurating health concerns and economic concerns is beset by controversy. Strictly, the NHS has a general policy of not funding drugs with an incremental Cost Effectiveness Ratio (iCER) over £30,000/QALY and routinely funding drugs under £20,000/QALY (Badano, John, and Junghans 2017). (Singer and Plant’s £25,000/QALY figure is a (telling) over-simplification here of the fact that between these figures “social values” are used to guide distribution.) This policy is justified on the grounds that this is the best way to maximize health outcomes, given a set budget; it is supposed to represent an “opportunity cost,” not the “value” of saving a life. Furthermore, this figure is, itself, highly contested, and, in practice, drugs are often funded even when their iCER is over £30,000; for example, when they treat end-of-life conditions (Badano, John, and Junghans 2017). And all of these concerns arise before we even get to the more fundamental question of whether QALYs are a sensible way of thinking about health-related-quality-of-life at all (Hausman 2015). This is not to say that a £25,000/QALY measure for doing CBA is clearly wrong; it is to say that any such choice is eminently contestable. Plausibly, these specific problems about Singer and Plant’s use of this figure generalize: there is no simple, uncontestable way of placing a value on human life (Wolff 2007).
The second problem is slightly more complex. Modern tools for cost-benefit analysis differ from Utilitarianism in their theoretical underpinning (Schmidtz 2001). However, they share a common feature: typically, they assess policies in terms of whether aggregate benefits outweigh aggregate costs, and as I noted in Section 1, assessing policies solely in terms of aggregate outcomes can lead to counter-intuitive results. Of course, most applications of CBA are bounded by broader ethical norms, such that their use does not threaten clearly mistaken results. Still, in our case, a focus on aggregate outcomes can seem to overlook ethically salient aspects of the pandemic. For example, straightforward application of CBA implies that the fact that the young “lose” from lockdown is simply irrelevant to the ethics of lockdown; all that matters is the overall balance of effects. In principle, then, a CBA of lockdown policies might justify imposing huge uncompensated costs on the young, just as long as doing so helped “enough” older people. These unequal distributions of burdens and benefits might seem unfair in themselves. Furthermore, they are even more worrying when they intersect with historical or contemporary forms of systematic oppression. In our case-study, for example, there are important concerns that COVID-19-related mortality rates may be far higher in areas of social deprivation (Rubin and Offit 2020). For proponents of CBA, this clustering of disadvantage is not ethically significant (even if it is useful to know for designing effective policies). However, this seems problematic; when we know that a high fatality rate among certain groups is, at least in part, a result of other, discriminatory, social policies, it seems we have good reason to pay particular attention to these groups’ claims in policy-making. Of course, there is scope for disagreement over the relevance of distributive questions in assessing health policy (Hausman 2007). Nonetheless, it seems problematic to insist, on the one hand, that policy-makers must be alert to the fact that policies involve trade-offs between different goods (say, aggregate health and aggregate wealth), yet to fail to take account of the fact that they involve trade-offs between distinct groups of people (say, young and old).
The proponent of CBA has two options in response to such challenges: either she can ignore them; or she can take account of distributional issues in some way. In Section 1, I suggested a very general ethical reason to follow the second route: that doing so follows from a concern to respect the “separateness of persons.” Moreover, concerns about distribution seem to play an important role in some appeals to CBA. For example, in developing his argument that lockdown measures are inappropriate in sub-Saharan Africa, Broadbent (2020) suggests that lockdown measures have been rapidly adopted because COVID-19 targets the privileged, asking, provocatively, “would we care about the increased risk of fatal pneumonia that COVID-19 might cause in Africa, if it did not also greatly increase the risk of fatal pneumonia for prime ministers, business people, and university professors?”, going on to suggest that “we don’t care about COVID-19 because of how many it kills, but whom.” Although Broadbent’s argument is couched in terms of aggregate consequences, these comments suggest that his underlying ethical concern is tied up with broader, philosophically distinct, worries about equity and social justice.
If they are to capture the full spectrum of ethical concern, then, cost-benefit analyses should be sensitive to distributional concerns. Broadly, there are two ways of doing this: by allowing distributional concerns to shape our answers to some of the puzzles about how to calculate “costs” and “benefits” (for example, by using life-years-saved, rather than lives saved as a metric for success); or by adding “equity weights” to our analysis of the goodness or badness of outcomes (Bobinac et al. 2012).
