Refugees and minorities: some conceptual and normative issues

ABSTRACT In many contexts, states have a duty to take special measures to protect minorities. Does this duty include prioritizing minority over majority refugees? To answer this question, we first show that a vulnerability-focused notion of ‘minorities’ is preferable to a numerical one. Given the vulnerability-focused notion, there is a presumption in favour of prioritizing minority over majority refugees. However, this presumption is sometimes defeated. We identify five conditions under which this is the case. In fact, surprisingly, under special circumstances, states should prioritize certain majority over certain minority refugees.


Introduction 1
The concept of minority status plays an important role in many areas of political philosophy, e.g. multiculturalism, affirmative action, discrimination and hate speech. Take the first example. On the standard understanding, multiculturalism is concerned with the presence of cultural, etc., minorities within a society as well as the special rights, forms of recognition and other political measures that minority status might justify (e.g. Kymlicka 1995;Taylor 1994;Galeotti 2002).
Many think that states ought to give some special weight to the protection of minorities within their own borders. Others think, and, indeed, offhand it appears plausible, inter alia, in the light of how moral theories imply minorities to be treated in an intra-state context, that the states ought to give some special weight to the protection of minorities among refugees (Vitikainen 2020). 2 In both cases, it is generally assumed that the reason why special groupfocused measures are justified is that the relevant beneficiaries form a minority. But what must a minority be for minority status to be a reason for special protection? In this article, we pose this question at the level of non-ideal theory, specifically with regard to refugees, even though some of what we say is applicable to minority protection in general. 3,4 To provide a satisfactory answer to our question, we first need to clarify what a minority is. Given that the distinction between minorities and majorities, literally understood, has something to do with numbers, the intuitively most obvious starting point is a numerical notion of minorities. However, this notion does not track what friends of minority protection are concerned with, and accordingly, we propose an alternative vulnerability-focused notion of minorities (From a numerical to a vulnerability-focused notion of minorities). On that basis, we go on to explicate certain important features of that notion (Dimensions of the vulnerability-focused notion of minorities). Minority refugees: the duty to rescue and the principle of irrelevant goods-The site of vulnerability then offer five reasons why, despite a presumption in favour thereof, somewhat surprisingly, states should not always give priority to minority refugees over majority refugees. Noticeably, this is so even on our vulnerabilityfocused notion of minorities, which tracks the moral concerns underpinning minority protection and, thus, seems more congenial to prioritizing minority over majority refugees than a numerical notion. Conclusion sums up our main claims.

From a numerical to a vulnerability-focused notion of minorities
We begin with the conceptual question of what a minority is. The simplistic numerical minority notion offers the following answer: A certain group of individuals constitutes a minority (/majority) if, and only if, together they constitute less than (/more than) half of the relevant population.
Given a particular background understanding of which population is relevant, this notion gives us intuitively proper classifications of a number of cases, e.g. it implies that LGBT+ people form a minority in Russia. 5 The notion is natural to apply in cases where a population is (believed to be) dichotomously divided into two exhaustive and mutually exclusive groupsthe majority and the minority. Finally, this simplistic notion is straightforward and makes it natural to speak about minorities in the way that it suggests. Take Andrew Altman's characterization of affirmative action as 'positive steps taken to increase the representation of women and minorities in areas of employment, education, and culture from which they have been historically excluded' (Altman 2018;cp. Lippert-Rasmussen 2020, 10-12). Surely, the reason why he uses the formulation 'women and [our italics] minorities' is that women are not a numerical minority.
However, the simplistic notion is over-inclusive. 6 Consider Shia Muslims in Lebanon. They make up 27% of Lebanon's populationa tiny bit more than Sunni Muslims and significantly more than any other Lebanese religious group. It would be wrong to count Shia Muslims as a minority on the purely numerical grounds that they make up less than half the Lebanese population, given that they form a group larger than any other religious group. 7 If we did, we would have to say, unhelpfully, that in Lebanon, everyone belongs to a minority (religious) group. This suggests that we adopt the modified numerical minority notion: A certain group of individuals constitutes a minority (/majority) if, and only if, together they constitute a group, which is smaller (/larger) than the (/second-) largest group in the relevant population.
