I feel very lucky that my essay has received such strong and challenging responses, but I also feel the pressure to do justice to my seven critics’ thorough comments. Each of them has chosen a different angle of attack. If this were a military encounter, my best strategy might be to let them direct their fire against each other while taking a middle position and ducking my head so that I am not hit by the bullets. Yet this is not a fight but a debate with friends and colleagues whom I admire too much to play such a game.

Instead of replying to each in turn, I will address their comments in the sequence that has provided the structure for my essay. This means that I will not be able to take up all points. However, that approach will allow me best not just to defend my theory, but also to clarify and correct my views where I feel this is necessary. Ideally, I would have liked to rewrite the essay itself, since my critics made me acutely aware of its gaps and weak spots. In a conversation like this one, earlier statements cannot be undone, but they can be modified subsequently when the force of a better argument is acknowledged. I will of course also try to convince my critics in turn. And I sincerely hope that this conversation will continue beyond this book.

The circumstances of democracy

That brings me straight to Joseph Carens's response. Carens's main question is what my question is. Clarifying this is really important for a good conversation. In a nutshell, my question is: Which principles should guide citizens of a democratic polity and their representatives when considering whose interests should count in their political decisions, whom to offer protection, and whom to include in their midst as citizens?

In sections 1 and 2 of chapter 1 I argue that this question can only be meaningfully asked in a bounded polity and that philosophical approaches that try to conjure away political boundaries or aim at delegitimizing them from a critical or ideal theory perspective are misguided. This is not mainly a conceptual point and I acknowledge Carens's critique that I have been less than clear about this. Not too much follows from saying that inclusion conceptually presupposes boundaries. Boundaries can always be stretched to include whatever has been initially left outside: after including all human beings in a global demos, one might consider including animals, all living organisms, or even inanimate things that humans attribute value to and – why not? – immaterial ones such as ideas. My point is rather that questions about inclusion belong to the stuff of which democratic politics is made, just as much as questions about how to deal with a diversity of interests, identities and ideas within a bounded democratic polity.

This seems, then, like an empirical statement and Carens takes me also to task in this regard when he asks whether we cannot imagine a fully insulated society that has never had contact with other human societies and that is still structured democratically. As a philosophical exercise this may be possible but it still seems to me futile given what we know about human history as well as problematic in setting our normative inquiry on the wrong track. Since Carens asks for actual examples of such communities, let's take the case of Iceland, an insulated and until recently ethnically homogenous society that boasts it has the oldest parliament with some continuity until today. Why was that Althing created in 930 AD? In order to bring an end to bloody wars fought between clans through laws that they would adopt jointly to regulate their relations. I suppose that it is only in retrospect and through a statist lens that we anachronistically attribute to the medieval Icelanders the idea of being members of a single self-governing nation or society. I assume instead that the politics of the Icelandic Commonwealth was shaped by external boundaries between clans and villages just as much as by internal conflicts within these, and that the dualism between internal diversity and external boundaries was present in the minds of those who engaged in law-making and politics as an alternative to war just as much as it was in ancient Athens or Rome. In order to grasp the general features of what I call the sphere of the political, empirical theories should always take this dualism into account. Where one of the two aspects is missing, our conception of political life becomes seriously flawed. One prong of this critique applies to international relations theories that consider states as homogenous entities endowed with a singular will while the other prong attacks theories of democracy that fail to consider how all democracies are constituted through their horizontal or vertical relations to other polities.

Neither is my claim in section 2.1 of chapter 1 that boundaries belong to the “circumstances of democracy” merely an empirical one. Consider whether Rawls's idea that a moderate scarcity of resources is part of the circumstances of distributive justice is an empirical or a normative claim. On the one hand, the suggestion is that moderate scarcity belongs empirically to the human condition because, given the malleability of human needs, their satisfaction always breeds new desires. Is there also a normative content in this idea? It is certainly not that justice commits us to maintaining a condition of moderate scarcity but the other way round: moderate scarcity is the condition under which distributive justice is possible and necessary. Yet this implies also a normative message for theorists (and political ideologists): theories of justice must not ignore this condition by stipulating full satisfaction of all human needs in a communist society as a goal of ideal theory. My proposition regarding external boundaries has a similar normative message for political theories of democracy. These go wrong when they assume – as, for example, does Robert Goodin (2007, 2013 ) – that a principle of moral equality of all human beings translates into an ideal of a single world polity that includes all humanity and whose internal boundaries represent only second best accommodations to the problem of how to administer such a polity. Instead, we should consider a horizontal and vertical pluralism of autonomous polities as a condition to which democracy responds and under which it becomes necessary in order to legitimize political rule towards free and equal citizens.

My claim is therefore somewhat stronger than a “normative realist” one that answers to the question of how democracies ought to respond to boundary problems in the current international state system. Carens characterizes my approach as interpretive and critical. But I think of it also as normative and aspirational in the sense that I aim to provide feasible inclusion principles for democracy as it should be, not just as it currently is.

I agree with Carens's view that the state system is unjust, as long as this comment is qualified by telling us what exactly is unjust or democratically illegitimate about this system. In my essay I suggest that the current principles of international law allow states to broadly ignore the interests of outsiders who are negatively affected by their decisions. Under my interpretation of the all affected interests (AAI) principle, this provides not only cover for possible injustices, which states may or may not commit towards outsiders, but also tarnishes the legitimacy of decisions taken by governments that have been democratically authorized by their citizens. Just as a legislature needs to be exposed to the articulation and mobilization of interests inside its territorial jurisdiction before it can legitimately adopt a law that affects such interests, so it must expose itself also to external interests when adopting a policy that affects these. I suggest some institutional remedies for this problem that could help to mitigate it within the current state system, but my critique is more fundamental than this in challenging the dominant interpretation of state sovereignty that underpins the current state system. I should have stated this more clearly than I did.

Carens puts much weight on how the state system contributes to global social injustice. I have little disagreement with him on this point. What I would like to point out is again that we need to distinguish the “background features of current arrangements that are morally problematic” (p. 107) from those that aren't. To my mind, global social injustice is partly generated de iure by the Grundnorm of equal sovereignty of states in international law that permits them to resist just claims for global redistribution and to arbitrarily control immigration into their territories, and partly by their de facto unequal power that allows some states to dominate others. It is not, however, a consequence of the fact that states have boundaries demarcating their territorial jurisdiction and special responsibilities for their citizens. The horizontal pluralism of polities is, in my view, a normatively desirable feature of the state system in the sense that it would be wrong to try to overcome it. This may not be obvious from a perspective of global justice, but it seems to me compelling from a perspective of democratic legitimacy of political rule.

I think it is possible to reconcile these two perspectives as long as our standard of distributive justice is a liberal one. Territorial and membership boundaries are certainly responsible for some cross-border social inequality. Wherever polities can decide autonomously about public investments, levels of taxation and spending on welfare programmes, there will be inequality of collective outcomes even if there had been perfect equality of initial resources. This kind of social inequality is democratically legitimate and should not be considered unjust. We should distinguish it from social inequality that is the result of redistribution-blockers for initial resources and of structures of asymmetric domination between states in the current international order.

