A recent practical effort to balance these concerns against each other appears in the Model State Emergency Health Powers Act (2001), available at <http://www.publichealthlaw.net/MSEHPA/MSEHPA2.pdf> (last visited May 3, 2005). The specific accommodation represented by the Act has been criticized by some civil libertarians. See, e.g., May, T., “Political Authority in a Bioterror Emergency,” Journal of Law Medicine & Ethics 32 (2004): 159–163, at 161 (2004); Parmet, W. E., “Quarantine Redux: Bioterrorism, AIDS, and the Curtailment of Individual Liberty in the Name of Public Health,” Health Matrix 13 (2003): 85–116; Annas, G. J., “Blinded by Bioterrorism: Public Health and Liberty in the 21st Century,” Health Matrix 13 (2003): 33–70, at 46–63; Goldstein, A., “Mr. Quarantine Meet Miss Liberty,” Time, April 18, 2002, at 19.CrossRefGoogle Scholar

Here I am using the term “quarantine” in its generic sense to refer to the broad class of practices that contain infectious diseases by segregating possible carriers from the general population. A narrow usage distinguishes isolation – which segregates only those who are actually ill – from quarantines proper – which segregate exposed persons within a community whether they are ill or not – from cordons sanitaire – which segregate entire communities (typically cities or towns) from the outside world.Google Scholar

Other diseases – including many gastrointestinal diseases and most sexually transmitted diseases – may be controlled by blocking the specific pathways through which the infections spread, for example by public and private sanitary practices such as good sewers and safe sex. Many of the comparisons that I draw between quarantines and vaccinations may also be drawn between quarantines and sanitation, although I shall not develop these points here.Google Scholar

One prominent example is the proposal to respond to a smallpox outbreak by a combination of “isolation of infected individuals, surveillance of contacts, and a focused selective vaccination program.” Henderson, D. A. et al., “Consensus Statement, Smallpox as a Biological Weapon: Medical and Public Health Management,” JAMA 281 (1999): 2127–2137, at 2136–37 (1999). I shall return to this proposal in later endnotes.Google Scholar

“Primary active immunization develops more slowly than the incubation period for most infections and must therefore be induced prior to exposure to the etiologic agent” Stites, D. P. Terr, A. I., and Parslow, T. G., Basic and Clinical Immunology 8th ed.(Norwalk, CT: Appleton & Lange, 1994) at 718. The head start needed by a vaccine may be substantial. The Ebola vaccine, for example, takes four weeks to confer immunity. See Clarke, T. and Knight, J., “Fast Vaccine Offers Hope in Battle with Ebola,” Nature 424 (2003): 602. See also, Sullivan, N. J. Geisbert, T. W., and Geisbert, J. B. et al., “Accelerated Vaccination for Ebola Virus Haemorrhagic Fever in Non-Human Primates,” Nature 424 (2003): 681–84 (2003). Earlier Ebola vaccines took six months and several boosters to confer immunity.Google Scholar

If a disease spreads sufficiently slowly and a vaccine against it confers immunity sufficiently quickly, it may be possible to contain the disease through post-outbreak vaccinations administered broadly, including to persons who have not been exposed. In such cases, vaccines are less inefficient than the main text contemplates and commensurately more appealing. This possibility only strengthens the case for vaccinations presented in the main text.Google Scholar

This means that the inefficiency of vaccinations increases as the chances of an outbreak decrease. It is therefore unsurprising that vaccinations are most prominently employed to protect against endemic diseases, for example the familiar childhood diseases, whose outbreak is virtually certain.Google Scholar

This means that the inefficiency of vaccinations increases as the infectiousness of a disease decreases. It is therefore unsurprising that vaccinations are most prominently employed to protect against diseases, including again the familiar childhood diseases, that are extremely infectious.Google Scholar

Note that these numbers reflect a within-quarantine rate of transmission of the infection that exceeds the rate of transmission in the general population in the absence of a quarantine. This is a reasonable assumption, because quarantines will typically confine persons who have merely been exposed in close proximity with persons who are already infected and therefore impose increased risks of infection upon persons who enter the quarantines free of the disease. This assumption will play a prominent role later in the argument, and I take it up in greater detail there.Google Scholar

A one in five chance of suffering two-and-a-half million infections imposes an expected aggregate burden of 500,000 infections.Google Scholar

