1. Two principles and three objections

Much critical response to sufficiency views focuses on two well-worn objections. This section clears the way for the rest of the article by distinguishing the excessiveness objection from those objections. In particular, it shows that, unlike the other objections, the excessiveness objection also applies to the now widely accepted position that I refer to simply as ‘sufficientarianism’. Just as the debate has shifted from the doctrine of sufficiency to sufficientarianism, so too should it shift from objections targeting the doctrine of sufficiency to the excessiveness objection, which retains its force against sufficientarianism.

Harry Frankfurt claimed that ‘what is important from the point of view of morality is not that everyone should have the same but that each should have enough’. He adds that ‘[i]f everyone had enough, it would be of no moral consequence whether some had more than others’ (Frankfurt Reference Frankfurt1987: 21). This suggests the following position:

The doctrine’s main rivals are egalitarianism, which requires that distributions are equalized in some significant regard, and prioritarianism, which gives a greater weighting to benefits for the worse off (Parfit Reference Parfit, Mason and Williams2002; McCarthy Reference McCarthy2017; Adler and Holtug Reference Adler and Holtug2019). The doctrine denies that equality or priority are distributive goals,Footnote ³ although it allows that there may be other goals (such as utility maximization) that are lexically posterior to sufficiency.

There are two common and, it is widely thought, telling objections to the doctrine of sufficiency. The inverted priority objection notes that, on this view, we focus on increasing the number of people who have enough (Frankfurt Reference Frankfurt1987: 31). It objects that such a view requires that, among people who do not initially have enough, a small benefit to the better off (those close to having enough) that gives them enough is more valuable than a large benefit to the worse off (those far from having enough) that does not give them enough (Shields Reference Shields2012: 102–103; see also Benbaji Reference Benbaji2005: 311; Arneson Reference Arneson and Kaufman2006: 28; Casal Reference Casal2007: 315–316). The doctrine of sufficiency is ‘willing to make extremely large trade-offs of the welfare of the worse-off individual to increase the welfare of the better-off individual by a small amount’ (Roemer Reference Roemer2004: 279). This is, it is claimed, a perverse reversal of prioritarianism and the Pigou–Dalton principle.

The indifference objection (Shields Reference Shields2012, Reference Shields2016; Nielsen Reference Nielsen2019) emphasizes the doctrine of sufficiency’s negative thesis (Casal Reference Casal2007: 299–300) that, provided everyone has enough, egalitarian or prioritarian distributive principles are irrelevant. It objects that, provided everyone has enough, the doctrine will overlook some being much better off than others merely on account of the families they are born into, or due to discrimination on the basis of race or religion (Casal Reference Casal2007: 311; Temkin Reference Temkin2003 a: 65–66; see also Arneson Reference Arneson2000: 347; Temkin Reference Temkin2003 b: 773–776; Holtug Reference Holtug2010: 233–234). Sometimes, the claim goes, the doctrine will permit the most blatant unfairness, and so fails to fulfil a basic requirement of distributive justice.

These are two powerful objections to Frankfurt’s doctrine. But the general idea that Frankfurt appeals to – that providing people with enough has special distributive importance – can find expression in the more moderate principle with which the article opened. This sufficientarianism says that providing enough, or as close to enough as is possible, is the lexically prior distributive goal.

Sufficientarianism is more moderate than the doctrine of sufficiency in two respects. First, where it is impossible to provide enough for all, it is concerned with providing as close to enough as is possible, and it is therefore invulnerable to the inverted priority objection.Footnote ⁴ Second, it says only that securing enough has lexical priority over equality and priority, not that those goals are irrelevant, so it is not vulnerable to the indifference objection (Brown Reference Brown2005). Both of these departures from Frankfurt’s view are in keeping with core sufficientarian intuitions and with Paula Casal’s ‘positive thesis’ of sufficientarianism, which ‘stresses the importance of people living above a certain threshold, free from deprivation’ (Casal Reference Casal2007: 297–298; see also Shields Reference Shields2012: 115).Footnote ⁵ Given this background, it is natural to understand those who are concerned with giving people enough of something as typically concerned also with giving those who cannot receive enough as much as is possible (see Crisp Reference Crisp2003 a: 757–758; Huseby Reference Huseby2010: 184–185). One’s commitment to alleviating deprivation is in no way hindered by allowing that, once as much deprivation alleviation as can be achieved has been achieved, other distributive goals may be pursued. I therefore take Casal’s positive thesis as the core, or (in her terms) ‘dominant thesis’, of sufficientarianism. (For simplicity of expression I hereafter understand ‘enough’ in a technical sense as ‘enough, or as close to enough as is possible’.)

While sufficientarianism is more moderate than the doctrine of sufficiency in these two respects, it has one remaining extreme feature, which is the target of a third objection. Sufficientarianism treats providing enough not just as weighty but as lexically prior to other distributive goals. Perhaps surprisingly, most sufficiency theorists endorse this absolutism. Roger Crisp holds that ‘absolute priority is to be given to benefits to those below the threshold at which compassion enters’, while Robert Huseby comments that ‘individuals below the maximal sufficiency threshold should have absolute priority over individuals above this threshold’ (Crisp Reference Crisp2003 a: 758; Huseby Reference Huseby2010: 184). Crisp and Huseby favour securing a sufficient level of welfare, but theorists proposing a sufficient level of capabilities also endorse absolutism. For instance, Martha Nussbaum claims that ‘a necessary condition of justice for a public political arrangement is that it deliver to citizens a certain basic level of capability’ (Nussbaum Reference Nussbaum2000: 71, emphasis added; see also Nussbaum Reference Nussbaum2011; Claassen Reference Claassen2018). Elizabeth Anderson’s capabilitarian theory similarly ‘guarantees effective access to the social conditions of freedom to all citizens, regardless of how imprudently they conduct their lives’ (Anderson Reference Anderson1999: 326–327; see also Anderson Reference Anderson, Brighouse and Robeyns2010). Sufficientarians rarely constrain their sufficiency principles by identifying other principles of the same fundamental weight, as egalitarians and prioritarians routinely do by endorsing thoroughly pluralistic views (Arneson Reference Arneson1999; Arneson Reference Arneson2000; Knight Reference Knight2009: Ch. 6; Segall Reference Segall2010; Segall Reference Segall2014; Temkin Reference Temkin, Knight and Stemplowska2011; Agmon and Hitchens Reference Agmon and Hitchens2019). Indeed, while Axelsen and Nielsen emphasize that, on their capability sufficientarian account, ‘justice is … pluralist in nature’, this is only in the sense that ‘being sufficiently well-off means being free from pressure against succeeding in each of the central areas’ (Axelsen and Nielsen Reference Axelsen and Nielsen2015: 407). Their sufficientarianism remains unconstrained: ‘different mechanisms of distribution regulate the different human freedoms, which sometimes involves a concern with relative shares to different degrees, but always for reasons of sufficiency’ (Axelsen and Nielsen Reference Axelsen and Nielsen2015: 409; see also 419).