Since Carens pushes me hard on this point and provides his own vision of a just world with distinct polities, I will take the risk and describe an ideal global boundary structure. In such a structure a plurality of independent states pool their sovereignty partly in larger regional unions and devolve it partly to autonomous regions and municipalities in their territory; submit to institutions of global governance with regard to issues that by their very nature affect all humankind; and keep their borders open for each other's citizens. I do not claim that this ideal world can be derived from the principles of inclusion I defend. It is just one among a larger set of possible ideal worlds that is very different from the state system without either abolishing political boundaries or internalizing them in a single global polity. The set of these possible ideal worlds is defined by the circumstances and conditions of democracy that I analyse in section 2 of my essay. The three inclusion principles that I propose in the third section would fully apply in any of these ideal worlds, which shows that they not only serve as interpretive and critical standards towards the current state system but also as general principles of democracy under ideal conditions. However, none of the three principles can identify a uniquely ideal world within this set. This is not their task. Doing so would require a much more comprehensive theory than I have tried to provide here. The three principles are meant to establish democratic legitimacy through inclusion in a world structured by political boundaries. This is not the same task as achieving global social justice although I would hope that these goals are fully compatible with each other.

Carens asks whether the limits that I impose on my inquiry are those dictated by a concern for feasibility and political relevance. Not quite. My thinking about specific boundary questions, such as national voting rights for emigrant citizens or local voting rights for non-citizen immigrants, has certainly evolved in response to empirical observations about how democratic states themselves have expanded their conceptions of the demos. And I sympathize with the call of utopophobes that political theory, in contrast with moral theory, should always be action-guiding and aim for improving existing political institutions (Frazer 2016). However, in contrast with normative realist inquiries, I recognize the usefulness of ideal theory in setting goals as long as it strengthens our efforts to build routes and bridges that allow us to approximate these goals. Feasibility and practical relevance are concerns that help in designing the bridges, but they are not the borders of our normative maps. The outer limits are those that I have tried to identify as circumstances of democracy: a double plurality that emerges from internal diversity within and external multiplicity of bounded self-governing polities.

The second set of conditions that I have set aside from my main inquiry are what I call the “contexts of democracy”: territorial jurisdiction and relative sedentariness. These I do not regard as limits for normative theory. If I could, I would like to develop a normative theory of democracy for a world in which jurisdictions are predominantly non-territorial and only a minority of people are sedentary within any set of political borders. I didn't do so in chapter 1 because I suspect that I would end up describing a dystopia rather than an ideal and the exercise might then discourage the building of bridges starting from current conditions.

Including (all?) affected interests

Does my theory rule out global democracy as an empirical possibility or a desirable goal? This depends partly on what we have in mind when speaking of democracy. Let us consider three ideas: democracy as popular self-government, as government directly accountable to citizens, and as a method for making collectively binding decisions. In chapter 1 section 3.4 I try to combine these three ideas with the corresponding inclusion principles into a comprehensive conception of democratic inclusion for democratic polities. Projecting such a comprehensive conception to the global level is in my view deeply problematic. However, this does not rule out a thinner conception of global democracy that relies only on the principle of including affected interests without aiming to build a global government and to forge humanity into a single political community.

Global democracy in this sense should be promoted by including externally affected interests in decisions made by particular governments and by democratizing global governance regimes through the inclusion of non-state actors and policy stakeholders (Macdonald 2008). Globalizing national decision-making and democratizing global decision-making in this way on issues such as climate change, refugee protection, global poverty relief, international criminal justice, trade and finance is not at all the same thing as building global democratic institutions with the power to set their own agenda and take binding decisions on an unspecified range of issues. The latter would only be democratic if there were a global political community that could authorize such a government and hold it accountable.

Carens seems to suggest that, given the seriousness and interrelatedness of the global collective action problems listed above, we should not rule out such a thicker ideal of global democracy. Yet if there were really a need for global government with such wide competences, then democracy itself might no longer be necessary to provide legitimacy for it. Those who value democracy only instrumentally might easily despair at its capacity to stop climate change or to promote egalitarian global justice. If such goals are seen to be of overriding importance then even democracy at world level might be regarded as an obstacle that needs to be removed. Consider a science fiction scenario under which this conclusion would be hard to reject. Suppose Planet Earth faced a hostile invasion by an alien species that threatened humankind with possible extinction but there were sufficient time to prepare for defence. Under these conditions of a global state of emergency, all states and their citizens should be ready to empower temporarily a global government that is free to impose its decisions on states and on private organizations, as well as directly on all individual human beings. The point of this scenario is that global government would be necessary under conditions that would also justify a suspension of democracy. Conversely, in the absence of such conditions the case for thick global democracy remains unconvincing. In other words, I cannot imagine a situation in which a global government with general legislative powers would become necessary and would have to meet also democratic standards of authorization and accountability. This dilemma between the conditions for democracy and the conditions under which a need for global government arises does not depend on whether one accepts the familiar Kantian probabilistic argument that a world state is likely to degenerate into anarchy or despotism (Kant 1795/1991).

As Carens rightly observes, I have not defeated the notion of a global polity as conceptually incoherent. But I have suggested an empirical argument that over the course of human history external boundaries have been a constitutive feature of democracy and a normative one against a thick conception of democracy at the global level. As long as coordination between interdependent polities that are not subjected to a global government can effectively address global collective action problems and provide democratic legitimacy for global policies, we should better stick to a pluralistic and thin conception of global democracy and resist the urge to merge humanity into a single demos.

Let me move on to David Miller, who is much more sceptical than Carens about the principle of including all affected interests and wants to “drop [it] altogether as a solution to the boundary problem” (p. 130). While Carens seems to think that the principle may provide some support for a global demos, Miller aims to limit its scope in two ways. First, he regards AAI as grounding at best a substantive right to ex post justification of decisions rather than an ex ante procedural right to representation in the deliberation about the decision. Second, he limits the material scope of externally affected interests that deserve a justification to those covered by a Global Harm Principle (GHP) that “considers as relevant cases where interests are affected negatively, cases where the setback to interests is serious and cases where the setback cannot reasonably be avoided by prudent action on the affected agent's part” (p. 131).

As I acknowledged in chapter 1, my discussion of the all affected interests and all subject to coercion (ASC) principles is certainly incomplete. My intention was to give these principles their due while demonstrating that they fail to answer to the democratic membership question. My main concern was to pave the ground for my multilevel theory of stakeholder citizenship in section 4. Ideally, in a book-length discussion, I should have devoted as much space to further working out the normative implications and real-world applications of the principles of including AAI and ASC. Miller's sharp comments provide a welcome occasion for adding some more substance and also some revisions to my discussion of AAI.