Either way, we face a problem. In assessing differences in the distribution of costs and benefits, we focus on differences between different categories of individuals. For example, in assessing lockdown, we might consider the likely unequal distribution of consequences for younger and older generations. However, in principle, there are multiple different ways we could categorize the population for purposes of ethical analysis. Rather than focus on young and old, we could assess policies in terms of whether they generate inequalities between different racialized groups, or inequalities between men and women, or between rural and urban areas, and so on (Bartley 2016). These analyses may lead to radically different assessments of the policies: focusing on inter-generational inequalities may make us skeptical of lockdown, whereas focusing on issues around social deprivation might make us more supportive. Just as I suggested that the world itself does not tell us which reference classes to use for purposes of calculating individual risk, there is no straightforward naturalistic reason to prefer one way of categorizing populations over another. We must make choices about which inequalities matter and, again, those choices are contestable.
Many countries have rushed into lockdown measures, justifying this approach on the grounds of a moral imperative to save lives. Clearly, these justifications are problematic insofar as they overlook the fact that lockdown measures might themselves “cost” lives, and have other, ethically significant costs in terms of on human wellbeing. Some of these costs are direct; for example, we know that economic hardship can have significant effects on life expectancy (Wilkinson and Pickett 2010). Other costs are indirect opportunity costs; for example, as resources are diverted away from other health and social needs. Clearly, any full reckoning of lockdown must take account of the full range of costs, as well as benefits. Versions of CBA supply our best-developed tools for doing such reckoning. However, we must be extremely careful in how we use and interpret those tools. First, any CBA is highly sensitive to how we count “costs” and “benefits” in the first place; and choices about how to do this are highly contestable. Furthermore, appeals to CBA face a second challenge: whether or not to take account of the distribution of costs and benefits, as well as their aggregate sum. If so, then they also rely on a second set of highly contestable choices, about how best to categorize the population. CBA may seem to cut through the ethical thickets, but only by sweeping away a series of questions.
3. THE GENERAL PROBLEM
Section 1 drew on the notion of the separateness of persons to suggest that the ethics of lockdown should be centered around the question of when interventions are justifiable. Section 2 explored an alternative: that we should simply consider the net consequences of lockdown. These are very different approaches. Oddly, however, both face a similar set of problems. Regardless of our preferred ethical framework for deciding what to do, our decisions must take claims about wellbeing and ways of categorizing the population as inputs. However, neither individual wellbeing nor categorization schemes are simply “out there” in nature to be discovered by our measuring machines (or, at least, if they are, we lack the right kinds of measuring machines). In describing our problem, we are already making contestable judgments about wellbeing and about how to categorize the population. Clearly, this is a very general problem. It is, however, particularly striking in the case of COVID-19; we cannot even get started on thinking about whether the trade-offs inherent in lockdown measures are justifiable if we can’t decide whether there are trade-offs at all. Should we just give up on thinking about the ethics of lockdown?
No. The problems above can be seen as merely a rather abstract way of reframing a familiar phenomenon, that real-life ethical debates are often less concerned with abstract ethical principles and more concerned with how to frame decision problems at all. To take a non-COVID-19 example, consider heated debates around the ethics of population screening for early stage cancer. One key question in these debates is how and whether our account of the possible costs should extend beyond the health-related harms of overdiagnosis and overtreatment to include the psychosocial harms of medicalization (Malm 1999; Marmot et al. 2013). No ethical conclusion automatically follows from saying that psychosocial outcomes are relevant to assessing population screening. Rather, decisions about whether or not to count these outcomes as “costs” are debates over how we should frame the ethics of screening in the first place. In turn, these debates are not simply settled by the world. Although empirical studies can show that screening has psychosocial harms, no study can show that we must measure or list those harms. Rather, to settle that question, we must draw on other, broader, technical, ethical and political considerations; for example, over whether these costs can be “adequately” well-measured to play a role in policy-making, how we should conceptualize the ends of medicine, and so on.