Ten million certain vaccinations at a per-vaccination burden equivalent to 1/40 the burden of infection impose an aggregate burden equivalent to 250,000 infections.Google Scholar

A one in five chance of confining one million persons to a quarantine in which 650,000 will become infected and all will face a burden of confinement equivalent to 1/10 the burden of infection imposes an expected aggregate burden equivalent to 150,000 infections.Google Scholar

The phrase comes from Defoe, whose account of the Plague of London reports that the practice of “confining the Sound in the same House with the Sick, was counted very terrible,” Defoe, D., A Journal of the Plague Year (1772) Wall, Cynthia ed., (New York: Penguin, 2003): at 150, and that “many people perished in these miserable Confinements, which ‘tis reasonable to believe, would not have been distemper’d if they had had Liberty.” Id. at 48. Defoe's belief is confirmed by modern accounts of the Plague, which conclude that “certainly, the practice of shutting up a family in their house when plague first struck must have increased the inter-household contact rate.” Scott, and Duncan, , supra note 16, at 226.Google Scholar

See CDC Guidelines, supra note 19, at 29.Google Scholar

More than 30 states forcibly detained tuberculosis patients in the years following World War II, for example, either in locked hospital wards or in prisons. See Levner, B. H., “Tuberculosis in Seattle, 1949–1973: Balancing Pubic Health and Civil Liberties,” Western Journal of Medicine 171 (1999): 44–45.Google Scholar

One model of a large smallpox attack in a population of 10 million persons, for example, concludes that controlling the attack through post-exposure vaccinations and quarantines would require quarantining 59,000 symptomatic cases. See Kaplan, Craft, , and Wein, , supra note 11, at 10935–36.Google Scholar

Here it is of course critical that vaccinations are not generally effective unless administered prior to exposure to the diseases that they protect against, so that within-quarantine rates of infection generally cannot be controlled simply by vaccinating those who have been quarantined.Google Scholar

Some of these burdens may perhaps be compensated, even if they cannot be eliminated. But problems of valuation and perhaps even incommensurability will render any compensation that is offered imperfect at best and at worst inadequate. Moreover, not all the burdens associated with quarantines can be compensated. Most starkly, when a disease is fatal, the infections caused or allowed by quarantines cannot be compensated.Google Scholar

The numbers in the example make the simplifying assumption that only a few quarantined persons are actually infected when the quarantine is imposed and that virtually all quarantined persons who eventually suffer the disease become infected within the quarantine. This assumption is necessary for equating the within-quarantine risks of infection with the ratio of total within-quarantine infections to quarantine size. The assumption fits with the perfectly effective quarantine that the example involves, which must sweep broadly enough so that no cases arise without it and must therefore sweep up many persons who have been exposed but not infected or have not even been exposed.Google Scholar

A secondary effect is worth mentioning in the margin. When vaccination policies fail to immunize everyone or fail to confer total immunity, some persons may become infected even under a vaccination regime. Vaccinations, in such cases, will reduce but not eliminate infections. Moreover, it may happen that the weaker force of infections in an imperfectly vaccinated population causes the (fewer) persons who are infected to catch the infection at older ages and therefore to become more seriously ill than the younger persons who would otherwise be infected. See Anderson, and May, , supra note 16, at 99. In this case, the greatest individual burdens may be greatest under a vaccination regime. I do not take up this unlikely possibility here.Google Scholar

See Euripides, Iphigenia at Tauris and Iphigenia at Aulis. The planned sacrifice never actually occurred, because Iphigenia was miraculously transported to Tuaris and an animal sent in her place.Google Scholar

This concern for fairness finds an interesting expression in the Christian adaptation of religious sacrifice, in which Christ died on the cross to relieve humanity of the burden of it sins: the conceit of Christ's divinity, which is necessary for the crucifixion's effectiveness, at once eliminates any ordinary concern for the fair treatment of the sacrificial offering.Google Scholar

Every economy of course also contains unproductive persons who are responsible for their state – who have squandered their talents, who refuse to exploit their talents, or even who have simply failed to develop their talents. The observations in the main text do not take a position on the complex of questions involved in separating the two classes of unproductive persons, save perhaps insofar as they assume (plausibly, I think) that the class of persons who are not responsible for being unproductive is not empty.Google Scholar

Taxes, as economists say, create dead-weight loss.Google Scholar

At least in an otherwise Pareto perfect economy.Google Scholar

See Rawls, J., A Theory of Justice (Cambridge, MA: Belknap Press of Harvard University Press, 1971): at 75.Google Scholar