Let me first concede that ex post justification of decisions towards externally affected interests is always necessary and often sufficient. Although Miller is likely to disagree, I think this can be said even of his Mediterranean beach-bar owner affected by the British Chancellor's decision to raise consumption tax to finance British health care with the unintended effect of reducing the number of British tourists and the money they spend abroad (p. 130). There is indeed no justice-based claim of people working in the tourist industry to a constant demand for its services. In cases like these, the justification is entirely generic and does not even have to be delivered to the affected person; it is implied in the UK's right to take decisions that have external effects on markets. If the owner of the beach-bar calls the British Consulate to find out why his business is suffering, he ought to be given an explanation, but there is no duty of the British government to notify him.

Justification needs to be more specific and addressed to those affected where external agents are not merely impacted but coerced. For example, if a government decides on a trade boycott of another country, it is not sufficient to say that, as a participant in global trade, the other country should accept this as a market risk. Yet, it would still be absurd to claim that the trade sanctions against Iran gave that country's government not merely a claim to specific justification (which was actually provided) but to participation in the decision about the sanctions.

In other instances, ex post justification is not enough and ex ante procedural representation is called for. Miller's GHP is a good candidate for identifying such cases, except that it does not take culpability into account. Such a narrower GHP would help us to identify those cases where externally affected interests are affected negatively and seriously and where the affected agent cannot avoid the effects and suffers them unjustly. It seems to me plausible that in such cases the agent's interests ought to be represented in the deliberation before the decision is taken.

David Owen goes a step further. He suggests in his comments in chapter 6, that such inclusion of externally affected interests has a compositional effect on the demos if we distinguish discursive and editorial demoi or modes of membership from authorial ones. As I will discuss further below, I hesitate to use the term “demos” in this broad way because this risks disconnecting discursive and editorial demoi from the citizenry in a way that clouds their sources of legitimacy. However, in terms of substance Owen's distinction seems sound to me.

Let us first consider how democracy operates internally when a decision is on the agenda that is likely to affect the interests of those within the jurisdiction very unequally. In these cases, liberal and pluralistic democracies do not claim that whatever the parliamentary majority decides is legitimate. Instead, the issue ought to be made public prior to the decision so that citizens can form an opinion on it and we expect the government and MPs to listen to experts and consult with organizations that represent the different affected interests. Not every citizen is, or needs to be, involved in this discursive representation of interests. I have doubts whether the conglomerate of those who are should be called a “demos” that in this case would be narrower than the authorial demos that elects the government.

The difference jurisdiction makes is important. What happens more or less spontaneously within a well-functioning democracy, where civil society associations, interest organizations and mass media constantly monitor public policy-making, needs to be brought about in the transnational realm through institutional innovations that make the voices of externally affected interests heard. But this practical difficulty, which results from the way the international state system works, does not support a normative judgement that ex post justification is enough even where externally affected interests meet the criteria of the modified GHP. These interests must be represented before a democratically legitimate decision can be taken. This can be argued on epistemic grounds: those whose interests are affected know better how they are harmed and may also know better how the negative impact could be avoided if the decision were modified. But it can also be argued on grounds of democratic justice: some outsiders have a moral right to participate in democratic deliberations about decisions the consequences of which should not be imposed on them.

Miller seems to agree when he writes that “there are occasions when people outside of the demos do have a right to be heard by those inside” (p. 131). But this has institutional implications for national democracies, which then need to engage in intergovernmental deliberations, invite external delegates to legislative assemblies, build transnational and global governance institutions and defer to their decisions, or submit decisions that affect several polities to comparable degrees to transborder referendums.

These are the kinds of institutional reforms of the international state system that taking a principle of including affected interests seriously would require. Miller rightly rejects the “fantasy that democracy could work on an ad hoc basis, with different constituencies being assembled to decide each issue as it arose” (p. 129). This is indeed the fantasy of authors who regard AAI as a principle for constituting the demos and determining its territorial jurisdiction. Yet once a demos is legitimately constituted, some externally affected interests must and can be included through issue-specific and ad hoc institutional arrangements that extend inputs into democratic decision-making beyond territorial jurisdictions.

Like Miller, Iseult Honohan wants to minimize the role of affected interests, but in her case this is because she defends the republican principle of non-domination that is supposed to do all the work with regard to democratic inclusion. I have strong sympathy for this approach, but there is something important that gets lost. If domination is interpreted as long-term subjection to the arbitrary will of another agent, then it is no longer clear why taking into account the interests of occasionally affected external interests should be necessary at all. One reply to this might be to accept AAI as a moral principle while denying that it has anything to do with democratic legitimacy.1 This is not, however, a good response. As pointed out by pluralist theories of democracy (e.g. Dahl 1989), democracy is not about the aggregation of equally weighted individual interests of citizens, but relies on input by specifically affected interests through the channels of organized representation of interests, such as trade unions and employer federations, non-governmental organizations, campaigns in online and offline public spheres, and so on. These are not deficiencies of democracy that prevent legislators from considering the common good, but essential devices for making legislation responsive towards internal interests affected by policies. As I have already pointed out, the problem is that for externally affected interests such channels are less institutionalized, are monopolized by powerful corporate business and global media, are much more effective when used by powerful states, or are blocked altogether because of states’ external sovereignty. Opening them up to global policy stakeholders (Macdonald 2008) and pluralizing them is not only a moral requirement, but also one of democratic legitimacy. For each specific policy decision, including affected interests is a moral requirement of justice addressed to the individual legislator or the collective assembly that takes the decision. But including affected interests is also a requirement of democratic legitimacy that calls for institutional responses. If we accept that including affected interests in the deliberation about policies is a condition for democratic legitimacy, then there is no reason why this should apply only to internally affected interests and not to externally affected ones as well.

In chapter 1I attempted to deflate Goodin's global version of AAI by suggesting that agenda-setting powers remain the privilege of representatives of an authorial demos and that only actually affected interests count. In response to Miller I have now added some further specifications. But Miller's GHP still relies on the intuitive moral appeal of AAI and he fails to explain why the way interests affected within a jurisdiction must be taken into account should be fundamentally different from responding to externally affected ones.

Let me add one further concession. I have adopted the standard formula of the principle that refers to including all affected interests and often this is meant to imply that all affected interests ought to be taken into account equally. As my discussion here has made clear, this is not sustainable for externally affected interests. I can save the “inclusion of all” requirement by extending the principle to the case of generic ex post justifications discussed above. But there is no way a requirement of equal inclusion can be defended. As Miller rightly points out, the degree of affectedness matters. Even a threshold interpretation does not imply that those interests that qualify have a claim to be represented equally. This is not only because stronger degrees of affectedness and harm give rise to stronger representation claims, but also because the appropriate response to such harms depends on the issue and the institutional context. Representing the universal interests of humanity in slowing down climate change will require a different institutional response compared with representing the interests of the people of Tuvalu in finding a new home if rising sea levels submerge their island state.