We have a problem—claims about wellbeing and systems of categorization need to be settled before we can even discuss our ethical problems—but also a sketch of a solution: we need to appeal to broader ethical and political concerns. How can we do that in the case of COVID-19, where we need to settle on an account of wellbeing, and on an account of how to categorize the population? There is no shortage of literature on how we should conceptualize “wellbeing.” A particularly interesting proposal has been made by Anna Alexandrova (2017), discussing work on the science of wellbeing (for example, in claims such as “beyond a certain level, increases in income do not improve wellbeing”). Alexandrova suggests two worries about such claims: “inattention” (that we might use controversial, value-laden terms without recognizing that we are doing so), which leads to a risk of “imposition” (that using these terms might lead us to overlook or deny or override others’ values). In turn, she suggests that the use of mixed claims can be justified only when underlying value assumptions are made explicit and deemed acceptable after consultation with affected parties.
Elsewhere, drawing on the example of stratifying the population by poly-genetic risk scores, rather than by age, I have suggested we need an “ethics of apt categorization” (John 2013). I suggested that, even if the poly-genetic scheme is scientifically sound and broadly in-line with the aims of a screening program, it may be problematic along other, broadly ethical dimensions: for example, that they might threaten to perpetuate past or on-going social injustice, and that they are not “intelligible” to those thus categorized.
Both of these specific proposals can be challenged. Nonetheless, they suggest considerations for thinking about the proper framing of the ethics of lockdown. Alexandrova’s concept of “inattention” provides a useful label for thinking about cases such as the £30,000/QALY threshold discussed in Section 2. Imagine we could show that, in the context of NICE’s resource allocation problem, given its goals and local, UK-specific ethical and political constraints, it is appropriate for NICE to use this threshold. It doesn’t follow from this that when considering lockdown measures in some other country we must use the same number to decide on whether lockdown is beneficial; lockdown measures don’t involve rationing problems, and local values may differ from UK values. In using this apparently “objective” number, we are, in fact, smuggling presuppositions about goals and values into descriptions of lockdown.
More generally, both approaches stress the importance of ensuring that ways of calculating wellbeing and of categorizing the population must be justifiable to those whose wellbeing we measure or whom we assign to certain groups. In this sense, they return to a key theme in Section 1, that we should focus attention on whether our interventions are justifiable. When we describe the effects of policies on people we are not directly intervening in the world, but we do know that those descriptions will guide interventions; as such, we need to ask whether our ways of describing things are justifiable to those we describe. Of course, in deciding how to frame the ethics of COVID-19, we cannot sit down and consult everyone—or even a sub-set of everyone—that what we are doing is acceptable. What we can do, however, is to draw on very general forms of knowledge, embedded in existing structures, to decide which accounts of wellbeing and which systems of categorization are likely to be justifiable.
First, consider wellbeing. When deciding to assess lockdown policies in terms of QALYs, we can give arguments as to why these measures should be used, even if they distort or mangle certain aspects of lived experience; for example, because they provide a useful, broadly neutral way of comparing interventions. By contrast, we might be far more nervous about construing individuals’ interests to include the “benefits” of living in a society where care is provided by all. This is not because we doubt that people do often care about living in such a society. Rather, it is because there are obvious dangers to including such concerns within our account of interests or benefits. Doing so threatens a kind of exploitation, where natural instincts of care are twisted into justifications for abuse. People can and do have ethically admirable motivations of solidarity, but it does not follow that, for purposes of ethical analysis, the warm glow of solidarity can unproblematically be treated as an aspect of prudential wellbeing. As such, we should, I suggest, reject the idea explored in Section 1 that living in a society where care is offered to the sick and ill is, itself, something that counts as promoting individuals’ wellbeing. Rather, we have good reasons to maintain a narrow sense of interests.
Second, consider the case of categorization, and the key question in Section 1: when we are thinking about the ethics of COVID-19, should we divide the population by age, allowing us to say that the old “benefit” more than the “young”? I suggest that categorizing the population this way is justifiable. First, the evidence for age-based differences in COVID-19 morbidity and mortality is very strong (Spiegelhalter 2020), and it is relatively easy to categorize people by age. Age is an epistemically robust category, which can be operationalized. Furthermore, we know that biological age correlates with a wide range of other normatively significant phenomena: for example, medical needs, social status, financial wellbeing, and so on. As such, using this category tracks a range of other, ethically significant concerns—such as concerns around economic precarity. Finally, we know that this category is “understandable”: that individuals’ self-identity is related, at least in part, to their biological age, such that the category maps onto (at least) some of their own reasons for action (John 2013). Therefore, there are strong reasons to think that it is justifiable to categorize the population by age.