While several writers have criticized this absolutist feature of sufficientarianism, in doing so they have not usually taken care to distinguish this objection from the indifference and inverted priority objections, which are not applicable to sufficientarianism (Arneson Reference Arneson2000; Temkin Reference Temkin2003 b; Arneson Reference Arneson and Kaufman2006; Casal Reference Casal2007). Even the title of Casal’s ‘Why Sufficiency is Not Enough’ and the subtitle of Arneson’s ‘Distributive Justice and Basic Capability Equality: “Good Enough” Is Not Good Enough’ foreground the indifference (or ‘enough is not enough’) objection. I will present, in some detail, a third kind of objection, the excessiveness objection, which unlike the other objections provides strong grounds for rejecting sufficientarianism.Footnote ⁶ This objection is roughly the mirror image of the indifference objection. Whereas the indifference objection holds that providing enough is not enough, and complains of sufficientarian indifference to other distributive principles, my objection is that providing enough is too much – it rides roughshod over other distributive values. The excessiveness objection aims not at features peculiar to the doctrine of sufficiency, as does the indifference objection, but at the very heart of sufficientarianism: its insistence on alleviating deprivation, understood as bringing individuals up to (or closer to) some particular threshold, come what may. Just as the doctrine of sufficiency is fundamentally flawed, so too, in a different way, is sufficientarianism.

2. The threshold

The definition of the sufficientarian threshold will play an important role in my presentation of the excessiveness objection. I will define the sufficientarian threshold as consisting of 100 units of whatever good it is that the sufficientarian distributes. That is, where a person has 100 units, they are considered to have enough.

Nothing important follows from the use of a specific number, of course, and all my examples could be adjusted where a sufficientarian considers that a higher or lower figure is appropriate. Still, it might be countered that it is inimical to sufficientarianism to present the threshold in a quantitative fashion such that, for instance, 99 and 100 look very similar though one is below and the other at the threshold. Four points to motivate my approach can be made here. First, 99 and 100 plainly are similar in terms of the amount of benefit, and therefore they will be similar from the perspective of theories that are centrally concerned with the amount of benefit, such as utilitarianism. But sufficientarianism says that something else – the threshold – is also very important, and from that perspective, 99 and 100 are not at all similar. Thus, the objection at hand, though seemingly supportive of sufficientarianism, fails to acknowledge the great moral difference between 99 and 100 posited by that view. Second, and confirming this first point, sufficiency theorists frequently make use of quantitative illustrations of the importance of their thresholds (Frankfurt Reference Frankfurt1987; Crisp Reference Crisp2003 a; Benbaji Reference Benbaji2005; Huseby Reference Huseby2010, Reference Huseby2020; Shields Reference Shields2012; Shields Reference Shields2016). It might be asked whether this is nevertheless inappropriate as applied to the – usually capabilitarian – sufficientarians who eschew numerical representations of the threshold. My third point, then, is that the use of numbers is simply a presentational convenience, and that the same arguments could be made without them, for instance with examples where a person will either have, or be marginally short of having, some capability to the requisite degree. My arguments could not be made in this way if the relevant capabilities were all binary. But in fact the key capabilities in the literature are firmly scalar. This appears, for instance, to be true of all ten of Nussbaum’s Central Capabilities: life, bodily health, bodily integrity, practical reason, affiliation, and so on, are all things one can have more or less of. Indeed, the first of these capabilities will largely be expressed numerically, in terms of years, given it includes ‘[b]eing able to live to the end of a human life of normal length’ (Nussbaum Reference Nussbaum2011: 32). Fourth, a sufficientarian that refused to allow those just below the threshold to be compared to those at the threshold, even in principle, would fall squarely into the sights of those that object that sufficientarianism tacitly relies on ambiguity (Casal Reference Casal2007: 315–316). The only plausible sufficientarian position, and the dominant position in the literature, is one that says that a threshold can, in principle, be identified and that some might fall just short of that threshold. That is the view I address.

What kind of threshold is the sufficientarian threshold? If sufficientarianism is to challenge utilitarianism, egalitarianism and prioritarianism, as most of its adherents suppose it does, it must use a threshold that is of non-instrumental significance. Recognizing instrumental thresholds as significant is not a distinguishing feature of sufficientarianism – all credible distributive justice theories do this. For instance, if people need 100 pills to be free of severe pain, and any pills beyond that have only decorative value, utilitarians, egalitarians and prioritarians would all recognize that the 100 pill threshold has particular importance, but for reasons of utility, equality and priority, not sufficiency. As Liam Shields (Reference Shields2016: 27) observes, ‘we might favour policies to ensure that people have secured enough of some things – enough education, enough wealth, or enough health are worthy policy goals – but if these policies are grounded in a non-sufficientarian distributive principle, perhaps a prioritarian principle of welfare, there would be no need for sufficientarian principles to account for this guidance’.

A complexity here is the fact that some apparently sufficientarian views consider the threshold to be important instrumentally rather than non-instrumentally. For example, Anderson’s fundamental position is egalitarian (specifically, relational egalitarian), and she endorses a form of capabilities sufficiency as a means of achieving equality (Anderson Reference Anderson1999, Reference Anderson, Brighouse and Robeyns2010). Even Frankfurt’s doctrine of sufficiency can be interpreted as an instrumental view, furthering broadly utilitarian goals through sufficiency of wealth (Frankfurt Reference Frankfurt1987). I will, however, focus on the more common non-instrumental view of sufficiency, which rivals or complements utilitarianism, egalitarianism and prioritarianism rather than serving them, and considers the threshold to have non-instrumental moral significance (Crisp Reference Crisp2003 a; Huseby Reference Huseby2010; Nussbaum Reference Nussbaum2011; Shields Reference Shields2016).