Even inside a jurisdiction, affected interests are included unequally, partly based on the degree of affectedness, partly on the numbers of those affected. What is problematic is when strongly affected interests are excluded because they lack legal status and voting power or resources for mobilization and when weakly affected interests of small numbers dominate policy deliberations because they have more resources to make themselves heard or to influence decision-makers. The same considerations apply to negatively affected external interests. But in their case the norm of state sovereignty adds another exclusion mechanism that systematically prevents them from being adequately included. And this is precisely where AAI provides a compelling moral critique and points towards feasible political reform.

Subjection or interdependence?

David Miller's comments prompted me to clarify that affected interests cannot be included equally and that including all affected interests comes at the price of thinning out the corresponding duties and responsibilities of democratic legislators. I claimed in chapter 1 that the principle of including all subject to coercion is more egalitarian than AAI. Yet Carens points out that those subjected to the laws, too, do not necessarily enjoy, or have a claim to, equal legal status and rights. In his approach, it is social membership that counts and this comes in degrees. Tourists, temporary migrants, foreign students and long-term resident foreigners are all subjected to the laws but do not have the same residence status or access to different types of social welfare rights. Doesn't this undermine the egalitarian thrust of ASC?

In response I should make it clear that I do not understand equal protection of the law as entailing equal enjoyment of all legal rights, but rather equal protection of those rights which persons subjected to government authorities can claim based on the kind of subjection to which they are exposed. In other words, equality here refers to protection of rights rather than to substantive rights per se. What is democratically illegitimate – and incompatible with the rule of law – is to subject persons to the same laws in such a way that some are offered full protection and others only some or no protection at all.

This is not meant to be a formalistic response. It would be a mistake to narrow the scope of ASC so that it covers only long-term residents. Tourists are also entitled to protection of their fundamental rights and should be able to contest administrative and judicial decisions that apply to them and restrict their freedom. And, as Carens has argued persuasively, irregular immigrants who are long-term residents have claims to regularization of their status that provides them then with equal legal rights. I suggest in chapter 1 that ASC should even be applied to would-be immigrants turned away at the border who obviously cannot claim most legal rights and certainly not membership in the polity, but who have a claim to protection of their human rights and to justification of their rejection that is a special responsibility of the state agents who exercise coercion over them.

A critical question is which of those statuses that come with restricted legal rights rather than with the full scope of equal protection exist only in a non-ideal world and which would exist in any pluralistic world that is structured by external relations between self-governing polities. Tourists will exist in any such world and their claims to legal rights will differ from those of long-term residents. Irregular migrants will exist only in a non-ideal world in which states are justified in controlling immigration for work and settlement.

My main disagreement in this regard seems to be with David Miller, who draws a distinction between coercion used against would-be immigrants and their vulnerability. On Miller's account, “[w]hat lends plausibility to ASC is the thought that a coercively enforced legal system shapes people's lives in a fundamental way, and potentially exposes them to domination” (p. 131). “When people are prevented from entering a country by immigration controls, a significant opportunity is often being denied them, but it does not follow that their lives are being shaped and potentially dominated by the legal system of the country they are trying to enter in the same way as those who are already living under that system” (p. 134). This limits the scope of ASC in a way that would leave lots of people against whom states use coercion without any claim to protection of their rights. In my view, liberal states have a general duty to justify immigration control. Contrary to most other theorists (Carens 2013; Oberman 2016), I do not ground this duty in a corresponding human right, but in liberal states’ duties to promote their own citizens’ opportunities for free movement, which creates a reciprocity-based duty to admit other states’ citizens to their territory. This does not entail that tourists who are denied entry visas are thereby dominated. The general duty to justify restrictions of free movement is enhanced if those who want to enter can raise specific claims why this state rather than any other should admit them. The weakest claims of this kind might be those referring to economic benefit. If admitting economic migrants would benefit the country of origin, the migrants themselves and the country of destination – in the language of development economists: if there is a “triple win” – then liberal states ought to create immigration programmes for such economic migrants. The claims of would-be immigrants are much stronger if their family members have previously been admitted to the country, and they become imperative if they are refugees who might perish if they were turned away. Coercion and vulnerability are thus not entirely different things, as they seem to be on Miller's account. Instead, the general duty to justify coercive migration control is enhanced by special responsibilities of states for particular migrants and by those migrants’ vulnerability. Where the responsibility and vulnerability is strong, migrants’ lives are indeed being shaped by a decision to turn them away and they are actually rather than just potentially dominated by the legal system of the country they are trying to enter.

While my disagreement with Miller concerns what one could call the “lower range” of ASC, I have another one with Carens, Honohan and Owen regarding the “upper range”. Can ASC – or Carens's principle of social membership – determine a threshold above which all those subjected to coercion have an equal claim to all rights of citizenship, including access to the legal status and to voting in all elections? Iseult Honohan suggests in her comments, and Carens has argued so in his own work (Carens 2013), that all those permanently subjected (as long-term residents) have this claim. This would imply that, at least inside a territorial jurisdiction, ASC does all the necessary work and my stakeholder principle becomes redundant. I will discuss the implications of this view for voting rights in contexts of migration and multilevel democracy below. Here I want to focus on Iseult Honohan's interpretation of ASC that aims to blunt the differentiation from the all citizenship stakeholders (ACS) principle that I have tried to sharpen.

According to Honohan, “the all subjected account … may define membership of the demos more clearly on the basis of a single principle, but the account of citizenship needs to be pluralist, mainly by building in a temporal cushion with respect to subjection” (p. 157). But she also writes that “temporary rather than continuing subjection does not constitute the same threat of domination as longer-term residence, and requires protection rather than membership of the demos” (p. 153). This is exactly my point and the reason why I want to retain a broader version of ASC that focuses on protection separate from a distinct principle that applies to membership claims.

The risk I see is that an exclusive focus on long-term subjection diverts attention from those forms of domination that would-be immigrants, transients and temporary migrants are subjected to precisely because they are (rightly) not seen as having a claim to membership. The most obvious case is refugees who seek admission and whose vulnerability to domination results from the lack of protection offered by their citizenship of origin.

Honohan acknowledges that the value of non-domination that is foregrounded in contemporary neo-republican theories can be interpreted in different ways. Some, like Philip Pettit, put more emphasis on people's interests in protection of their rights by the government to which they are subjected, while others, such as Habermas, Lovett and Honohan herself, emphasize relations of interdependence and the connection between individual and collective autonomy. “There seem, then, to be two alternatives in considering non-domination as the basis of ASC: we can interpret it in terms of interests, in which case it collapses back into AAI, or in terms of facilitating autonomy, when it leans towards a stronger norm of participation in collective self-government” (p. 244).