Of course, none of these considerations show that we must categorize the population by age when considering the ethics of COVID-19. After all, this is a socially charged division, and one which carries with it risks of stigma and stereotyping, as, for example, when the young are accused of being “selfish” or the elderly are viewed as “drains on resources.” Therefore, using this category might risk creating or perpetuating forms of division and disadvantage. Nonetheless, it seems difficult to imagine how we might do health policy in a completely “age-blind” manner, given that age co-varies with so many other normatively relevant variables. Therefore, I suggest that we can justify using this category to think through the ethics of COVID-19. We should explicitly frame our decision problem in ways that acknowledge that lockdown affects different age-groups very differently.
The previous two sections set up a very general problem: thinking about the ethics of lockdown requires us to describe the likely effects of that policy. In turn, decisions about how to describe the effects of lockdown may implicitly reflect contestable value judgments: for example, saying that lockdown promotes feelings of social solidarity implies that such feelings are an ethically salient aspect of wellbeing; comparing the effects on the young and the old presupposes that we should care about equity between young and old. We could multiply similar examples beyond those I have discussed above: Should we treat a feeling of security as an ethically relevant effect of lockdown? Should we assess inequalities in the effects of lockdown on rural and urban dwellers? And so on. In this section, I have sketched a very general way of thinking about these problems. In turn, I have argued for a concrete claim: that we should think of lockdown as creating a problem of intergenerational justice, rather than as a policy which only has winners. My more general aim, however, has been more general: to suggest that we cannot do the ethics of lockdown without also thinking about how we describe that policy in the first place.
Even if we have some account of how to frame the ethical decision problem of assessing lockdown, we still have a further problem: how should we actually decide whether lockdown is justified? Furthermore, we still face a problem left over from Section 1: what is the relationship between an account of whether lockdown measures are ethically justified and the actual justifications we are given to adhere to those measures? In this concluding section, I take up these questions in reverse order.
Imagine that I am right and that our ethical reasoning about lockdown measures should be framed using an account of wellbeing and ways of categorizing the population such that the young are “losers” and the elderly are “winners.” Does it follow that it is wrong for the state to present measures asif everyone is a “winner”? This is tricky question, because an obvious corollary of my approach is that different accounts of “interests” might be appropriate in different contexts, much as the £30,000/QALY threshold might be appropriate when making resource allocation decisions at NICE, but inappropriate when thinking about the ethics of lockdown. Assume, then, that lockdown measures are, in fact, ethically justified, even though they have both winners and losers. Imagine further, however, that we know that the best way in which to get people to adhere to these measures is by presenting them as if they had only winners. Is there anything wrong in this scenario?
It’s complicated. We can and should distinguish between whether a policy is, in fact, justified and the reasons which we may legitimately give people to adhere to that policy. After all, what is important is that people do what they ought to do. In real life circumstances, where time is short, cognitive resources low, and emotions running high, trying to explain precisely why they ought to do those things may lead to ethically sub-optimal outcomes. Given our legitimate ethical ends, it may be permissible, for purposes of public health communication, to endorse an account of interests according to which lockdown is in the interests of each. To think otherwise requires an odd kind of moralism. However, having said that, as Section 3 argued, assessing whether policies are ethical requires that our framing of those policies are justifiable. One key way of ensuring this is through actual deliberation and discussion with those affected by our decisions. Blurred messaging—which presents our interests as if they are not in conflict—may block this important process. Furthermore, we run a risk of a form of inattention, insofar as it is easy to confuse the ways in which interests are presented in messaging with how they ought to be presented in ethical reasoning. Blurred messaging is not straightforwardly false or deceitful, but it does create significant epistemic and practical barriers to thinking properly about the ethics of lockdown.
What, though, about the first challenge, and the key motivating question of this paper: is lockdown ethically justified? Broadly, there are two ways of thinking about this question. One, discussed in Section 2, is to use the tools of cost-benefit analysis. A second, briefly mentioned in the Introduction, is to think through such policies using extant frameworks for public health ethics, which, typically, consider a wide range of different ethical principles and values. Perhaps disappointingly, I will not take a stand here on which of these approaches we should use. Rather, my concern has been to point to a very different sort of consideration: that, however we approach the ethics of lockdown, we need to do ethical work in deciding how to describe the effects of lockdown in the first place. In turn, I want to suggest that this process is an important, albeit easily overlooked site of ethical and political contestation. It is also a process that cannot be reduced to a question about which ways of describing the world have the best consequences, nor to the application of general ethical principles.