Notice that although efficient markets sacrifice the unproductive in the service of efficiency, there is no suggestion that they use the unproductive literally as means, so that the ideas concerning human inviolability mentioned in note 30, supra, cannot account for our intuitive opposition to unfettered free markets. Indeed, if these ideas apply at all to the economic case, then they apply on the other side of the argument, as in Robert Nozick's suggestion that redistributive taxation drafts the productive into the service of the unproductive. See Nozick, R., Anarchy, State, and Utopia (New York: Basic Books, 1974): at 172, 228–29. I do not accept this suggestion, for reasons that I have set out in Markovits, D., “How Much Redistribution Should There Be?” Yale Law Journal 112 (2003): 2291–2329, at 2325–2326. The merits of this argument do not matter here, however, although it is worth mention that our distributive intuitions against minimizing aggregate burdens by concentrating them seem able to stand up even to our Kantian intuitions concerning the inviolability of persons.Google Scholar

Finally, ultimately, overall.Google Scholar

The reference to a utilitarian in this context is natural, since the utilitarian concern to maximize aggregate well-being broadly understood presents the most historically prominent general foundation for the impulse to minimize aggregate suffering from disease that underwrites quarantines' intuitive appeal. The dictum, incidentally, is attributed to Bentham by Mill. See Mill, J. S., Utilitarianism, Sher, George ed. (Indianapolis, IN: Hackett Publishing Co., 1979): at 60. Although it is commonly thought that Bentham never actually adopted this form of words himself – and Mill certainly provides no citation – Bentham in fact did utter almost the exact phrase Mill reports, saying that “every individual in the country tells for one; no individual for more than one.” See Bentham, J., “Rational of Judicial Evidence,” in Bowring, J., ed., The Works of Jeremy Bentham, vol. 7 (Edinburgh and London: W. Tait, 1843): vol. 6: 201–585 at 334, and vol. 7: 1–644. I would like to thank Gerald Postema for pointing this reference out to me.Google Scholar

Rawls, , supra note 34, at 27.Google Scholar

The marginalist conception of equality may have secondary or indirect distributive implications, as in the familiar utilitarian suggestion that, in light of the diminishing marginal utility of money, equal distributions of a fixed total of wealth will tend to produce greater total well-being than unequal distributions. But the utilitarian concern for distributions is necessarily incidental and contingent only – there is of course no diminishing marginal utility of well-being, and utilitarians will accept unequal distributions of well-being whenever such inequalities contribute to increases in total well-being. This distinction is practically important, moreover, because the utilitarian concern for equality even of wealth runs out when the diminishing marginal utility of wealth does not apply or is outweighed by other considerations. Utilitarians will not, for example, support redistribution in favor of persons whose handicaps render them inefficient converters of wealth into well-being (so that their marginal utility of wealth is low everywhere on the scale).Google Scholar

This way of thinking was brought to prominence in modern moral philosophy by Anscombe, G. E. M. and Taurek, J. See Anscombe, G. E. M., “Who is Wronged?” The Oxford Review 5 (1967): 16–17 and Taurek, J., “Should the Numbers Count?” Philosophy and Public Affairs 6 (1977): 293–316.Google Scholar

This assumes, of course, that the risks of exposure, quarantine, and infection fall equally on all persons. I take up this assumption, and the weakness it introduces into the case for measuring individual burdens ex ante, in a moment.Google Scholar

This is entailed by the assumption that the quarantines are more efficient than the vaccinations.Google Scholar

In the extreme case, in which the identities of the persons who will be exposed and quarantined are fixed in advance of any actual outbreak of disease, the pattern of individual ex ante burdens associated with quarantines will precisely match the pattern of individual burdens measured in medias res. In this case, the priortarian concern that quarantines are unfair will apply identically from the ex ante point of view as at the moment of the quarantine decision.Google Scholar