This notion accommodates the previous counterexample, but it too is over-inclusive. Suppose that Sunni Muslims form 26% of the Lebanese citizens. 8 It is implausible that on this purely numerical ground it forms a minority, given that it is almost as large as the largest group -Shiasand that all other groups are significantly smaller, e.g. Maronite Catholics 20%, Greek Orthodox Christians 8%, and the Druze Community 5%.
At this point, we suggest that we change tracks. While more refined numerical notions of a minority are possible, it is more fruitful to articulate the concern that underpins our responses to the cases that we have described directly. After all, the best explanation of why a certain group of people qualifies as a minority group is, we contend, that they are likely to be vulnerable in a way that other groups are not. Call this comparative vulnerability. In light thereof, we suggest the vulnerability-focused notion of minorities (/majorities): A certain group of individuals constitutes a minority (/majority) if, and only if, its members are significantly more (/less) vulnerable compared with members of other relevant groups in a way, which is morally significant. Vulnerability, whether comparative or non-comparative, can be generic or relational; the latter being a specific form of generic vulnerability. An agent suffers generic vulnerability if he or she 'suffers a substantial risk of a significant loss in the relevant metric (welfare, resources, capabilities)' (Vrousalis 2013, 133). Non-relational vulnerability contrasts with, for example vulnerability due to threats from other agents and is sometimes relevant in cases of natural forces, e.g. drought. 9 Relational vulnerability necessarily involves reference to the power of other agents and is thus often political or social. B is relationally vulnerable to A if '(i) B lacks some desideratum x that is a requirement for, or a constitutive feature of, B's flourishing (in which case x is the object of B's need), (ii) B can only obtain x from A, and (iii) A has it within his discretion to withhold x from B' (Vrousalis 2013, 134). The notions of generic and relational vulnerability need to be filled out by a specification of which metric is morally 6 We also believe that it (and the other numerical notion we introduce below) is under-inclusive, but we postpone this issue until later in this section. 7 This is not to deny that we might do so on other grounds. 8 No official census has been made since 1932. Hence, reliable accurate figures are unavailable. 9 While anthropogenic climate change refugees are vulnerable as a causal result of the power of agents and how they have exercised it, i.e. by emitting greenhouse gases, their vulnerability, e.g. to flooding, can be described without any reference to the power of agents.
relevant and which needs are necessary for flourishing. Generally, our discussion focuses on minorities afflicted with relational vulnerabilities. If, plausibly, vulnerability is often linked to group features, such as ethnicity, religion or gender, this gives us a pro tanto reason to address the predicament of such a group. In the first instance, vulnerability is a property of individuals and accordingly provides reasons for the protection of these individuals. Nevertheless, we focus on vulnerabilities linked to groups. This is so for two reasons. First, most of the types of vulnerabilities that matter are group related, e.g. threats to basic security are often due to membership of ethnic, religious or other types of socially salient groups. Secondly, the question of this article concerns justifiable decision procedures at the level of non-ideal theory rather than ultimate criteria of rightness in ideal theory. Such real-world decision procedures have to focus on more practical criteria, such as group membership, even if individual vulnerability is what ultimately matters.
The vulnerability-focused notion of minorities has the welcome implication that the fact that a certain group forms a minority is a pro tanto reason to do something to address their predicament. So, the move to the vulnerability-focused notion of minorities not only addresses the intuitive problems with the numerical notions described above, it also provides a rationale for linking minority status to special protection. Still, one might ask why we want our notion of minorities to focus on vulnerability rather than, say, oppression or domination. We take domination, e.g. on the republican understanding as one agent's arbitrary or discretionary power over another agent (Pettit 1997), to be one form of relational vulnerability. Oppression is a more clearly moralized notion and is sometimes used more or less interchangeably with domination but also often as a structural phenomenon where the reference to particular agents is not required. Vulnerability includes many forms of domination but is a broader notion that can also include threats due to nonhuman factors. Accordingly, we conjecture that the concerns, which guide our classificatory intuitions with regard to the minority-majority distinction, are captured by the vulnerability-focused notion of minorities. Obviously, however, vulnerability in this sense and numerical size can be pried apart. This forms the basis for two potential challenges to our vulnerability-focused notion of minorities.