To illustrate these claims, consider two examples. First, consider the question of whether, when doing a cost-benefit analysis, of the sort discussed in Section 2, we should assess lockdown measures in terms of the number of lives lost or the number of life years lost. We cannot answer that question simply by looking it up in the Book of Nature. Nor, however, can we answer it simply by looking at the consequences of our decision; to do that would simply be to push the question back a level. Rather, that question must be addressed through a process of trying to figure out which measure is “apt,” where that process is essentially concerned with the question of which measure can best be justified to those affected by it.
Second, imagine that we seek to assess lockdown by asking how it relates to a mid-level framework for public health ethics, such as that proposed by Nancy Kass (2001). Kass lists six important considerations, covering a wide range of different ethical considerations: whether a program’s goals are proper; its means are effective; what burdens the program has; whether these burdens can be minimized or avoided; whether the program is implemented fairly; and whether its benefits and burdens are fairly balanced. Using the framework articulated in Section 1, I understand these considerations as putative justifications we might offer to individuals for interventions that affect them. I suggest that a striking feature of Kass’s list is that many of these considerations turn, again, in large part on how we describe the program in the first place. For example, in this paper, I have focused on one aspect of Kass’s sixth consideration: whether benefits and burdens are fairly balanced. On the face of it, it seems that the benefits and burdens of lockdown are not fairly balanced, given the very different effects on the young and the old. However, as I have suggested, we cannot simply take this appearance for granted; rather, we need to argue that it is ethically proper to conceptualize lockdown in ways that distinguish the young and the old, and that adopt a narrow account of the benefits and burdens of those policies. Before we can apply Kass’s framework, we need to decide how to describe the policy at all.
This may all seem very depressing. We all already know that there are many different ethical frameworks which we could use to assess lockdown, and that choice of a framework will be controversial. We also know that, even if we choose a particular framework, there are often disagreements over how to interpret particular principles and their application to specific cases. I seem to have suggested a further problem: that how we describe the world will also be controversial. Does that mean that the ethics of lockdown is impossible? I suggest not. Rather, my suspicion is that by the time we have decided how best to frame the ethics of lockdown, we will have gone a long way to settling the ethics of lockdown.
Thanks to Joseph Wu for discussion of related ethical issues; to Anna Alexandrova and Jeff Skopek for feedback on an earlier version of this paper; to the audience at a special seminar COVID-19: HPS perspectives, hosted at the University of Cambridge, for useful questions; and to an anonymous reviewer and the editors for their speedy, insightful, and challenging comments. Particular thanks to Charlotte Goodburn for discussion of this paper and to Isaac and Clara for tolerating such discussion.
Alexandrova, Anna. 2017. A Philosophy for the Science of Well-being Oxford: Oxford University Press.
Badano, Gabriel, Stephen John, and Trenholme Junghans. 2017. “NICE’s Cost-Effectiveness Threshold.” In Measurement in Medicine: Philosophical Essays on Assessment and Evaluation edited by Leah McClimans, chapter 10. London: Rowman and Littlefield.
Bartley, Mel. 2016. Health Inequality: An Introduction to Concepts, Theories, and Methods, Second Edition. London: Polity Press.
Bayer, Ronald and Amy L. Fairchild. 2016. “Means, Ends and the Ethics of Fear-based Public Health Campaigns.” Journal of Medical Ethics 42 (6). http://dx.doi.org/10.1136/medethics-2016-103621.
Bobinac, Ana, N. Job A. van Exel, Frans F. H. Rutten, Werner B.F. Brouwer. 2012. “Inquiry into the Relationship between Equity Weights and the Value of the QALY.” Value in Health 15 (8): 1119–26.
Broadbent, Alex. 2020. “Lockdown is Wrong for Africa.” Mail and Guardian, April 8. Accessed May 14. https://mg.co.za/article/2020-04-08-is-lockdown-wrong-for-africa/.