I will, then, assume that there is a non-instrumentally significant sufficientarian threshold. Sufficientarians have provided various specifications of such a threshold – for instance, in terms of contentment (Huseby Reference Huseby2010; Hassoun Reference Hassoun2021), absence of hypothetical compassion (Crisp Reference Crisp2003 a), dignity (Nussbaum Reference Nussbaum2011), and freedom from duress (Axelsen and Nielsen Reference Axelsen and Nielsen2015). My purpose is to explore sufficientarianism under the assumption that some such threshold is plausible, rather than to defend or criticize any particular version of the assumption. Indeed, a major result of my argument is that the assumption is less important than might be supposed. I argue that even if we accept that there is a non-instrumental, normatively significant sufficientarian threshold, we should reject standard forms of sufficientarianism.

I conclude my discussion of the threshold by considering two important issues regarding it. First, does the threshold mark the point of satiation? Second, is the threshold sensitive to relative deprivation?

According to Lasse Nielsen (Reference Nielsen2019: 806–807),

Satiable-value sufficientarianism presents an interesting theory about how goods work, but it cannot be our focus here. The high priority given to attaining the threshold, and the accordingly low priority given to benefits in excess of the threshold, is the distinctive feature of sufficientarian theory, at least as the vast majority of the literature understands it. This priority has no application if the threshold cannot be exceeded. To examine sufficientarianism we must therefore allow that the threshold can be exceeded.

Some sufficientarian theories are sensitive to relative deprivation. Huseby (Reference Huseby2010: 182), using a contentment-based threshold, writes that ‘there is reason to think that it is harder to be content if, for instance, many are much better off’. Axelsen and Nielsen (Reference Axelsen and Nielsen2015: 407) note that ‘relative deprivation may, in important ways, influence one’s absolute level of freedom and, thus, make one’s freedom insufficient in absolute terms’. Nussbaum’s concept of ‘multiple realizability’, according to which her basic capabilities are to be ‘concretely specified in accordance with local beliefs and circumstances’, is sensitive to similar concerns (Nussbaum Reference Nussbaum2000: 77). In practical terms, these considerations may mean that sufficiency is ‘moderately averse to inequalities’ (Huseby Reference Huseby2010: 178), because inequality may lower the worse off’s contentment, freedom or capabilities. I assume throughout that sufficientarianism is in this way sensitive to relative deprivation. But the key point for present purposes is that this inequality aversion is empirically contingent. While sufficientarianism may have something of an empirical tendency to favour equality, this is only insofar as it actually promotes sufficiency; where sufficiency is best achieved by a more unequal distribution, it has no inequality aversion at all. Likewise, though I set the threshold at 100 for presentational convenience, this is quite consistent with it varying in the real world across time and space on account of relative deprivation.

In short, then, I assume that there is a non-instrumentally significant sufficientarian threshold, which does not mark the point of satiation, and which is sensitive to relative deprivation. In the following sections I present four versions of the excessiveness objection. If we accept even one of these objections, we should reject sufficientarianism.

3. Magnitude of Advantage

In this section and the next I present two versions of the objection that point to sufficientarianism’s inefficient character. A view is inefficient if it foregoes large benefits or advantages (I use these terms interchangeably) for the sake of small ones.Footnote ⁷ Benefits are the ‘currency’ or ‘metric’ of the theory (typically welfare, resources, or capabilities – my examples are indifferent between these measures). That a theory is inefficient in this sense is not in itself much of an objection to it – any non-utilitarian theory is to an extent inefficient. But I suggest that the particular forms of inefficiency that sufficientarianism entertains are unusually objectionable. Many prioritarians and pluralistic egalitarians would accept large inefficiencies in exchange for strongly skewing distributions towards the worse off or making them much less unequal. But this proportionality is alien to sufficientarianism. It takes even the most minimal improvement in its favoured dimension to justify massive inefficiency. This is true also of absolutist versions of egalitarianism and prioritarianism, but that makes sufficientarianism’s absolutism no more palatable.

Sufficientarianism implies that, if the threshold is 100, and we can have individual A at 200 units and B at 99 units (200, 99), or both at 100 units (100, 100), we should bring about (100, 100). That is, it says that, of the options in Figure 1, we should prefer the second world, as more people are at the threshold (indicated by the dashed line). This is counterintuitive (Arneson Reference Arneson and Kaufman2006: 29; Casal Reference Casal2007: 298, 317). We can refer to this example as Magnitude of Advantage.

Figure 1.

Magnitude of Advantage.

Something that might be said in favour of the second world is irrelevant here. The second world is one of pure equality, whereas the first world contains a large inequality. But the sufficientarian cannot present that as a reason for favouring the second world. When considering whether the sufficientarian’s approval of the second world is plausible, we must discount egalitarian considerations, as the sufficientarian discounts them, at least in cases where, as here, the lexically prior issue of providing enough is still in play.

Magnitude of advantage cases are the best known version of the excessiveness objection. Current sufficientarian theories have been developed, in part, to avoid them. It will be instructive to survey two variants of sufficientarianism that might be seen to resist magnitude of advantage cases.

First, some sufficientarians endorse conceptions of advantage that are not reducible to a single measure of advantage, contrary to the apparent assumptions of magnitude of advantage cases. The most prominent theories of this sort are capabilities sufficientarian theories. Nussbaum (Reference Nussbaum2000) aims to secure ten ‘central human capabilities’, Anderson (Reference Anderson1999) aims to provide access to ‘three aspects of individual functioning’, while Axelsen and Nielsen claim that there are ‘distinct and separate thresholds within different aspects of human life that are all elements in being free to succeed’ (Reference Axelsen and Nielsen2015: 413–414; see also Nielsen and Axelsen Reference Nielsen and Axelsen2017). But it is a simple matter to show that these forms of sufficientarianism also disregard magnitude of advantage. Making up the smallest shortfall in any one of the basic capabilities would justify foregoing a benefit of any size whatsoever for someone initially only very slightly better off who is at the threshold. Indeed, magnitude of advantage-type cases are likely to be even more prominent under capabilities sufficientarianism than under other sufficientarian views, as the former have a threshold corresponding to each basic capability. Each threshold is taken to justify foregoing massive benefits for the sake of tiny ones.Footnote ⁸

Second, Crisp Reference Crisp(2003 a: 758) might deny that his view disregards magnitude of advantage by pointing out that, on his view, only ‘nontrivial benefits’ can justify assigning absolute weight to providing enough. On this view, if we find (200, 99) clearly superior to (100, 100), that is because the difference between 99 and 100 is trivial and as such lacks absolute weight. However, the difference between a trivial and a nontrivial benefit may be very small indeed (Holtug Reference Holtug2010: 231). It is hardly plausible that an infinitesimal increase in the size of a below-threshold benefit is sufficient to change it from carrying almost no weight (because it’s ‘trivial’) to carrying greater weight than any amount of above-threshold benefit. Crisp’s view remains implausibly insensitive to the magnitude of advantage.