First, if minority status is not conceptually tied to the numerical size of a group, why use the term 'minority' as opposed to, say, 'vulnerable groups'? This challenge can be answered. It is possible to explain why 'minority', e.g. in the context of refugee protection, is an apt term, even if purely numerical facts do not form the supervenience base of minority status. Sociologically speaking, generally larger numbers translate into less vulnerability of the members of a group, e.g. Sunnis in Saudi Arabia relative to Shias, and, accordingly, group size is highly negatively correlated with vulnerability, even if sometimes they come apart, e.g. as when white South Africans were a small, non-vulnerable numerical minority under Apartheid. In short, the normative concern underpinning minority rights is not conceptually tied to being a numerical minority, even if it so happens that the relevant concern is rarely triggered in the case of numerical majorities but often triggered in the case of numerical minorities, thus, making the term 'minority' an apt one in most contexts involving the protection of vulnerable groups.
Secondly, if minority status is conceptually tied to vulnerability and if, as seems conceptually possible, a numerical majority may be vulnerable relative to a numerical minority, we should, in such a case, say that the latter forms a minority in the relevant sense and is in need of minority protection. While this may seem strange in purely linguistic terms, there are historical cases where this has occurred, e.g. in relation to colonialism. In such cases, special rights protecting a numerical majority could naturally be seen as cases of minority rights. Suppose that the majority of people in a country consisting of indigenous people recently colonized by a numerical minority of European settlers find themselves in a precarious situation where their language and culture are in danger of perishing. In this situation, labelling special laws protecting indigenous culture and language 'minority rights' would seem linguistically appropriate, even when they form a numerical majority in the relevant state. 10 However, there is a unified and more basic reply to both challenges just addressed, to wit that, from a moral point of view, the interesting issue is not whether to chart the proper use of the term 'minority', which is probably not completely fixed in ordinary language. The interesting issue is to see if there is a notion of minorities, which is such that it reliably tracks a relevant normative concern (cp. Laborde 2017 on interpretative concepts). We believe a vulnerability-focused notionto be explored further in Dimensions of the vulnerability-focused notion of minoritiesdoes so to a higher degree than a numerical notion of minorities. Specifically, we note that, other things being equal, it seems morally more important to address vulnerabilities of numerical majorities than those of numerical minorities for the simple reason that, again other things being equal, it is morally better to help more rather than fewer people. Hence, if there is a special case for minority protection, that case must derive directly not from the numerical characteristics of minorities but from the special difficulties they face, e.g. their vulnerability. This also means that we make the conclusion we eventually want to establishthat sometimes it is morally permissible to give majority refugees preferential treatmentharder to justify on a vulnerability-focused notion of minorities than on a numerical notion and, thus, more interesting.

Dimensions of the vulnerability-focused notion of minorities
In this section, we unfold a vulnerability-focused notion of what a minority is along six different dimensions. First, while many are inclined to think of minority status as a matter of either-or, the vulnerability-focused notion suggests otherwise; the reason being that vulnerability allows for degrees. We could understand 'significantly more vulnerable' as involving a precisely defined level of comparative vulnerability such that any level of comparative vulnerability, which falls below that level, is insufficient for qualifying as a minority status, whereas any level at or above the threshold qualifies as a minority status. However, any such threshold is arbitrary, and in any case, the normative concern that motivates minority rights tailors the location of the underlying scalar degree of vulnerability. That being said, the latter either-or distinction might still play a useful heuristic role, so we shall set aside the issue of continua of vulnerability. 11 10 Recall also our case of Apartheid where it would seem odd to call rights protecting the completely dominant, numerical minority 'minority rights'. 11 A similar issue arises for numerical notions of minorities.