Childress, James F., Ruth R. Faden, and Ruth Gaare. 2002. “Public Health Ethics: Mapping the Terrain.” Journal of Law, Medicine & Ethics 30 (2): 170–78.
Cohen, I. Glenn, Norman Daniels, and Nir Eyal (Editors). 2015. Identified Versus Statistical Lives: An Interdisciplinary Perspective. Oxford: Oxford University Press.
Daniels, Norman. 2007. Just Health. Cambridge: Cambridge University Press.
Dupré, John. 2020. “Following the Science in the Corona-19 Epidemic.” Nuffield Council on Bioethics, April 29. Accessed May 14, 2020.https://www.nuffieldbioethics.org/blog/following-the-science-in-the-covid-19-pandemic.
Fleurbaey, Marc, and Alex Voorhoeve. 2013. “Decide as You Would with Full Information.” In Inequalities in Health: Concepts, Measures, and Ethics, edited by Nir Eyal, Samia A. Hurst, Ole F. Norheim, et al. Oxford: Oxford University Press
Fuller, Jonathan, and Luis J. Flores. 2015. “The Risk GP Model: The Standard Model of Prediction in Medicine.” Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences 54: 49–61.
Hájek, Alan. 2007. “The Reference Class Problem is Your Problem Too.” Synthese 156 (3): 563–85.
Hausman, Daniel M. 2007. “What’s Wrong with Health Inequalities?” Journal of Political Philosophy 15 (1): 46–66.
–––––– 2015. Valuing Health. Oxford: Oxford University Press.
Hope, Christopher. 2020. “The Story Behind ‘Stay Home, Protect the NHS, Save Lives’ – The Slogan that was ‘Too Successful’.” The Daily Telegraph, May1. Accessed May 14, 2020. https://www.telegraph.co.uk/politics/2020/05/01/story-behind-stay-home-protect-nhs-save-lives/.
John, Stephen D. 2013. “Cancer Screening, Risk Stratification and the Ethics of Apt Categorisation: A Case Study.” In Ethics in Public Health and Health Policy, edited by Daniel Strech, Irene Hirschberg, and Georg Marckmann, 141–52.Springer International.
–––––– 2014. “Risk, Contractualism and Rose’s Prevention Paradox.” Social Theory and Practice 40 (1): 28–50.
Kass, Nancy E. 2001. “An Ethics Framework for Public Health.” American Journal of Public Health 91 (11): 1776–78.
Liddell, K, Skopek, J, Palmer, S and Sagar, F (Forthcoming) ‘Who Gets the Ventilator? Important Legal Rights” forthcoming in Journal of Medical Ethics
London, Alex J., and Jonathan Kimmelman. 2020. “Against Pandemic Research Exceptionalism.” Science 368 (6490): 476–77.
Malm, H. M. 1999. “Medical Screening and the Value of Early Detection When Unwarranted Faith Leads to Unethical Recommendations.” Hastings Center Report 29 (1): 26–37.
Marmot, Michael G., D. G. Altman, David Cameron, J. A. Dewar, S. G. Thompson, and Maggie Wilcox. 2013. “The Benefits and Harms of Breast Cancer Screening: An Independent Review. British Journal of Cancer 108 (11): 2205.
Mitchell, Mary. 2018. The Construction of Well-Being. PhD Thesis Submitted to University College London. Accessed May 14, 2020. Archived at https://discovery.ucl.ac.uk/id/eprint/10064726/1/Mitchell%20The%20Construction%20of%20Well-Being.pdf.
Nuffield Council on Bioethics. 2007. Public Health: Ethical Issues. London: Nuffield Council on Bioethics.
Otsuka, Michael, and Alex Voorhoeve. 2009. “Why it Matters that Some Are Worse Off Than Others: An Argument Against the Priority View.” Philosophy & Public Affairs 37(2): 171–99.
Rawls, John. 1971. A Theory of Justice. Cambridge, Mass: Harvard University Press
Reichenbach, Hans. 1949. The Theory of Probability, Second Edition. Berkeley: University of California Press.
Rubin, David, and Paul A. Offit. 2020. “We Know Crowding Affects the Spread. It May Affect the Death Rate.” The New York Times, April 27. https://www.nytimes.com/2020/04/27/opinion/coronavirus-crowds.html
Schaeffer, G. Owen and Angela Ballantyne. 2020. “Downloading COVID-19 Contact Tracing Apps is a Moral Obligation.” Journal of Medical Ethics Blog, May 4. Accessed May 14, 2020. https://blogs.bmj.com/medical-ethics/2020/05/04/downloading-covid-19-contact-tracing-apps-is-a-moral-obligation/.