4. Number of Beneficiaries

Sufficientarianism has another troublingly inefficient implication. Suppose that we can choose between a first world that contains 200 units of advantage for ten people and 99 for one person, and a second one that contains 100 for all eleven people. This is represented in Figure 2, where the width of the bars indicates the number of people that fall in the categories of A (now a group) and B (still one person). Sufficientarianism is of course committed to favouring the second world, as it brings an extra person to the threshold level. In light of the fact that the first world is better for ten times as many people as is the second world, this is counterintuitive (Arneson Reference Arneson and Kaufman2006: 29; Casal Reference Casal2007: 298–299, 313–314). Call this Number of Beneficiaries.

Figure 2.

Number of Beneficiaries.

There are more extreme versions of this problem. Indeed, we can increase the number of people in A by any amount whatsoever – even so it includes billions of people – and sufficientarianism will still endorse the second world. It is not plausible that we should concentrate our efforts on providing a tiny benefit to one individual, the rest of humanity be damned.Footnote ⁹

It might be asked why the number of beneficiaries matters. What is it about benefiting ten people, or ten billion, that makes it, ceteris paribus, preferable to benefiting one? Perhaps the intuitive phenomena in support of favouring more rather than less are so strong that an argument hardly needs to be given. But as some disagree (Taurek Reference Taurek1977), a little more might be said. The basic idea is, as Derek Parfit puts it, that ‘[e]ach counts for one. That is why more count for more’ (Parfit Reference Parfit1977: 301). This in turn appeals to the even more basic idea that ‘equal concern and respect’ should be shown to every person – an idea that is ‘highly abstract’ and that can be appealed to by most modern theories of justice (Dworkin 1975: 50, Reference Dworkin1983: 24–25). Sufficientarians presumably have their own interpretation of equal concern and respect as implying special concern for those below the threshold. The present complaint, then, could be described as saying that sufficientarians provide an impoverished sense of equal concern and respect, according to which the interests of many – perhaps all of humanity but one – are liable to be disregarded on account of the absolute priority for those below the threshold.

It may seem that my assumption that we should take into account the number of beneficiaries is problematic in the light of cases in which a large disadvantage for a badly off person can be prevented, but only at the cost of small disadvantages for a large number of well-off people. Crisp writes that

Huseby (Reference Huseby2010: 187; see also Huseby Reference Huseby2020: 217) similarly remarks that weighted prioritarianism ‘allows that one person’s excruciating and lasting pain can be outweighed by a large number of minor disappointments’. This is, he says, implausible.

There are, I believe, two good reasons for thinking that the aggregative assumption of the argument I am making is not the problem it may seem to be. First, it is possible to settle the weightings in a weighted view (be it prioritarian, egalitarian, sufficientarian, or something else) in such a way that, while benefits are still aggregated, there is a strong presumption against severe disadvantages for a few individuals being justified on the basis of minor conveniences for a multitude. Crisp himself suggests the ‘number weighted priority view’, according to which ‘the number of beneficiaries matters less the better off they are’ (Crisp Reference Crisp2003 a: 754). While benefits are still aggregated on this view, adding more better off beneficiaries adds little to the value of a distribution. Several other modified versions of aggregation have been suggested (see Casal Reference Casal2007: 320–321; Hirose Reference Hirose2014; Voorhoeve Reference Voorhoeve2014; Tomlin Reference Tomlin2017; Lazar Reference Lazar2018; Privitera Reference Privitera2018).

Second, as Nils Holtug contends, in some cases we are quite willing to accept a very high risk of excruciating pain or even death for a small number of people, in order to achieve small benefits for a large number (Holtug Reference Holtug2010: 236–237). Building projects such as major bridges seem to have this feature: some will predictably suffer death or serious injury, all for the sake of a shortened commute for many more. If the critic cannot identify some feature which makes aggregation more objectionable than ordinary practice, they seem committed to requiring that major building projects are usually gravely unjust (see Widerquist Reference Widerquist2010). That seems a less plausible position to take than the one I am taking.

Opponents of aggregation might counter that bridge building typically provides large benefits as well as small benefits, as it enables people to lead lives that they otherwise could not, and that without such investments we might all suffer. It therefore does not count against the argument that we should not risk major hardships for the sake of many small benefits. It can further be claimed that the widely discussed transmitter room case, in which we must choose between relieving an individual of electric shocks (a large disadvantage for one person) or maintaining live coverage of the World Cup final (a small advantage for many), is a better test of our intuitions.Footnote ¹⁰

I agree that intuitively it seems that we ought to avoid the electric shocks, and that bridge building may confer large benefits. But there are other cases in which many small disadvantages clearly can outweigh large, sufficiency-threatening disadvantages. Suppose, for instance, that a large airport proposes to replace its shabbiest terminal with new facilities. This will make no difference to the routes or frequency of flights, and employment levels will be unchanged even in the short term – were the existing terminal to be retained, it would require a labour-intensive overhaul just to keep it standing. The only benefits of the new terminal are that each of the ten million passengers that pass through it per year will experience a shorter queuing time and more pleasant surroundings. It therefore realizes no large benefits for particular individuals, nor does it relieve anyone’s suffering. The health and safety assessment shows that, while the overhaul to the existing terminal could be conducted with zero risk of serious injury, a new terminal building will probably result in a handful of serious injuries during construction. Nevertheless, I think that, intuitively, we would be morally justified in going ahead with the new terminal, on the basis that the risk of unintended suffering and insufficiency for a small number of construction workers is counterbalanced by the convenience afforded to many millions of passengers.