Secondly, consider next the notion of being 'morally significant in a particular way' (see our vulnerability definition in the previous section). Presumably, we need moral principles to tell us when a certain vulnerability is morally significant. In this sense, our vulnerabilityfocused notion of minorities is a 'normatively dependent concept' (Forst 2013, 32-35). One cannot identify minorities in the relevant sense independently of normative considerations and determine, which vulnerabilities matter. Since there are many competing accounts of what the right moral principles are, there are many different competing accounts of which groups form minorities. A right libertarian, for instance, might acknowledge very few minorities compared to egalitarians (cp. Kukathas 2007;Kymlicka 1989). If, to mention a third possible normative principle, our fundamental concern is with equal civic status and that citizens can be assured that their equal status will be respected, the relevant minority might be those who are especially vulnerable to having their equal status put in question (cp. Waldron 2013 on hate speech). Our general concept of minority status does not tell us which of these conceptions is best, and we invite the reader to simply assess our arguments on the basis of those vulnerabilities that he or she finds morally relevant.
Thirdly, on the vulnerability account, which groups form a minority is context sensitive in that it depends on the empirical facts of the relevant case, and different normative principles will pick out different empirical facts as being relevant. Jews in Tunisia, for instance, might form a (tiny) minority, even if Jews in Israel do not. More generally, when certain groups are thought of as minority groups, it is because we assume a particular context in which these groups are subjected to certain comparative vulnerabilities (cf. the definition of vulnerability above and Sally Haslanger (2000, 41) on gender and race).
Fourthly, consider next the qualification 'in a particular way'. Many groups are vulnerable in many different ways. Coal miners are vulnerable to their industry becoming obsolete in response to climate change. Fashion designers are vulnerable to changes in fashion, which render their particular creative skills irrelevant. Yet, we would not say that coalminers and fashion designers are minorities in the relevant sense, even though, obviously, they are both vulnerable in certain ways. These ways are relevantly different from the ways in which racial, religious, political or sexual minorities are vulnerable. While fashion designers, unlike coal miners, might be more vulnerable than other groups to changes in fashion, they are not vulnerable to being worse off overall than most other groups. After all, fashion designers are fairly well off and secure on most socio-economic dimensions, e.g. income and status, and not just on any vulnerability, but the vulnerability to being worse off than other groups on a suitable measure of overall advantage is crucial to minority status. Moreover, the means, which one could use to address the vulnerability of fashion designers and coalminerse.g. funding of retraining programmes and unemployment benefitsare very different from the sort of means that one typically employs to address the vulnerabilities of minorities, e.g. promoting small languages by making them official languages in which the citizens can communicate with the state. There seems to be little reason to address the vulnerability of fashion designers and coalminers through group rights, group recognition or other groupspecific measures that take up much of the debate in the literature on multiculturalism. 12 Similarly, there are some vulnerabilities, e.g. vulnerability to extreme weather events, which make less sense to mitigate by granting asylum than others do, e.g. vulnerability to religious persecution.
We now turn to the fifth feature of our vulnerability-focused notion of minorities, to wit, that whether a group is vulnerable depends on which other groups are the relevant groups to compare with. Since being a minority, based on our definition, is a matter of being especially vulnerable to certain forms of treatment compared with other groups, we need to specify which other groups to compare with. This essential comparativeness has important implications. 13 Given that our minority concept not only serves a normative function but is also in itself normatively dependent, this will go for the specification of the relevant comparisons as well. Hence, to determine which groups to compare and in which respects for the purpose of deciding whether some of these groups should have minority status, we need to invoke the normative principle that gives normative force to minority status in the first place. This is an important point to be aware of since it makes minority status dependent on the scope of normative principles. In the context of refugees for instance, minority status of a certain group of refugees will depend, inter alia, on whether one compares their degree of vulnerability with other groups of refugees from their state of origin or with refugees in general.
A sixth important feature concerns the site of vulnerability. One suggestion is that we compare the vulnerability of different groups prior to their seeking refuge. Another suggestion is that we compare the vulnerability of different groups qua refugees. These two vulnerabilities might come apart. Members of a religious minority might be subjected to severe persecution in their country and, thus, be much more vulnerable than other groups in their country. Yet, it might be that refugees from this minority find it very easy to be granted asylum in several recipient countries in which their religion forms the majority religion and, thus, are eager to admit co-religionists to flee persecution. Qua refugees, they might be less vulnerable than refugees from the dominant majority in their country.