This seems to me a problematic contrast. First, all three principles (AAI, ASC, ACS) can be restated in terms of political interests, but they refer to interests in different elements of political life: policies, government protection, and membership and self-government. Regarding the three as complementary rather than alternative or reducible to a single principle should help to save republicanism from the charge that it has illiberal implications. In other words, a republicanism focused on self-government must be tamed and enhanced by a liberalism defending the inclusion of affected interests and equal protection of the law also for those who are not members. This should not be difficult for neo-republicans since they can recognize all three principles of inclusion as responding to risks of arbitrary domination. Outsiders who are seriously and innocently harmed by a policy in whose making their interests have not been taken into account are exposed to the arbitrary will of another agent, even if this is not a permanent condition. Persons inside a jurisdiction whose fundamental rights are not protected by the government are dominated even if they do not have a claim to membership. While all three inclusion principles can be endorsed as serving the same goal of avoiding domination and promoting self-government, it does not follow that asserting this goal as a coherent normative ideal of democracy reduces the three principles to a single one. Republicans who think of ASC as a principle that is sufficiently broad to cover all domains of democratic inclusion are as wrong as liberal theorists who regard AAI as the overarching inclusion principle from which all others are mere derivatives (Beckman 2009).

Honohan provides, however, very helpful interpretations of a republican version of ASC. She responds to the question about membership claims by referring to a condition of “future-oriented interdependence in continuous subjection” (p. 155). Such interdependence creates “circumstances where collective action is required for the possibility of self-government” (p. 152).2 Notice how, in this account of subjection, it no longer serves as the justifying reason why governments that restrict the autonomy of individuals have also to offer these individuals protection of their autonomy. Instead, continuous subjection to coercive government power becomes a condition under which interdependence among people is likely to grow to the point where they consider themselves as members of a political community with a claim to collective self-government. For membership claims, it is therefore interdependence rather than subjection that does the justificatory work. And this is how it should be, since interdependence is another, and possibly better, name for the genuine links that I seek to trace through the citizenship stakeholder principle.

There are, however, still some differences of view that emerge from Honohan's emphasis on residence as the proxy for forward-looking interdependence. In my view, the subjection of residents within a territorial jurisdiction to the same government may create the horizontal interdependence that gives rise to a sense of shared membership in a political community, but is not always necessary (as in the case of diasporas participating from distance in projects of national self-determination) or sufficient (as in the case of colonies and occupied and annexed territories).

It is also good to see that sometimes disagreements about the interpretation of principles matter for specific policy recommendations. This proves that political theory is after all sometimes capable of helping citizens and policy-makers think through the normative choices that they face. So here are my disagreements with Honohan's concluding suggestions for specific citizenship policies.

Honohan endorses de Schutter's and Ypi's (2015) proposal for automatic citizenship attribution to long-term resident immigrants, which is based on the perception of an “unfair asymmetry in the distribution of political obligations between citizens and immigrants” (p. 153). There are indeed many native citizens who are concerned that immigrants free ride on the benefits provided to all residents in their host country. Instead of campaigning for imposing citizenship on immigrants, however, these citizens mostly want to make it harder for immigrants to get permanent residence and to become citizens.3 This observation does not of course address the normative question of whether immigrants should be compelled to become citizens. My response to this proposal is that it is plausible only from an immigration-centric perspective that fails to consider a just distribution of political obligations and rights for individuals with strong political ties across state borders. The non-citizen residents of one state are at the same time the non-resident citizens of another state. We need to consider their claims to membership (including their claims not to be pressed into another membership against their will) in a transnational context rather than a purely national one.

Transnational migrants do not stand in the same relation to their host state as those residents (including second generations of migrant origin) who have obtained their citizenship at birth. Those who argue that automatic citizenship at birth for natives and optional naturalization for migrants is unfair towards the former are inspired by a version of ASC that focuses on long-term residence rather than a stakeholder perspective, which considers how migrants and sedentary citizens are involved in different membership relations. This may be unfair towards Honohan since she considers the transnational context when arguing that the increasing acceptance of multiple citizenship by states “reduces the difficulties with respect to mutual obligations between independent states” (p. 154). Toleration of multiple citizenship would indeed make the consequences of involuntary naturalization less dramatic for immigrants. But it seems to me implausible to assume that countries of origin will then readily accept compulsory recruitment of their emigrants as host country citizens or ought to do so as a matter of justice. It is also not clear that multiple citizenship reduces the difficulties with sorting out citizenship obligations (just think about military service and diplomatic protection). Finally, it seems incoherent to argue that immigrants have to be naturalized against their will in order to equalize the obligations between long-term residents who are equally subjected to government power, while accepting that as multiple citizens immigrants retain additional rights and obligations that native mono-citizens do not share with them. In my view, these normative puzzles can only be resolved by considering from the very start the membership claims of migrants from a transnational rather than from an immigration-and-residence perspective.

Honohan's second policy proposal concerns ius soli and ius sanguinis, both of which she proposes to make conditional upon residence after birth. These suggestions follow indeed coherently from a strong emphasis on residence as the best proxy for continued subjection and therefore, in her view, also for membership claims. For me, the unattractive implications of such reforms cast instead some doubts on the principle from which they are derived.

Both proposals abolish a main feature of birthright citizenship, which is that it is not conditional or provisional once it has been awarded. Instead, in most current instances, it is the parents and not the child who have to meet conditions. In the case of ius sanguinis, this is often a condition that parents must have been born in the country of citizenship in order to transmit it to a second generation born abroad. In the case of ius soli, it is mostly a condition that one of the foreign parents must have resided for some time in the country for the child to get citizenship at birth. These rules serve as proxies for the parents’ genuine links to the country, not for the child's future residence. “Provisional ius soli membership for those born in the state, depending on continuing residence”, as suggested by Honohan (p. 155), would instead undermine the idea that citizenship is a lifelong status and give to the state a power that liberal governments should only have in extreme and rare cases: the power to deprive citizens of their membership (Bauböck and Paskalev 2015). The same objection applies to “conditional ius sanguinis membership for those born to citizens abroad, retained only if residence in the state is subsequently established” (pp. 155–156). I assume Honohan would like to qualify this proposal with a condition that withdrawal is only possible if the person has acquired another citizenship. In the international state system, avoiding statelessness and securing a stable assignment of state responsibilities for individuals is an overriding imperative. States should therefore not be able to shed their responsibility for citizens based on a weakening of links for which residence serves as the proxy. Yet in combination with her advocacy of provisional ius soli, Honohan's proposal amounts to turning minor children born to either non-resident or non-citizen parents into citizens on probation who need to establish their links to one state or the other through continuous residence. Those who move too often or between too many countries will fall into the traps of residence-based citizenship. Imagine a child born in country A to parents who are native-born citizens of country B. If the family settles in country C before Honohan's continuous residence requirement has been met, the child could end up being stateless until she qualifies for naturalization in C.

From a stakeholder perspective, the links that citizenship status is meant to protect are primarily those that protect individuals’ “right to have rights” in a world of independent states. This does not entail that rights of citizenship cannot be conditional upon residence. In fact, most rights are anyhow since states can only guarantee them within their territorial jurisdiction. As I will explain below, the core citizenship right of political participation in democratic elections should be attached to present or past residence. In other words, in contrast with membership in the citizenry, membership in the demos need not be awarded as an unconditional birthright. Instead of giving states the dangerous power to strip children born abroad of their citizenship status, I propose that such birthright citizens should not acquire membership in the demos at the age of majority unless they have connected to the state where they want to vote through a sufficient period of past or current residence. The same logic applies to the phasing out of birthright citizenship itself. If we are concerned about equality between citizens independently of how they have obtained their status and about the power of states to strip citizens of their membership, then it is much better if the third generation of emigrant origin no longer acquires citizenship by descent than if the second generation is put on provisional citizenship.