It might be objected that in the airport example, those reduced below sufficiency had chosen to expose themselves to risk (i.e. by being construction workers), and that it is only that extraneous feature of the case that makes aggregation acceptable. While that might be a valid concern for some opponents of aggregation, it cannot coherently be raised in support of sufficientarianism which is, as we shall see, committed to preventing insufficiency even where those below the threshold are responsible for being there.

The airport intuition seems hard to reconcile with our reaction to the transmitter room case. I suggest that this is on account of misleading features of the transmitter room case. As Iwao Hirose suggests, it ‘obscures several important assumptions and implicitly manipulates people’s intuitions’, for instance because ‘we know the identity of the victim in the transmitter room whereas the World Cup viewers are unidentified masses’ (Hirose Reference Hirose2014: 46, 48). By contrast, the cognitive bias towards an identified victim and against statistical victims is absent in the airport case as victims and beneficiaries alike are anonymous. We should also recall the familiar point that intuitions are an untrustworthy guide where very large numbers are involved (Broome Reference Broome2004). Where such distortions are not present, as in Number of Beneficiaries, aggregation seems acceptable. I conclude, therefore, that there are strong grounds for the position that we should take into account the number of beneficiaries, and that sufficientarianism is unacceptable in this respect.

5. Responsibility and desert

In this section and the next I present two more versions of the excessiveness objection to sufficientarianism. These focus on sufficientarianism’s unfair character. While the complaint of inefficiency concerns the amount of advantage in a distribution, the complaint of unfairness concerns how advantage is distributed.

In this section I consider responsibility and desert, which have been thought relevant to distributive justice in several different ways. The kind of responsibility that is our focus here is agent responsibility or ‘responsibility as attributability’, which tells us which things an agent is responsible for bringing about (Scanlon Reference Scanlon1998: 248). Luck egalitarians seek to equalize advantage, except insofar as individuals are responsible for raising or lowering their advantage levels (Arneson Reference Arneson1989; Cohen Reference Cohen1989; Knight Reference Knight2009; Lippert-Rasmussen Reference Lippert-Rasmussen2016). Desert theorists hold that ‘in so far as one lives well, one deserves to fare well, and in a fully just world one would fare well’ (Temkin Reference Temkin, Knight and Stemplowska2011: 53; see also Feldman Reference Feldman2016). Luck egalitarianism allows the upshot of choices to stand.Footnote ¹¹ Desert is usually (but not always) construed as rewarding choices or characters that are directed towards moral goals (see Arneson Reference Arneson2001, Reference Arneson, Knight and Stemplowska2011; Vallentyne Reference Vallentyne and Olsaretti2003).

Fortunately, as sufficientarianism standardly disregards agent responsibility and desert, we do not need to pick out any one feature of individuals’ actions as the distributively relevant one.Footnote ¹² To illustrate the point, consider a simple world in which half of the population make choices that both create benefits for each of them and further moral goals. For instance, we can imagine that they work tirelessly in rewarding and socially beneficial jobs. The other half of the population have made choices that have bad prudential results and that are contrary to morality. Suppose, for example, that despite being naturally talented, they choose to lounge around, thereby denying both themselves and the wider world of the benefit of their talents. Or more extremely, we can imagine that they deliberately burn down their own houses. The members of both groups are responsible for making the described choices, and initially had identical abilities. Members of the first group could have easily chosen to be members of the second group, but they did not, and vice versa. Following one of Larry Temkin’s examples, call these groups saints and sinners, respectively (Temkin Reference Temkin1993: Ch. 9).

Suppose we can choose between a first world that contains 200 units of advantage for the saints and 99 for the sinners, and a second world that contains 100 units for the saints and 200 for the sinners. The saints deserve (or are, on responsibility grounds, entitled to) 200 units of advantage, as indicated by the shortest dashes on Figure 3, and the sinners deserve 90 units of advantage, as indicated by the longest dashes. Sufficientarianism is committed to saying that the second world is better than the first world, as it has a net increase in those at or beyond the threshold. But it is deeply implausible that a small increase in advantage, even for people below the threshold, is sufficiently valuable for those who are less deserving (and responsible for bringing about bad things) to receive more than twice as much benefit as those who are more deserving (and responsible for bringing about good things). Call this case Responsibility and Desert.

Figure 3.

Responsibility and Desert.

Distributive justice is not served by giving the saints 100 units (half of what they deserve) and the sinners 200 units (more than twice what they deserve), where we could instead give the saints 200 (exactly what they deserve) and the sinners 99 (slightly more than they deserve). Yet this is what sufficientarianism requires. The threshold is not so important that just any benefit below it justifies a severe violation of considerations of responsibility and desert. This is particularly clear if it is the case that a third world, where the saints and sinners both have 200 units, would have been available, if only the sinners had not frittered away the necessary resources (Arneson Reference Arneson2000: 348–349; Temkin Reference Temkin2003 b: 772–773). When assessing fair shares, what people have done matters, and not just (as sufficientarianism might allow) as a tiebreaker.

Sufficientarians are aware that their treatment of responsibility and desert strikes many as counterintuitive, but they have not appreciated the severity of the problem. Robert Huseby considers ‘a situation in which I am badly off, but you are even worse off. My present state is completely due to bad luck, while you are responsible only for that portion of your shortfall that distinguishes your state from mine’ (Huseby Reference Huseby2010: 193–194). Huseby suggests that a unit of advantage that can either bring the very badly off person up to the level of the merely badly off person, or improve the merely badly off person’s position, should be used in the first way, even though the very badly off person is responsible for their shortfall relative to the merely badly off person. I may well agree. If the two people’s initial advantage levels were at, say, 20 and 21, and desert (or responsibility) scores at 20 and 21, it is not obvious whether it is more important to give the extra unit to the worse off (and below-threshold) person, or to the more deserving person. Both things matter, and where we can provide a modest improvement in one at the cost of a modest loss in the other, either option is reasonable. But that is not at all the case that is difficult for sufficientarians. The difficulty comes where a modest improvement in the sub-threshold worse off’s position can be achieved only at a huge cost in terms of responsibility or desert sensitivity, as in Responsibility and Desert. Huseby (Reference Huseby2010: 193) does not subject his stance that ‘responsibility considerations come into play only when two individuals are equally badly off’ to the real test.