In this section, we have drawn attention to six features of the vulnerability-focused notion of minorities, to wit, that: 1) its supervenience base is scalar, 2) it is normatively dependent, 3) it is contextual, 4) only certain kinds of vulnerability bear on minority status, 5) it is inherently relational and 6) different sites of vulnerability might matter.

Minority refugees: the duty to rescue and the principle of irrelevant goods
We will now apply the vulnerability-focused notion of minorities to a discussion of the ethics of admission of refugees. We ask if it is always morally justified to give priority to minority refugees over non-minority refugees given our vulnerability-focused notion of minorities. We will assume that states do not admit all refugees that ought to be admitted because, say, in the light of costs, the states do not have a duty to admit all refugees who ought to be admitted, or even though they do, they will not and, thus, priorities have to be made (see footnote 2). We argue that, despite our vulnerabilityfocused notion of minorities, in some cases it is justified, perhaps even required, not to give minority refugees preference over majority refugees.
In the course of defending this claim, we make two assumptions. First, we assume that admitting refugees is a matter of preventing harm to others. 14 This is not always the case. If, say, a state unjustly wages war against another state and in the course of doing so causes citizens of the latter to starve and flee the country, then it fails to prevent its action from actively harming these citizens by way of their precarious nutritional predicament if it declines to admit them as refugees and thereby providing them access to food. Moreover, on the assumption that agents, including states, have special duties to prevent their actions from unjustly harming others, such a state might have a reason to give priority to nonminority refugees from that state over minority refugees from other states fleeing oppression, etc. This is also the reason why we set aside such cases. It seems too obvious that in such cases, states may have a reason not to give priority to minority refugees for the predicament of which it bears no responsibility over non-minority refugees whose predicament it wrongfully caused.
The second assumption we make is that in cases where an agent is under a duty to help others whose predicament the agent bears no responsibility for, there is a prima facie reason to help as much as possible. Suppose that you are on the ocean in a boat with plenty of space. Suppose also that you discover some people to the left who are drowning and an equal number of people to the right who wears life vests and is able to swim to an island close by but who will later suffer from significant, lifelong health problems due to their ordeal in the water. Assuming that you have a duty to rescue but cannot rescue more than one group, you have a duty to rescue the group of people who will lose the most if you do not help them. On the assumption that a state's duty to admit refugees can be thought of as a duty to rescue, the state should similarly admit refugees such that by doing so, it helps the most. Given our vulnerability-focused notion of minorities, this would seem a good reason to admit minority refugees rather than majority refugees.
We believe that in many cases, this assumption is warranted. Suppose that a totalitarian state is such that all groups of citizens are subjected to human rights violations. However, the totalitarian state is particularly systematic and brutal in its human rights violations of members of a particular religious minority. In this case, recipient states ought to give priority to the minority refugees from that state over the non-minority refugees, given that it will not admit all refugees from that state, which it ought to admit. Consider now a different case. A totalitarian state severely violates the human rights of the citizens it suspects of opposing it. It does so largely, but not completely, irrespective of group membership. Members of a certain linguistic minority experience slightly worse repression than members of the majority linguistic community do, e.g. they are slightly more likely to suffer arbitrary detention. Though all citizens of the totalitarian state suffer severe repression, members of the relevant linguistic minority suffer slightly worse repression, and, thus, giving priority to the linguistic minority refugees would help more when we hold the number of refugees admitted fixed, so, along the lines described above, there would seem to be a prima facie reason to favour the former minority group. However, this case might be an exception from the presumption in favour of minority refugees. All the refugees in question flee their country to avoid having their basic human rights violated, and fairness arguably requires that all get an equal chance of enjoying this basic good. Members of the linguistic minority will benefit slightly more from being admitted than non-minority refugees, given that they face slightly worse repression, but that extra benefit is what, in a bioethical context, Kamm (2013, 424-426) has labelled 'an irrelevant good'. 15 If we embrace this view, it is difficult to see why a similar view should not apply in our refugee case. In short, in cases where minority and non-minority refugees flee severe repression, and the former group experiences only slightly worse repression, we should, all other things being equal, give equal priority to minority and non-minority refugees. One implication of this is that when we must prioritize between male and female refugees who flee from the sort of severe human rights violations typical of prolonged civil wars, e.g. prolonged torture and arbitrary detention for extended periods of time at unknown detention facilities, one should not give priority to female refugees from such countries, if in addition to being vulnerable to such gender-neutral human rights violations, they are also subjected to comparatively mild forms of sexist discrimination on the labour market, e.g. implicit gender biases, which means that sometimes a slightly worse qualified male applicant is selected instead of a slightly better qualified female applicant. 16