Stakeholder citizenship and liberal nationhood

David Miller engages most extensively with my search for a distinct normative principle that provides legitimacy to individual membership claims and to institutional membership rules in liberal democracy. Miller sees my account as a pluralistic one and urges me to consider combining several principles, which would open my approach also for a liberal version of the nationality principle that Miller defends. However, I am monistic in this respect. I propose that AAI, ASC and ACS each address a specific aspect of democratic inclusion, but that only ACS applies to membership issues. The price that I need to pay for this claim is that my statement of ACS remains necessarily quite general so that it can apply to the broad range of problems in which membership in a particular polity is at stake.

As Miller rightly points out, membership problems may either concern individual inclusion or territorial jurisdiction and collective self-government claims. Chapter 1 deals mostly with the former and only very briefly with the latter. Individual claims to citizenship generally presuppose uncontested territorial borders and collective identities as a stable background. When immigrants apply for acquisition of citizenship what needs to be considered is the democratic legitimacy of naturalization requirements and not that of jurisdictional boundaries. By contrast, when a national minority demands stronger political autonomy or a referendum on secession, what needs to be considered is the legitimacy of existing federal or other arrangements for devolution and power-sharing as well as how changes would affect individual memberships. My claim is that ACS provides a general guideline for liberal responses to both types of problems because it focuses on the relational nature of membership. This relation is not just one between particular individuals and polities, but is a triangular one between individuals, a particular polity and other polities. When specifying rules for individual membership in a particular polity we need to keep the conditions in mind under which that polity can be self-governing in its relations to other polities. And when considering how jurisdictional conflicts between polities should be sorted out we need to consider whether present or proposed future arrangements provide adequate conditions for self-government of all polities involved, as well as whether these arrangements satisfy individual claims to membership.

Let me first address Miller's critique of ACS as a principle for determining individual citizenship claims under conditions of stable jurisdictional boundaries. Miller considers two interpretations of ACS: a utilitarian one according to which individuals should be recognized as members of those polities where their interests in autonomy and well-being are best protected; and a psychological one according to which individuals should be recognized as members of those polities that figure most prominently in their personal identities (p. 136).4 I agree with Miller that both interpretations are problematic.

On a utilitarian interpretation, the strength of individuals’ claims to a particular membership depends on how well that polity would protect their autonomy and promote their well-being. I suspect that if this view prevailed, Swedish citizenship might become hugely oversubscribed, which would lead to a collapse of Sweden's capacity for self-government. What I have suggested instead is that citizenship ought to trace individuals’ genuine links to a particular polity. If my strongest links are to a polity that is currently not flourishing, then this will affect my well-being. Yet since my claim to membership in that polity is a relational one, it does not follow that I have now a stronger claim towards another polity that is currently doing better. What follows instead is that my polity needs to do better in order to make my own life go better. But it does not follow either that I am forever stuck with this particular membership. If the polity is a city inside a liberal democratic state, I can always move to another city where being a resident is sufficient to establish my claim to membership. And if the polity is an independent state, then it still has to guarantee free exit and ought to promote reciprocal free movement with other states. After some time of living in another country, my autonomy and well-being will no longer depend only on how my country of origin is doing, but will also be shaped by my country of residence. Rather than having a right to choose another citizenship that makes me better off, it is the changing circumstances of my life that give me a claim to a new citizenship. Whether I moved in order to improve my economic opportunities or did so for other reasons does not matter. What matters is whether the links that I establish with the polity match the conditions under which that polity can be self-governing. And these conditions differ for municipalities and independent states.

The psychological interpretation of stakeholdership is wrong for the same reason. Miller's Francophile (p. 137) does not have a claim to French citizenship because his subjective preferences are irrelevant for determining the membership boundaries of a self-governing territorial polity. This does not mean that it is not important that citizens share a sense of belonging. It is indeed important for the flourishing of a polity that citizens feel attached and attribute non-instrumental value to their membership. However, these psychological aspects are a consequence of experiences of collective self-government rather than a criterion for selecting members.

The secluded monk and the wealthy cosmopolitan who collects passports may not feel any particular psychological attachment of this kind. They are free riders on the benefits of collective self-government to which they do not contribute. This does not disqualify them from being members in those polities where they fulfil the criteria of residence, birthright or derivative citizenship, since these criteria refer to links that can be established independently of their subjective attitudes. The citizenship that they receive is objectively important for their autonomy and well-being – although they do not value it or value it only instrumentally – if it confers membership in a stable democratic polity. And in order to make the polity stable and democratic, the criteria for awarding citizenship must trace individuals’ relevant objective links.

The general idea of triangular relational correspondence can be best illustrated by considering the different membership criteria for states, municipalities and sub- or supranational regional polities that I discuss in section 4 of chapter 1. It is perhaps easiest to understand for derivative citizenship. I am a citizen of the European Union because I am an Austrian citizen by birth and Austria is a member state of the EU. Member states are the constitutive polities of the EU as a union of democratic states. Such a union tries to combine the self-government of member states as independent countries in the international arena with supranational democracy at the union level. In order to do so, it links the citizenship between the two levels so that every citizen of a member state is also a citizen of the union. The fact that I have been living in Italy for several years activates some of my special rights as an EU citizen, but it does not affect my membership status. Instead, I have a stake in EU citizenship because I am a citizen of a state that has accepted the conditions under which the EU produces binding supranational legislation with supremacy over national law and direct effect on all EU citizens.

A similar story can be told about regional citizenship in federal states whose federal union and territorial integrity are conditional upon regional self-government. In order to govern themselves as autonomous provinces within a stable and self-governing federation, citizenship of the former must be linked with that of the latter. If the relevant relation between polities is, however, one between municipalities and the states within which they are embedded, then such a derivative link seems to me neither necessary nor justifiable and residence becomes instead the basis for claims to citizenship at the local level. Finally, if the relation is one between independent states in the international system, then it is imperative that citizenship be based on birthright and constructed as a lifelong status.

In a bilateral conception of the relation between individual and polities it might seem absurd to claim that individuals’ stakes in citizenship can be fully and equally met by rules that attribute to them membership because they are already members in another polity, because they have been born to citizens or in a political territory, or because they are current residents in such a territory. These very different relations between individuals and polities seem to exemplify alternative conceptions of political community that cannot be reconciled with each other. It is the triangular relation between individuals and conditions for internal and external self-government that explains why these rules are appropriate interpretations of stakeholder citizenship that complement each other in a multilevel citizenship architecture.