Here it is instructive to note the contrast between the use of the ex ante point of view in the argument that quarantines satisfy prioritarian distributive principles and Rawls's use of the ex ante point of view in the Original Position. Although the Veil of Ignorance obscures all facts that distinguish persons from one another in the real world, so that the parties to the Original Position cannot know the individual circumstances they will experience once the Veil is lifted, it does not deindividuate outcomes in the manner that I have criticized. The parties behind the Veil do not know just the aggregate benefits and burdens that will arise in the society that they will enter when the Veil is lifted but rather the distinct parcels of benefits and burdens that all individual persons in that society will possess. Moreover, the parties reason as if the worst of these parcels will be individually theirs. In stark contract to the argument that I have criticized, the ex ante point of view, in Rawls's hands, emphasizes rather than obscures the differences among persons that will arise ex post.Google Scholar

This way of putting the point recalls Stephen Perry's argument that (at least insofar as the underlying outcomes proceed deterministically) a risk of loss, rather than being itself a harm, is merely a reflection of epistemic limitations concerning knowledge of certain harms. See Perry, S., “Risk, Harm, and Responsibility,” in Philosophical Foundations of Tort Law, Owen, D. ed., (New York: Oxford University Press, 1995): At 321–46 Perry argues, on the basis of this claim, that risk itself should not be actionable in tort, that is, that risk is not salient for purposes of corrective justice. And indeed, it seems that treating unresolved risks as harms acquiesces in our epistemic limitations in an unjustified way.Google Scholar

Indeed, it is critical to prioritarian distributive principles that persons maintain a coherent identity across their many experiences and that this identity, and not just the individual experiences, is a proper subject of moral concern. One line of attack against the prioritarian view of distributive justice, most prominently presented by Derek Parfit, seeks to cast doubt on precisely this claim about personal identity. See Parfit, D., Reasons and Persons (New York: Oxford University Press, 1984).Google Scholar

These considerations, incidentally, also have implications for the management of non-communicable diseases (such as anthrax), for which the question of quarantines obviously does not arise. When such diseases are unlikely to affect many people, the aggregate costs of vaccinations (which must be administered generally) may exceed the aggregate costs of the relatively few infections that the vaccinations would prevent. But fairness may require making vaccinations available in such cases even though they are inefficient. The greatest individual costs of suffering the diseases may exceed the greatest individual costs of the vaccinations that protect against them. And although this relation between individual burdens would be reversed if the burdens were assessed ex ante, so that the costs of contracting the diseases were discounted by the unlikelihood of doing so, the arguments in the main text explain why distributive arguments may not assess individual burdens from this ex ante point of view. Some people may of course decline the vaccinations they are offered (including for the reason that their expected burdens exceed their expected benefits). Those who decline vaccinations and then become infected cannot, for reasons elaborated below, make distributive claims based on the burdens of their infections.Google Scholar

Many prioritarians, for example T. M. Scanlon, do in fact accept this conclusion. See Scanlon, , supra note 46, at 229–41. A notable exception to this approach is G.E.M. Anscombe's suggestion that although “because they are more” is a perfectly intelligible reason for saving the larger group, “it doesn’t follow that a man acts badly if he doesn’t make it his reason.” Anscombe, , supra note 40, at 16–7. For Anscombe, either course in such a case – saving the smaller or the larger group – “seems O.K.” Id. at 16.Google Scholar

This account of the prioritarian's reasoning in saving the larger number, and the sense in which it does without reference to aggregate burdens, appears in Kumar, R., “Contractualism on Saving the Many,” Analysis 61 (2001): 165.CrossRefGoogle Scholar

Note that this argument explains the claim, made in note 17, that the prioritarian distributive preference for vaccinations over quarantines is not undermined by the fact that vaccinations cause a small number of individual harms that are as severe as the diseases they protect against. These individual harms are counterbalanced, one-by-one, by the more numerous and equally individually severe harms caused by the diseases in case vaccinations are foregone, until the distributive objections to vaccinations are all eliminated while distributive objections to quarantines remain in force.Google Scholar

In the case of quarantines, these two groups will typically overlap – some of those who will be infected under a quarantine will also be infected without a quarantine. This can only strengthened the case for quarantines. The persons who will be infected in every event and who therefore cannot be saved may be taken out of the moral calculus altogether. And doing so skews the ratio of numbers saved still further in favor of quarantines.Google Scholar

Determinism, on one understanding, is the view that such causal explanations, going arbitrarily far back, are always possible.Google Scholar