Creating and downsizing minorities
In this section, we explore how refugee policies might affect which minorities exist and their size such that states should not always prioritize minority over non-minority refugees out of a concern for minority protection. First, giving priority to minority refugees might affect the situation of minority members other than the refugees. 17 One way it could do so is by downsizing the minority in the country from which minority members flee, thuslet us supposemaking the remaining members even more vulnerable. In this case, not giving priority to minority refugees might be better for non-fleeing minority members, in which case a concern for minorities might be a reason not to give priority to these refugees among all refugees. In fact, it could be a reason to give priority to majority refugees since doing so might over time turn what is presently a minority into one that is not. Some might accept this point but suggest that while in principle it is correct, its relevance for real-life cases is dubious. We concede that while there might be some truth in this reply, it does not take sufficiently into account the cumulative effects of migration over time. The situation of Jewish and Christian minorities in the Middle 15 Suppose that a doctor can save only one of the two patients from a life-threatening infection. As a side effect, the treatment will also cure one patient of a minor skin complaint, but not the other patient. In light of the relative insignificance of being relieved of the minor skin complaint compared to having one's life saved, Kamm thinks that the extra benefit the doctor could bring about by treating the former patient should not determine whom he or she treats. He or she should flip a coin, giving each of them an equal chance. The good of being cured for a minor skin complaint is an irrelevant good when it comes to choosing which of the two patients to save from death. 16 Kamm's 'Sufficient Good Only Option Argument' motivates a stronger claim (Kamm 2013, 436-441). Suppose that minority as well as non-minority refugees flee mortal danger. Suppose moreover that majority refugees will be oppressed in various ways if admitted by the state to which they have fled, whereas minority refugees will not. (Suppose that we cannot eliminate this wrongful oppression, but we can affect who are admitted.) In this case, giving priority to minority refugees would bring about more good. However, arguably we should not give priority to minority refugees, given that everyone's life is at stake and that, so Kamm contends, the majority refugees might have as great an interest in surviving as the minority refugees (despite the fact that they will enjoy worse lives if admitted). 17 The present line of argument is parallel to related arguments in the 'brain drain literature', implying that we should assess the permissibility of allowing, say, doctors from poor states to migrate to rich states, partly in the light of how that will affect other groups within poor states (cp. Oberman 2013).
East, where, since the Second World War, many of these minorities have been decimated due to migration in large part, illustrates this. 18 In any case, migration might not only affect minorities in the countries of origin but also in recipient countries, and here we think that relevance to real-life cases is more apparent. Accepting refugees might affect which groups form minorities in the recipient countries. Take Lebanon again. It is a special case in the sense that refugees or descendants of refugees do not get Lebanese citizenship. But suppose they did. Given the massive influx of refugees from Syriaapproximately 1.5 million in a country with 4.5 million citizensgiving priority to minority refugees from Syria would significantly affect the relative sizes of different religious groups in Lebanon. Given that, by far, the most Syrian refugees are Sunni Muslims, we could imagine, for instance, a situation where giving priority to minority refugees from Syria -Sunnis form the numerical majority in Syria, but given that the regime is Alawite, arguably they form a minority on our vulnerability-focused notion of minoritieswould over time turn Lebanese Shias into a minority group (or, if you think they already are, making them even more of a minority group). If so, it might be permissible for Lebanon not to give priority to minority refugees in the interest of avoiding creating a new minority in the recipient country, which, if history is any guide to go by, would then become more vulnerable to various forms of oppression, neglect, etc. 19 In short, giving priority to minority refugees will not only affect refugees applying for asylum. It will alsoand especially as an accumulated effect of demographically significant migrationaffect which minorities exist and how large they are, and in turn, these two factors will affect the level of oppression of minorities. Hence, in a way, which might initially seem paradoxical, a concern for minorities might not always motivate giving priority to minority refugees.