It should be emphasized that sufficientarians’ disregard for responsibility and desert is not a quirk of the literature. Where absolute priority is assigned to providing enough, as sufficientarianism requires, other values are necessarily relegated to a subordinate role. Near the end of her critical discussion of sufficientarianism, Casal suggests a ‘sufficiency-constrained luck egalitarianism, which allows that some inequalities in outcome may arise justly but denies that individuals’ having less than enough is ever justifiable by appeal to voluntary choice’ (Casal Reference Casal2007: 322). Though this is a more moderate line than most sufficientarians would consider, it still fails to give enough weight to responsibility or desert. Indeed, confronted with Responsibility and Desert, sufficiency-constrained luck egalitarianism would be obliged to favour the second world, in which sinners have twice as much as saints, as that is the only way to ensure that everyone has enough. As I said, this seems problematic.

Two defences of this lexical priority for sufficiency over responsibility might be offered. First, it might be suggested that it would prevent individuals from making bad choices, which would threaten their sufficiency, in the first place (Casal Reference Casal2007: 322–323). There are, however, insuperable difficulties with this response. There are already significant disincentives to making many kinds of bad choice, but that does not prevent many people from making these choices nevertheless. Furthermore, some bad choices, such as a refusal to work, can be disincentivized only at great cost to personal liberty or by threatening the withdrawal of state support (e.g. unemployment benefit payments), which would seem especially unpalatable to sufficientarians. Eliminating all bad choices is not a credible option.

A second defence is suggested by a reviewer for Economics and Philosophy:

The reviewer astutely notes that this assumes a ‘methodological outlook according to which the content of normative theories should depend on (and seek to subvert) the broad contours of the political landscape’, and ‘suspect[s] that at least some sufficientarian theorists downplay responsibility because they want to avoid fanning the fires of individual responsibility’ that are ‘burning our welfare states to the ground’. I believe this outlook may not only explain some sufficientarians’ hostility to responsibility and desert, but also their wider defence of the seemingly unpropitious absolute priority for sufficiency. In reply, I must say (as the reviewer went on to anticipate) that I do not share this methodological outlook. In the first place, it is in the present case self-defeating. Taking a reasonable position (a special interest with ‘those who have too little’) to extremes (where some have little, they can make any choices they like without consequence, everyone else’s interests being irrelevant) does nothing to advance one’s cause. On the contrary, where the left have achieved sustained success, it is by infusing their goals with the concerns of society as a whole, as is in fact exemplified by the welfare state. There is, moreover, the deeper point that the proper purpose of political philosophy is not to bring about pre-ordained political goals. It is rather to question those goals themselves and arrive at accounts of fundamental political concepts, such as justice, that are best supported by relevant reasons. To favour a theory of distributive justice because it is thought to further favoured political causes is to put the cart before the horse.

As sufficiency-constrained luck egalitarianism cannot be successfully defended, it might instead be proposed to constrain sufficientarianism with luck egalitarianism.Footnote ¹³ This view aims to ensure that no one is below the threshold unless they are responsible for being below it. Lippert-Rasmussen (Reference Lippert-Rasmussen2016: 28) presents, without endorsement, a version of this view: ‘it is unjust if some people do not have enough through their bad luck’. But such a position entails rejecting the core of sufficientarianism that providing enough, or as close to enough as is possible, is the lexically prior distributive goal. Indeed, this ‘responsibility-constrained sufficientarianism’ rather requires that considerations of responsibility are lexically prior, as people’s claim to have enough is contingent on these considerations. Responsibility-constrained sufficientarianism does not seem to be proposed in the literature, most likely because it is subject to many of the objections facing luck egalitarianism, including the harshness objection, which highlights that the view is liable to abandon those who are worse off as a matter of their own choice or fault (Fleurbaey Reference Fleurbaey1995; Anderson Reference Anderson1999; cf. Knight Reference Knight2015). Given the challenges facing lexical priority for sufficiency (sufficiency-constrained luck egalitarianism) and lexical priority for responsibility (responsibility-constrained sufficientarianism), a more promising approach is to balance considerations of sufficiency and responsibility against each other, as proposed by the ‘responsibility-catering weighted sufficientarianism’ discussed in section 7 below.

6. Above-threshold distribution

If we can bring about a world with individuals A and B at 200 units and individual C at 99 (200, 200, 99) or a world with individual A at 300, and individuals B and C at 100 (300, 100, 100), sufficientarianism will say that we should favour the latter. It will, that is, favour the second world in Figure 4. This is, again, counterintuitive. The first world seems like a fairer world. The second world may seem fairer in one respect (the worst off person, C, is slightly better off), but that is outweighed by the substantial unfairness of the second world’s distribution between A and B, who are both at or above the threshold. The first world contains identical amounts of benefit for A and B, while the second world contains three times as much benefit for A as there is for B. Call this Above-Threshold Distribution.

Figure 4.

Above-Threshold Distribution.

The second world would be disfavoured by many even before they consult any specific principle. But there are also principled bases for this. One principled basis is the inequality in the second world, on account of the large relative difference in A’s and B’s positions. The second world both feels more unequal, and has a much higher Gini coefficient (0.2667 versus 0.1365 for the first world – 0 indicates full equality and 1 full inequality). Another principled basis for objection is the fact that B’s absolute (non-comparative) position is so much worse in the second world than it is in the first. This is offset to some extent by the absolute positions of A and C being better in the second world. But these are unlikely to outweigh the loss to B given that B is much worse off than A, and the benefit to C is very small.