The same triangular conception should also help to address the questions about national self-determination and secession that Miller raises. In section 3.3 of chapter 1 (pp. 42–43) I propose two criteria for establishing the legitimacy of self-government claims that contest the borders of current jurisdictions or the division of powers between regional and state governments: a criterion of compossibility of entangled self-government projects that considers how they impact on each other, and a criterion of representativeness of claims that considers the inclusiveness and internal support of the project.

The criterion of representativeness of claims relies on the bilateral version of the stakeholder principle. It requires that claims to enhanced self-government or secession must not only be internally supported by a relevant majority in a territory at a specific point in time (which would be sufficient for an associative secession theory) or be made on behalf of a culturally distinct historical community (which it would have to be under a nationalist conception of self-determination). I suggest instead that “the democratic people itself is constituted through representation of its claims to self-government” (p. 43).5 This means that there is a lot of historical contingency in the construction of the people on behalf of whom self-government claims are raised, but these claims are normatively constrained by a requirement that they represent a political community that includes all citizenship stakeholders rather than a regionally dominant nationality.

The second criterion of compossibility should not be understood merely as a question about whether the polities involved could continue to exist as distinct from each other and with some degree of self-government, but also about whether their relations to each other can be regarded as free of domination. Algeria-in-France or Ireland-in-Britain could have become territories with some degree of home rule and a shared French or British citizenship for all inhabitants, but given the colonial nature of the relation, they would very likely have remained dominated in a way that undermined their self-government. Separation is not always necessary, however, in order to overcome a history of domination. In plurinational democracies, such as Belgium, Britain, Canada, India or Spain, the historical domination of minorities has been overcome to a significant extent, even if not fully, through federally nested self-government. In such arrangements, the territorial integrity and unity of the encompassing state cannot be unconditional but must instead depend on constitutional guarantees for territorial autonomy of regions where minorities are concentrated and democratic practices of power-sharing that involve these regions in the government of the federation. Vice versa, where a plurinational democracy meets these conditions, national minorities are bound to respect its territorial integrity, since unilateral secession rights are incompatible with a right of self-government of the larger polity that encompasses the minority territory.

Miller finds it puzzling that I accept remedial secession in cases where primary rights to self-government have been persistently violated. Isn't this like saying “that I have no primary right to own property, but I do have a remedial right to take back stolen possessions” (p. 139, n. 9)? Not necessarily, if we distinguish between self-government as a right that can be exercised either as territorial autonomy within states or through independent statehood, on the one hand, and self-determination as a right to unilaterally decide between these options by changing the status or borders of a jurisdiction, on the other hand. This would be analogous to saying that my property rights do not include a right to use force against others who might be interested in my property, except if they actually attempt to steal it.

I fully agree with Miller that it is incoherent to attribute self-government rights to minorities while denying them to the larger citizenry of a state that includes these minorities. The compossibility test for self-government claims is not biased in this way but explains how territorially entangled projects can still be treated symmetrically: minority autonomy is the condition for territorial integrity of the larger polity and respecting the integrity of the larger polity is the condition for territorial autonomy of the minority.

Miller concludes by suggesting that “the nationality principle underlines the role of a shared identity in creating social and political trust, thereby facilitating the accommodations and compromises that are essential if democratic decisions are to be accepted as authoritative by all concerned” (p. 140). There is, however, a prior question as to whether the nationality principle is still able to create and sustain those jurisdictions within which democratic decisions need to be taken so that citizens can govern themselves in the contemporary world. I suspect that the once progressive role of the nationality principle in this respect has been largely exhausted since the times when it triggered decolonization and devolution in plurinational democracies. In the current highly interdependent world, democratic self-government can only survive if individuals can see themselves as citizens not only of nation-states, but also of local, regional and supranational polities that cannot be imagined as nations.

The test for this historical hypothesis is empirical rather than normative. In today's Europe the nationality principle is politically invoked mostly in order to deny immigrants access to citizenship, to reject solidarity between the member states of the EU and to break up democratic states or make their governments dysfunctional.

Miller is nevertheless right that there is a “tradeoff between thicker and more motivationally powerful forms of national identity and thinner and weaker, but more inclusive, forms” (p. 141). But this tradeoff should not be regarded as a static one. Growing interdependence between countries and growing mobility across their borders mean that national identities become not only ever more exclusionary but are also increasingly mobilized for the purpose of blocking emerging collective identities at local and supranational levels. The goal of liberal republicans should be instead to strengthen collective identities based on shared citizenship that can support social solidarity and democratic self-government across different levels and across borders of democratic polities.

Demos and citizenry

Nearly all theorists who have addressed the democratic boundary problem have considered inclusion in the demos as if it means the same thing as access to citizenship. Towards the end of my brief discussion of the citizenship stakeholder principle in section 3.3 of chapter 1 I suggest a distinction between the citizenry and the demos (pp. 44–47). The demos consists of adult citizens who can vote or be elected, while the citizenry is a broader category that includes minor children and others who are not capable of exercising the rights and duties of members of the demos. The comments by David Owen and by Sue Donaldson and Will Kymlicka challenge this distinction from two contrasting perspectives. While Owen wants to broaden the concept of the demos, Donaldson and Kymlicka question the distinction itself.

Owen identifies for each of the three inclusion principles a specific form of membership of the demos that provides democratic legitimacy for the corresponding process. The discursive demos provides legitimacy to political decisions through the input of affected interests; the editorial demos provides legitimacy to coercive subjection to government power through securing liberties and exposing this power to contestation; and the authorial demos provides authorization for democratic government and thereby enables the wider community of citizens to govern themselves collectively.

I find Owen's proposal congenial because it keeps the three inclusion principles apart and identifies for each a distinct set of persons whose inclusion fulfils specific legitimacy requirements. I still have some doubts about the terminology because the very concept of democracy refers etymologically to the rule of the demos. In my view neither the discursive nor the editorial demos is engaged in ruling in any meaningful way. Yet I do not want to put too much emphasis on terminology, not least because I do not have a better one to offer. So I will accept for the sake of this discussion that the citizenry (we might also call it the democratic people) is the political community that has a claim to self-government, whereas the demos is a functional entity that is constructed for specific purposes of democratic legitimation. The question I want to consider here is the relation between these different demoi and the citizenry.

The relation of the discursive demos to the citizenry is the most open-ended: the set of those who have a claim to be included changes all the time, depending on the decision on the agenda. It is a discursive relation in the sense of Habermas's discourse ethics (Habermas 1986). The discursive demos can be wider or smaller than the community of citizens. There is no specific requirement of congruity between the discursive demos and the citizenry, and thus also no implication whatsoever for membership in the latter.

The relation of the editorial demos to the citizenry is a much closer one: government power is legitimately exercised within territorial or personal jurisdictions. It is a relation of control by governments over subjects and contestation of governments by subjects. Being subjected to government power on a long-term basis gives rise to claims for membership in the citizenry. But, as I have argued, this is neither a strictly necessary nor a sufficient condition and governments must not themselves draw the boundaries of the citizenry through subjecting individuals to their power. They must instead enable a political community to be self-governing in its relation to other political communities. And this requires specifying membership and boundary principles for different types of polities and individuals’ relations to these. The inclusion claims of immigrants and emigrants in a state that is embedded in the international state system are therefore different from those of residents in a municipality nested within a state.