Some might object that to decide asylum policies on the basis of the sort of demographic considerations is to use refugees as mere means to benefit others (some of whom might not even be existing individuals but are people who will come into existence depending on our choices regarding refugee policies). For example, we reject applications for asylum, even though the asylum seeker's individual situation warrants giving priority to that asylum seeker as part of the policy of aiming at benefiting other members of the minority group to which the asylum seeker belongs.
While there might be some force to this objection, it is weakened by the following two observations. First, our duty not to treat persons as mere means might not be absolute. Perhaps if enough good is at stake, we might justifiably treat some refugees as mere means to protect other members of the minority to which they belong. Secondly, arguably the case we have described does not involve treating minority refugees as mere means. After all, the very rationale for the rule of prioritization is precisely an ex ante concern for the interests of 18 Generally, in consistency with our previous discussion of this matter (cf. From a numerical to a vulnerability-focused notion of minorities), we stress that we are not claiming that reducing the size of a minority, numerically speaking, in all cases increases its vulnerability. However, members of a religious minority are more likely to be vulnerable the smaller it is relative other religious groups, e.g. because of fewer people to push for its particular political concerns being accommodated or interests being promoted. 19 Some might object that Lebanese politicians among others should not take for granted that, say, minorities will be oppressed, etc., like they have been in the past. Rather, they should act to prevent such oppression, etc., in the future. We agree. However, pursuing this objection is simply pursuing a different line of argument than the non-ideal one that we pursue.
all members of the relevant minority, and, possibly, this shows that the relevant priority setting does not involve treating any members as mere means.

Culpable minority refugees
An additional reason for not giving priority to minority refugees appears in situations where minority refugees are complicit in the oppression of the majority in their country of origin. If so, it seems morally perverse to give them priority over majority refugees. Consider Sunni refugees from Iraq who actively supported Islamic State, though without actually committing any war crimes, etc. Suppose, moreover, that a state will admit either some such refugees or an equal number of Iraqi Shia refugees. 20 Suppose, realistically, that, unlike in the recent past, Sunnis now form a vulnerable minority in Iraq. In this case, if we give some moral weight to responsibility, we should possibly not give priority to these Sunni refugees (as opposed to non-complicit Sunni refugees). Arguably, we should not do so, even if they did not play any counterfactual, causal role in the plight of the relevant Shia refugees, who will be granted asylum in their stead. No doubt that people will have different views on the degree of complicity, which will warrant a lowering of the priority in the refugee queue that one is entitled to, but our claim is simply that at least for some such, a possibly high degree lowering this priority is morally required. 21 This is how we think about the duty to rescue in general. Return to our example of the two groups of people who are swimming in the water. If the members of the group who will drown if not rescued are complicit in the situation of the group that can reach land through their own effortssay, they unjustifiably sunk their shipthen the idea of which of the two groups we should rescue changes, i.e. we should now rescue the non-culpable individuals despite the fact that they do not face the threat of drowning but only a lesser, but still sufficiently, serious threat of, e.g. having a limb amputated due to hypothermia.

Vulnerable minorities or vulnerable refugees
In the discussion so far, we have discussed examples of refugees who are vulnerable because they belong to a minority group in their country of origin. However, it might be suggested that the correct focus is on how vulnerable refugees will cope if denied asylum compared with if they are granted asylum. If refugees, who are denied asylum, are sent back to their country of origin, whether they belong to a minority there, generally bears on that question. However, discrepancies could arise as to when having fled a country, this could be taken as a sign of disloyalty to the regime in the country of origin such that returned refugees will now be highly vulnerable whether they are minority or majority people. In such cases, giving priority to minority refugees might be wrong.