Notice that these reflections fix the point in time at which in medias res assessments of individual burdens may properly be made by connecting it to the conceptual structure of prioritarian distributive justice. The in medias res perspective does not arise just anywhere in the middle of the chain of events associated with an outbreak of disease but is rather the point at which all distributively salient differences between persons have resolved themselves and the only differences that remain are unconnected to any of the intuitive concerns of distributive justice. I have proposed that this happens after initial exposure to a disease but before persons finally become infected. But for some diseases it may happen earlier. Nothing rules out that a disease's patterns of exposure may disrespect all intuitively salient distributive categories and so appear as a matter of irreducible luck (and SARS may present patterns of exposure that approximate this case, although it is far from clear that the Western response to SARS was free from nativist anti-Asian sentiment). For such diseases, distributive assessments of individual burdens may perhaps appropriately be made from a point of view before any outbreak has occurred and in a way that spreads risks of exposure evenly across all persons (in which case the in medias res point of view, identified by the structural test set out here, coincides with the ex ante point of view). And such diseases may perhaps fairly be combated using quarantines.Google Scholar

Gambles, incidentally, also present everyday cases of Acts of God, or irreducible luck. The players at a roulette table, for example, will in the end enjoy distinct outcomes depending upon the spin of the wheel and the path of the ball. But these distinct outcomes are completely unrelated to any intuitively compelling difference between the persons who bear them – indeed, this is precisely roulette's charm for the player. And this is another reason for which a prioritarian might take an ex ante view of games of chance.Google Scholar

Indeed, such redistribution renders gambling impossible – a gamble depends, as a conceptual matter, on leaving luck within the gamble to lie where it falls. Any effort to redistribute luck within a gamble drafts some gamblers into the service of others and therefore undermines the choices of all gamblers.Google Scholar

The defense against a risk may invoke concrete methods (including the many ways of taking care) that actually prevent the risk's bad outcome from occurring or instead invoke more abstract methods (most notably contracts, including familiar forms of insurance) that provide compensation in case the bad outcome does occur. A similar discussion of converting ordinary risks into gambles (which emphasizes the special case of insurance) appears in Dworkin, R. “What is Equality? Part 2: Equality of Resources,” Philosophy & Public Affairs 10 (1981): 283–345, at 297.Google Scholar

This approach, it is worth noting in the margin, is consistent with Perry's observation, reported earlier, that a risk of loss is not itself a harm but only a reflection of epistemic limitations concerning knowledge of certain harms. See note 47, supra. The distributive burden of an unresolved risk arises not out of the risk itself – which might after all never eventuate – but rather out of the interplay between the risk and the risk-bearer's epistemic limitations. The risk makes these limitations immediately more costly to persons who come to face it: the risk places an unchosen charge on the epistemic limitations of persons who fall subject to it; and this charge applies even to the unresolved risk, and whether or not it eventuates.Google Scholar

If foregoing vaccinations involves forbidding individual persons from getting vaccinated, then the failure to be vaccinated less obviously resembles a deliberate gamble, at least on the part of persons who would have sought out vaccinations but for the prohibition. At the very least, the resemblance depends upon a supplemental argument, for example about democratic authority, establishing individual responsibility for collective decisions. But there is little justification for prohibiting vaccinations in the kinds of cases at issue (where vaccinations surely fall within the range of the medically reasonable), and this complication may therefore be set aside. The live question is whether to make vaccinations voluntary or mandatory. One argument in favor of making vaccinations mandatory is that persons internalize all of the costs but not all of the benefits of getting vaccinated, so that the rate of voluntary vaccination among persons who rationally pursue their self-interests will fall below the optimal level. See, e.g., Bauch, C. Galvani, A. and Earn, D., “Group Interest versus Self-interest in Smallpox Vaccination Policy,” Proceedings of the National Academy of Sciences 100 (2003):10564.Google Scholar

This thought may be brought within the orbit of distributive justice by proposing that the worse-off should not be given lexicographic priority (so that arbitrarily large gains to efficiency must be sacrificed to secure arbitrarily small benefits for the worse-off) but instead only some more moderate preference (so that sufficiently great gains to efficiency can justify sufficiently small additional burdens for the worse-off). This suggestion does not fit comfortably within prioritarian theory, however, because it is difficult to see how something less than lexicographic priority for the worse-off may be justified without making reference some idea of aggregates of benefit and burden of the sort that the prioritarian view is inclined to declare irrelevant to moral evaluation. It remains intuitively appealing nevertheless.Google Scholar

Compensation remains attractive in this connection, although as I remarked earlier, see note 24, supra, many of the harms that quarantines impose are not compensable.Google Scholar