Schliesser, Eric, and Eric Winsberg. 2020. “Climate and Coronavirus: The Science is Not the Same.” New Statesman, March 23. Accessed May 14, 2020. https://www.newstatesman.com/politics/economy/2020/03/climate-coronavirus-science-experts-data-sceptics.
Schmidtz, David. 2001. “A Place for Cost‐benefit Analysis.” Philosophical Issues 11 (1): 148–71.
Sen, Amartya Kumar. 1970. “The Impossibility of a Paretian Liberal.” Journal of Political Economy 78 (1): 152–57.
Singer, Peter, and Michael Plant. 2020. “When Will the Pandemic Cure Be Worse Than the Disease?” Project Syndicate, April 6. Accessed May 14, 2020. https://www.project-syndicate.org/commentary/when-will-lockdowns-be-worse-than-covid19-by-peter-singer-and-michael-plant-2020-04.
Sorell, Tom. 2013. Emergencies and Politics. Cambridge: Cambridge University Press
Spiegelhalter, David. 2020. “What are the Risks of COVID? And What is Meant by “the Risks of COVID”?” Medium, May 13. Accessed May 14, 2020. https://medium.com/wintoncentre/what-are-the-risks-of-covid-and-what-is-meant-by-the-risks-of-covid-c828695aea69.
Suikkanen, Jussi. 2020. Contractualism. Cambridge: Cambridge University Press.
Thompson, Christopher J. 2018. “Rose’s Prevention Paradox.” Journal of Applied Philosophy 35 (2): 242–56.
Verweij, Marcel F., and Angus J. Dawson. 2013. “Public Health Ethics.” In The International Encyclopedia of Ethics, 4220–30. London: Wiley-Blackwell.
Wilkinson, Richard, and Kate Pickett. 2010. The Spirit Level: Why Equality is Better for Everyone. Bloomsbury Publishing.
Wilkinson, Timothy M. 2007. “Contagious Disease and Self-Defence.” Res Publica 13 (4): 339–59.
Williams, Bernard. 1995. “Ethics.” In Philosophy: A Guide Through the Subject, edited by A. C. Grayling, 545–82. Oxford: Oxford University Press.
Wilson, James. 2011. “Why It’s Time to Stop Worrying About Paternalism in Health Policy.” Public Health Ethics 4 (3): 269–79.
Wolff, Jonathan. 2007. “What is the Value of Preventing a Fatality?” In Risk: Philosophical Perspectives, edited by Tim Lewens, chapter 3. London: Routledge.
and Anver de Shalit. 2007. Disadvantage. Oxford: Oxford University Press.
 At least, as long as we focus on individuals’ ex-ante interests.
 Maybe you think that this is all too fast, for two reasons. First, you might think that fluoridation is significant because it limits choice, a topic I haven’t discussed. Second, you might think, following Wilkinson and Pickett (2010), that more equal distributions might help each. In response: first, even if you think fluoridation has “hidden” costs, note that the costs are the same for everyone—the two policies differ; second, it may well be that equality is good for all, but, if so, pick another example—it’s simply wishful thinking to assume that all valuable public health policies are all good for everyone.
 It is worth noting here that it is not uncontroversial that we should think about interests in these ex-ante terms (Fleurbaey and Voorhoeve 2013). I defend this general stance in John (2014).
 Note that in this section, I focus on consequentialist accounts of the ethics of lockdown, rather than its communication, but if we start from consequentialism, there is a fairly straightforward argument for viewing public health communication in highly instrumentalist terms (Bayer and Fairchild 2016).
 Note that my claims about “wellbeing” can be read as reflecting a long-standing concern in political philosophy, that there may be a gap between “ethical” accounts of “wellbeing” and political notions of “advantage.” For a useful, theoretically self-conscious, attempt to construct an account of “advantage,” see Wolff and de Shalit (2007). For in-depth discussion of many related issues, see Mitchell (2018). One thing I have tried to do here is to remind the reader of the importance of these debates; another is to extend them beyond “wellbeing” to categorization.