Should the Numbers Count?
Consider the following situation:
Drug Distribution – There are six individuals in need to be administered a drug in the next ten minutes, otherwise their condition is such that they would die for sure. You have enough doses to save all six individuals, but you cannot reach them all on time. You can either go south to save individual A, or go north and save individuals B, C, D, E, and F. You have no agent-related or agent-neutral reason to save any one individual in particular.
In an article with the same title as this post, the philosopher John Taurek somehow counterintuitively argued that we actually do not have any reason to prefer to save individuals B-F rather than individual A. Taurek’s argument is complex and combines several more specific claims:[1]
(i) Given the assumptions made, it is certain that no one will suffer more if we choose to save A rather than B-F.
(ii) There is no coherent impersonal sense according to which it is worse if five persons die instead of one, it would only make the situation worse for more individuals.
(iii) If we want to show equal concern, we should give to each individual the same chance of being saved, and so toss a coin.
In his book What We Owe to Each Other, Thomas Scanlon briefly discusses and rejects Taurek’s claim. Scanlon reasonably argues that a principle justifying the use of a lottery to determine who to save (giving 1/6 chance to each person to be saved) could be reasonably rejected by any member of the B-F. Someone belonging to the larger group could object that her claim to be saved has not been appropriately taken into account since the fact that she has a claim does not make any difference compared to a scenario where she would not be present. On the other hand, if A would not have been present in Drug Distribution, then what ought to be done would be obvious, that is, to save individuals B-F. But then, that means that A’s claim does make a difference.
Scanlon however considers that this reasoning cannot apply when different persons are not suffering from identical or similar kinds of harm. Scanlon discusses in particular the following case that is now quite notorious among moral philosophers:
Jones’s Accident – Jones has suffered an accident in the transmitter room of a television station. Electrical equipment has fallen on his arm, and he cannot be rescued without turning off the transmitter for fifteen minutes. If this is not done, he will greatly suffer for the next hour. If the transmission is turned off, however, the broadcasting of the final match of the World Cup of football will be stopped for fifteen minutes, slightly harming the millions of people watching the match.
If the numbers count, then there is some threshold number of people watching the match beyond which we should not turn the transmitter off. Scanlon contends that this is implausible. Jones, and any individual that could potentially end up in a similar situation to Jones’s, would have a reasonable ground to reject an aggregative principle taking into account the number of people slightly harmed. This rejection would be based on the fact that the seriousness of the harm received cannot be compared with the harm that each individual who cannot watch the match for fifteen minutes suffers. On the other hand, it seems that the latter do not have any reasonable ground to reject a principle justifying helping Jones. In particular, the kind of argument used to allow for giving some weight to the numbers in Drug Distribution is not relevant here because the fact of not helping the larger number of individuals does not indicate any failure to take their claims into account.
This may sound too radical. Scanlon indeed suggests that we should take the numbers into account when and only when the different harms suffered by individuals, though different in intensity, remain in the same range or are comparable. But this can lead to troubles regarding the consistency of the underlying “betterness” judgment. This is the core topic of Larry Temkin’s fascinating book Rethinking the Good. Temkin discusses in particular what we can call “spectrum cases” which have the following formal structure. Suppose that you have a state of affairs A where one person is intensely suffering. You have then a state of affairs B where a small number of persons who are greatly suffering, but slightly less than the person in state A. Construct on this basis a series of state of affairs C, D…, Z where the numbers of persons suffering is strictly increasing but the harm suffered is less serious than in the previous state. Adjust the two variables (number of persons and intensity of suffering) such that when comparing two adjacent states of affairs in the series, you always conclude that the one which comes first in the series (i.e., the one with the lowest number of persons suffering) is the best. Hence, you conclude that A is better than B, which is better than C, …, which is better than Z. Now, comparing A and Z, we have something that looks similar to Jones’s Accident. If we agree with Scanlon (and Temkin, by the way), we should conclude that Z is better than A. But we have a problem: we have just constructed a case where our intuitively plausible judgment is intransitive and even cyclical.
Based on this kind of case, Temkin argues that transitivity is not a plausible constraint that our moral judgments should respect. The problem however is not really with transitivity[2] but with what philosophers call aggregation and economists would rather call “additivity” (of welfare, of claims, of utility…). Now, consider a last case:
Exiting a Pandemic – A pandemic has led the state to enforce coercive measures restricting individual liberties to limit the casualties of a relatively deadly virus. A vaccine is now widely available and virtually lessens the risk of death from the virus to 0. Based on this, the state has decided to release most of the coercive measures, despite the fact that a small fraction of the population is not eligible for vaccination due to health-related reasons. As a result, these persons face a significant risk of death without fault on their part.
Any resemblance with what is happening in most countries since the beginning of the year is not a coincidence. Democratically elected governments have basically decided that the numbers do count. Vaccinated people have a claim to having individual liberties that make them able to enjoy life. People who can’t vaccinate for health-related reasons have a claim to stay alive and to be protected from the virus, while also having the same individual liberties. Admittedly, the latter is stronger than the former. Individual liberties are morally important – for some, they even have a lexical priority over other goods and values – but saving lives is arguably even more important. This has been indeed the underlying justification for lockdowns and other coercive measures implemented at the beginning of the pandemic. But now that very few people are at risk of death, it seems that betterness judgment has shifted. So, the numbers count ultimately, at least in matters of public policy.
Sure, that doesn’t prove anything about what is the correct moral judgment (if there is one) in this kind of situation. Maybe what governments have decided is plainly wrong. Or maybe it is not and that we should not in any case rely on our intuitions about big numbers because they are not trustworthy (but on what should we rely then?). Also, there is a difference between Exiting a Pandemic and Jones’s Accident: in the former but not in the latter, the harm is only probabilistic. Maybe the fact that the risk of death for non-vaccinated people is less than 1 makes it acceptable to put them at risk. But, beyond the fact that adding probabilities to the mix makes our intuitions even less trustworthy, it is a possibility for the large majority of us who are now (mostly) free from any risk our ethical judgment is polluted by self-interest considerations.
So, I don’t have a definitive answer to the question “should the numbers count?”. But we have recently taken collectively a stance about it.
[1] I’m here following Frances Kamm’s discussion of Taurek’s argument. See Morality, Mortality: Death, and Whom to Save from It, Offord University Press, 1993, pp. 75-6.
[2] Readers who are familiar with this topic will immediately see that there is a formal trick to recover transitivity by properly “indexing” the options that are compared. In the example discussed in the text, state of affairs A is not the “same” state of affairs whether it is compared with state B or with state Z.