The above-threshold distribution objection is, as far as I know, novel.Footnote ¹⁴ It is related to, yet far more powerful than, the indifference objection mentioned in section 1, which for the sake of comparison we should now briefly revisit. Shields offers this definition: ‘The Indifference Objection: sufficiency principles are implausible because they are objectionably indifferent to inequalities once everyone has secured enough’ (Shields Reference Shields2012: 104). Casal notes that, ‘[e]ven when everyone has enough, it still seems deeply unfair that merely in virtue of being born into a wealthy family some should have at their disposal all sorts of advantages, contacts, and opportunities, while others inherit little more than a name’ (Casal Reference Casal2007: 311). Temkin asks us to ‘[s]uppose, for example, that two people with “plenty” both applied for a job. Would it not matter if we discriminated against one of them on the basis of his race or religion? Surely it matters’ (2003a: 65–66). In these and many other examples in the literature, sufficiency views are objected to on the basis that they are indifferent to unfair inequality provided everyone has enough. And this is indeed a strong objection to those sufficiency views, such as Frankfurt’s and Crisp’s,Footnote ¹⁵ that ignore inequalities in these circumstances. But it is a simple matter to accommodate this concern by instead endorsing sufficientarianism, which allows that equality or priority may be invoked once sufficiency is satisfied.Footnote ¹⁶

The above-threshold distribution objection contrasts with the indifference objection both in the circumstances in which it applies, and the nature of its complaint. The indifference objection only applies where everyone is assured of enough (e.g. 200, 200; 300, 100); its complaint is that in such circumstances sufficiency is indifferent between equality and inequality. The above-threshold distribution objection applies where everyone is not assured of enough: this covers cases where we can choose whether everyone will have enough (as in Above-Threshold Distribution, i.e. 200, 200, 99; 300, 100, 100) and those where some will certainly not have enough (e.g. 200, 200, 98; 300, 99, 99). And its complaint is that in such circumstances sufficientarianism requires severe violations of above-threshold distributive justice. This objection, unlike the indifference objection, cannot be sidestepped with the sufficientarian strategy of allowing supplementary distributive principles to be applied provided enough has been secured. This is because the supplementary principles have no application in the circumstances in which the objection applies (i.e. where some may not have enough). In cases such as Above-Threshold Distribution, supplementary principles are not consulted as the selection of the objectionable second world is already determined by the lexically prior sufficientarian objective. The above-threshold distribution objection applies with just as much force against sufficientarianism as it does against the doctrine of sufficiency.

Sufficiency theorists do not seem to have directly addressed this problem for their view, presumably because they have, to date, been confronted with the issue of above-threshold distribution as it is manifested in the less troubling indifference objection. They have, however, provided an indirect defence of their view, by objecting to views such as prioritarianism that are concerned with above-threshold distribution.Footnote ¹⁷ The best known such objection is Crisp’s Beverly Hills case, in which fine wine is to be distributed to well-off individuals. We can benefit 10 ‘Rich’ by two units each, or 10,000 ‘Super-rich’ by two units each. Crisp notes that one form of prioritarianism would prefer to give the wine to the Rich, which is ‘absurd’. He goes on to claim, more strongly, that ‘what the Beverly Hills case brings out is that, once recipients are at a certain level, any prioritarian concern for them disappears entirely’ (Crisp Reference Crisp2003 a: 755).

There are importantly misleading features to this case. The name of the case and of its subjects suggests that money is what picks out these subjects as ‘well off’, and that they exist in our world (Temkin Reference Temkin2003 b: 769–770). Crisp also states that they are well off in welfare terms, and clarifies that the Beverly Hills case is intended as a self-contained world. But our intuitions may be influenced by the names deployed nevertheless. This is important for two reasons. First, if the subjects are all very well off in monetary terms, it seems likely that when it comes to advantage – welfare, on Crisp’s account – the Rich and the Super-Rich are equally well off (Crisp Reference Crisp2003 b: 121). In that case, the prioritarian (or egalitarian) can agree that there is no reason to prioritize benefits for the Rich. Second, if the subjects live in our world, a world containing extreme poverty and many millions of easily preventable deaths per year, and they are very rich, it is likely that their desert and responsibility scores are low relative to their advantage levels. And if neither is due the wine – if they are due much lower advantage levels than they actually have – it is hard to feel any pull towards giving the Rich priority when it comes to assigning goods (Temkin Reference Temkin2003 b: 771–772). Luck egalitarians and ‘responsibility-catering prioritarians’ (Arneson Reference Arneson1999, Reference Arneson2000) can explain that intuition.

To really test our intuitions, I propose a different, less messy case. Suppose Advantaged is at 999 units of advantage, making her the second most advantaged person who has ever lived, and Most Advantaged is at 1000 units of advantage, making her the most advantaged person who has ever lived. No one else exists in their world, and both are clearly above Crisp’s compassion threshold. Yet I think we do have more of a reason to help Advantaged than we do Most Advantaged. Not much more of a reason, to be sure, as their advantage levels are so similar and so high, but more of a reason nevertheless. Crisp just states that ‘when people reach a certain level, even if they are worse off than others, benefiting them does not, in itself, matter more’ (Crisp Reference Crisp2003 a: 755; see also Crisp Reference Crisp2003 b: 121–122). I disagree and so do many others, and Crisp gives no real argument to the contrary. Benbaji advances a similar case in support of sufficientarianism but acknowledges that it is ‘by no means conclusive’ (Benbaji Reference Benbaji2005: 316).

It should be added that, by insisting on the irrelevance of above-threshold distribution, Crisp and Benbaji take on board one of the least attractive features of Frankfurt’s doctrine of sufficiency and become vulnerable to the indifference objection. Even those who believe that Advantaged should have no priority over Most Advantaged may change their minds if we add some further features to the story. Suppose that Advantaged has one less unit of advantage than Most Advantaged because a (now dead) third party racially discriminated against her, in favour of Most Advantaged (Temkin Reference Temkin2003 a: 65–66). If I have a unit to assign, it seems surely right that I should give it to Advantaged, undoing the harm of the wrongful discrimination. Huseby responds to this kind of case by claiming that sufficientarians ‘have ample reason to reject such practices without resorting to telic Egalitarianism’, and, I take it, without resorting to (telic) prioritarianism either (Huseby Reference Huseby2010: 191). But deontic restrictions, or the rejection of ‘disrespectful practices’, will not do the job here. To say that the act of discrimination was an act of wrongdoing is insufficient. There is something clearly bad about the outcome in which Advantaged is worse off purely as a result of discrimination that should be captured by our theory of distributive justice. Contrary to the supposed moral of the Beverley Hills case, it makes things better to assist a victim of discrimination rather than the beneficiary of the discrimination, even though both are above the threshold.

7. Weighted sufficientarianism

Providing enough has four counterintuitive results. It treats the smallest below-threshold benefit as justifying (1) the loss of any amount of above-threshold benefit; (2) the loss of benefits to any number of potential beneficiaries; (3) any degree of disconnection between what individuals receive and what they are deserving of or responsible for bringing about; and (4) any above-threshold distributive pattern, no matter how unequal or disadvantaging for the (above-threshold) worse off. If we reject even one of these results, we should reject sufficientarianism. Though it is unnecessary to my overall argument, I believe there are strong grounds for rejecting all of these results.