Some might object that the present view introduces a kind of objectionable moral incentive to flee. Suppose that initially one has a weaker case for being given asylum than minority citizen from one's country of origin. At least in cases where only a limited number of refugees can be granted asylum, it seems as if majority citizens might have a duty not to flee so as to give the minority citizens a greater chance of getting asylum and, thus, that it would be absurd if, as a result of violating that duty, the majority citizens then acquire a stronger claim to asylum.
We have some sympathy for this challenge. However, we think that it does not apply to all cases. If even majority citizens suffer severe repression, etc., it might be unreasonably demanding to require them not to flee to escape repression so as not to take up places that would otherwise have been allocated to minority refugees. In such cases, we do not consider it morally absurd to think that by choosing to flee, and hence acquiring additional vulnerabilities, one might thereby affect the priority among asylum seekers that one is entitled to.

The site of vulnerability
Consider finally the site of vulnerability. In most discussions of refugees, it is assumed that the relevant site of vulnerability is the nation state from which the refugees are fleeing. However, an obvious alternative is to adopt a global perspective instead. Which groups form minorities might depend on which site of vulnerability we assume. The Yezidis form a religious minority in Iraq, numbering roughly half a million people whose plight at the hands of Islamic State received worldwide attention recently. Suppose, unrealistically, that events following the defeat of Islamic State had resulted in two small areas of Iraq being carved out to form two small, independent states for the Yezidis whose citizens in both states were almost exclusively Yezidis. Hence, in these two states, the Yezidis constitute a majority. However, if we shift our focus to a global level, we might think that the Yezidis form a minority. Globally speaking, the group is extremely small compared to its complements, e.g. Christians and Muslims, and the existence of the two Yezidi states would in all likelihood be very fragile. Suppose that one of these two states degenerates into a despotic regime, resulting in many Yezidis fleeing. A concern for minorities when pitched at a global level could justify the democratic Yezidi state giving priority to refugees who do not belong to a minority in the country from which they flee, i.e. Yezidi refugees from the despotic Yezidic state, over refugees who belong to non-Yezidi minorities from the state from which they flee, e.g. Sunni refugees from Iraq. 22 One way to put it would be in Kymlickean terms, i.e. that in order to preserve their cultural structure, retaining a Yezidian character of the two small nations is important. In the light of recent events, it might also be argued that the Yezidis living elsewhere are saferthey have an exit optionwhen there are stable Yezidi states to which they can flee if persecuted in the countries in which they live.

Conclusion
In this paper, we have argued that numerical notions of minorities are unhelpful from the perspective of a concern for minority protection. In their stead, we have proposed and analysed a vulnerability-focused notion of minorities, which offhand seems much better placed to capture the relevant concerns. However, even given this notion of minorities, which renders giving priority to minority refugees much easier to justify, we have suggested that there are at least five reasons why minority refugees should not always have a higher priority than majority refugees. 23 First, in some cases, the additional vulnerability endured by minority refugees is not significant relative to the vulnerability of non-minority refugees. Secondly, the case of demographically significant refugee migrations giving priority to minority refugees might worsen the situation of existing minorities or create new vulnerable minorities over time. Thirdly, some minority refugees are complicit in making majority refugees flee. Fourthly, strictly speaking, we should be concerned with the vulnerability of people after having fled rather than minority refugees that qualify as such due to their greater vulnerability of members of this group in their country of origin. Finally, if we adopt a global perspective on vulnerability, it might be justified for small groups, globally speaking, within their own state not to give priority to minority refugees when, over time, that will increase their vulnerability. We should end by reminding the reader that we believe there is a presumption in favour of giving priority to minority refugees simply on the grounds that doing so will do more good, other things being equala presumption that will often be borne out by the facts. Still, it might be important to know what might sometimes defeat that presumption.

Disclosure statement
No potential conflict of interest was reported by the authors.