As we have noted, sufficientarianism is not the only view so insensitive to relevant considerations. Absolute forms of egalitarianism and prioritarianism – those that insist on outcome equality or maximization of the condition of the worst off, no matter the costs – are subject to relatives of most of the above criticisms. Absolute egalitarianism is also subject to the levelling down objection – it will decrease inequality even if this makes everyone worse off.Footnote ¹⁸ We have as strong reasons to reject these views as we do to reject sufficientarianism.

Nevertheless, some versions of egalitarianism and prioritarianism do, I maintain, fare much better than standard sufficientarianism. This is, for instance, true of Arneson’s responsibility-catering prioritarianism. On this view, a distribution is more just, the more it (a) increases the overall amount of advantage, (b) improves the condition of the worse off, and (c) makes distributions sensitive to individual responsibility. These three considerations are to be balanced against each other; none of them has lexical priority. Thus, while the view gives some priority to the worse off, as per (b), this is not absolute and is constrained by considerations of overall well-being levels. The view is therefore sensitive to the magnitude of advantage and the number of beneficiaries. (c) is presumably the basis on which Arneson (Reference Arneson2000) treats his view as a form of (luck) egalitarianism. It shows that the view does not disregard responsibility and desert, as sufficientarianism does.

A successful sufficientarian theory will have to be similarly responsive to a range of distributive considerations. It should reject lexical priority (Benbaji Reference Benbaji2005; Casal Reference Casal2007: 298–299; Dorsey Reference Dorsey2012; Herlitz 2018; Huseby Reference Huseby2020). Shields describes sufficientarianism as holding that ‘once people have secured enough there is a discontinuity in the rate of change of the marginal weight of our reasons to benefit them further’ (Shields Reference Shields2012: 108; see also Shields Reference Shields2016). On this account, it is not a necessary feature of sufficientarianism that lexical priority be given to below-threshold benefits. The importance of the threshold can instead be reflected by assigning less than lexical priority to below-threshold benefits.Footnote ¹⁹ This weighted sufficientarianism is the sufficientarian counterpart of the standard version of prioritarianism. By giving less than lexical priority to below-threshold benefits, weighted sufficientarianism allows them to be balanced against other considerations, just as prioritarianism allows benefits for the worse off to be balanced against other considerations. Weighted sufficientarianism remains consistent with the positive thesis of sufficientarianism; it is committed to the alleviation of deprivation, just not to the exclusion of everything else.

Of particular interest, given the various forms of the excessiveness objection, is a ‘responsibility-catering weighted sufficientarianism’ that says a distribution is more just, the more it (a) provides enough, or as close to enough as is possible, (b) increases the overall amount of advantage, (c) improves the condition of the worse off, and (d) makes distributions sensitive to individual responsibility. (a) is the distinctively sufficientarian aspect of the view, while (b)–(d) ensure that the view accommodates the concerns with magnitude of advantage, number of beneficiaries, above-threshold distributive distribution, and responsibility and desert that standard sufficientarianism neglects.

Given the profound difficulties faced by standard sufficientarianism on account of its absolutist insistence that providing enough is lexically prior to other distributive goals, sufficientarians may well be tempted to endorse weighted sufficientarianism. Just as sufficientarians rejected elements of the Frankfurtian doctrine of sufficiency in the face of the inverted priority and indifference objections, so it might be thought that they can reject the absolutism found in most contemporary sufficientarian writings in the face of the excessiveness objection. Were such a move to be widely adopted by sufficientarians, the plausibility of their theories would improve significantly.

While my main aim in this article has been to encourage a move away from standard, absolutist sufficientarianism, I will mention one challenge for weighted sufficientarianism, concerning the weightings of its distributive values. Consider a crude version of weighted sufficientarianism that says that 1 unit of above-threshold benefit has moral value of 1, while 1 unit of below-threshold benefit has moral value of 101. This view implies that if, as in Magnitude of Advantage (Figure 1), we can have individual A at 200 units and B at 99 units (200, 99), or both at 100 units (100, 100), we should bring about the latter. According to the view under consideration, the first world has moral value of (100*1)+(199*101) = 20,199 and the second world has moral value of 200*101 = 20,200. This version of weighted sufficientarianism thus makes the same judgement as standard (non-weighted) sufficientarianism in Magnitude of Advantage, preferring the second world, a judgement which I earlier suggested was counterintuitive. That some versions of weighted sufficientarianism remain subject to the excessiveness objection may not be surprising, but it does show that shifting to weighted sufficientarianism does not itself guarantee protection against the excessiveness objection. Weighted sufficientarianism must be combined with weightings that do not give too much weight to below-threshold benefits. Moreover, a full weighted sufficientarian theory seems to require a justification for its specific weighting, just as prioritarianism does for its weighting. That standard sufficientarianism needs no such justification is an appealing feature of that view, but it is a feature that weighted sufficientarians must do without.

Prioritarianism captures the idea of benefits to the worse off having greater value by transforming the value of benefits with a strictly increasing, strictly concave function. The most widely discussed prioritarian functions, which have certain theoretical advantages, are Atkinson social welfare functions (Adler Reference Adler2012: Ch. 5). One simple Atkinson function sums the square root of individual advantage levels (Hirose Reference Hirose2015; Ord Reference Ord2015; Adler Reference Adler2019). Weighted sufficientarians will presumably use a prioritarian function to capture the idea of giving priority to the worse off. But as they do not give absolute priority to the sufficientarian threshold, they will have to balance that threshold against the prioritarian function. Weighted sufficientarianism therefore faces the distinctive specification problems of both prioritarianism and sufficientarianism, and a further specification problem of its own. Like prioritarianism it must specify a prioritarian weighting function. Like sufficientarianism it must specify a unique threshold at which individuals have enough. And uniquely it must specify how much weight taking an individual to that threshold has relative to the output of the prioritarian function.Footnote ²⁰

Considerations of parsimony are, of course, far from decisive grounds for rejecting weighted sufficientarianism. But they do underscore that even weighted sufficientarians must have good grounds for placing special weight on bringing individuals to the threshold, grounds that have been assumed rather than argued for here.