Finally, the relation of the authorial demos to the citizenry is the closest one; it is a relation of representation. This may seem odd, since I am not talking about those who are elected but those who vote. Let me explain. One aspect of this representative relation is that the members of the demos represent those citizens who are not included in the former. In nineteenth-century democracy, when casting their votes, the male heads of households were assumed to represent their wives and children, who enjoyed merely “passive citizenship”. In contemporary liberal democracies the adult members of the citizenry are also members of the demos, so the relation between demos and citizenry seems mostly one of identity rather than representation, apart from the remaining exclusion of minor children. (I will say more about this when discussing Donaldson's and Kymlicka's critique.) The second aspect is that citizens when they vote are, or ought to think of themselves as, representatives of the larger community of citizens of which they are also individually members. In other words, the members of the demos should think of themselves not merely as electing representatives but as being representatives. The reason for this is that elected representatives do not merely represent their voters’ particular interests but have the power to shape an open political agenda for the community of all citizens. Only if voters see themselves as representing the common interests of this community will they elect representatives who carry out this mandate and hold these legislators accountable if they fail to do so. This is obviously a democratic aspiration rather than a condition for electoral participation; it does not rule out self-interested voting, but it constrains it normatively.

The representative relation between citizenry and authorial demos suggests furthermore that the demos must be a subset of the citizenry. Representatives elected by the demos must themselves be citizens, and so must those who are eligible to vote. If the authorial demos were composed of, or included to a significant extent, non-citizens this would undermine its functional task of representing the citizenry through authorizing and holding accountable their government. As I will argue in the next section, this leads me to resist Owen's suggestion that the principle of including all subjected to coercion might apply to determining a legislative demos that would then be composed also of non-citizen residents.

This interpretation of the role of the authorial demos makes it possible to resolve the problem of representation of citizens who are not included in the demos. In order to serve its functional purpose of authorizing a democratic government, this demos must consist of citizens who have certain cognitive capacities. The demos should therefore not be understood as if it were a distinct form of political community; instead its members ought to be regarded as the trustees of those citizens who are not included in the demos, such as minor children and mentally disabled resident citizens.

Donaldson's and Kymlicka's critique focuses on this point. They argue that my theory buys into a “linguistic capacity contract” that has historically served to support unjust exclusions and that is still invoked today when denying political rights to minor children or mentally handicapped persons and citizenship status to domesticated animals.

Donaldson and Kymlicka write: “On Bauböck's model … our duty to support and enable the political participation of our fellow members of society depends on whether they fall above or below some stipulated threshold of cognitive or linguistic competence” (p. 175). This is not my view. We have duties to support the political participation of minor children and the mentally handicapped in proportion to their capacity and desire to participate (a point also emphasized by Honohan). A linguistic capacity threshold is only relevant for those who are members of the demos, that is, those who are called to participate in democratic elections and who qualify as candidates. Donaldson and Kymlicka also attribute to me the view that “we owe [minor children] participation not in virtue of what they are – not in virtue of their interests or membership as children – but in virtue of what they will become (adult citizens)” (p. 177, n. 23). Again, this is not what my distinction between citizenry and demos suggests. I have argued that “newborn babies are attributed citizenship not just because we regard them as future citizens” (p. 46), but because they belong to a transgenerational political community. Such a community would no longer be transgenerational if it did not include those recently born and those close to death regardless of their capacities and contributions. On my account, minor children are not, however, members of the demos.

Why? Donaldson and Kymlicka suggest that a principle of stakeholder citizenship should first identify those who have a claim to membership and then search how these could be enabled to participate as fully as possible in the governing of the polity. In this view, a lack of (full) linguistic and cognitive capacities does not justify exclusion but calls instead for enabling forms of inclusion. On this account, any hard boundary separating the demos from the citizenry looks suspicious. After all, children, cognitively impaired people and domesticated animals can and do communicate their needs. Even if these communications are addressed to their immediate care-givers, they can still be fed as input into political deliberations and decisions that apply to these individuals and regulate their lives.

Donaldson and Kymlicka have forcefully made this argument in their recent work (Donaldson and Kymlicka 2011, 2017 ) and I find this part of their theory convincing. I remain, however, unconvinced in two respects. First, I think that the distinction between demos and citizenry is essential for liberal democracy and, second, I do not think that the rights of domesticated animals are best expressed in the language of citizenship.

Let me start with what seems obvious. Linguistic and generic cognitive capacities are indispensable as a qualification for ruling a human polity. Candidates for public office who seek a mandate to represent citizens must have these capacities and this justifies age thresholds for candidacy rights. Citizens, including those of minor age, would not want to be ruled by individuals who lack this capacity.6

Maybe this is enough and we need to set thresholds only for candidacy rights and not for participation in elections? Yet if only rulers need to be linguistically capable, then citizens would be subjected to rule by an elite that they could not hold accountable. For Aristotle, democratic citizenship meant ruling and being ruled in turn, which implied that the same capacity requirements applied to citizens and those exercising public office. Yet Athenian office-holders were appointed through lotteries. For representative liberal democracy, this condition can be somewhat relaxed, but not altogether. Those citizens who vote do not have to be trained as professional politicians but must be capable of communicating with other citizens and understanding what candidates and parties stand for. They are also collectively responsible for their choices (Hobden 2015). Not to expect the members of the demos to have these capacities means accepting that they must be governed paternalistically by those who have them.

This does not entail that only those who have the relevant capacity can be citizens. First, as suggested by Donaldson and Kymlicka, capacities can be created by overcoming socially constructed and reinforced incapacity. Instead of presupposing that citizens are literate, economically independent and have enough leisure time to be politically informed and engaged, the democratic state has duties to develop these capacities and provide citizens with these resources. Second, individuals whose lack of these capacities cannot be overcome by such means can still be included in a wider self-governing political community as citizens. These citizens (minor children and the cognitively severely handicapped) will then indeed be governed paternalistically. Paternalistic government that is oriented towards the well-being and autonomy of individuals will not only protect them and provide them with resources but also promote their participation in accordance with their capacities, in the ways Donaldson and Kymlicka suggest. But this does not qualify them as full members of the demos who are responsible for their political choices and have to hold those who rule over the citizenry accountable.

Donaldson and Kymlicka may respond that my account fetishizes elections from which citizens without the necessary cognitive capacities are inevitably excluded over deliberations in which they can be included. Yet democracy is not just an ongoing deliberation about the common good; it is also the exercise of coercive power that is always in danger of being abused. Minor children (below the age of adolescence) have to trust that their parents have their well-being in mind, but rulers can never be trusted in this way. Even parents of minor children have to be checked by laws and political authorities that they do not abuse their paternalistic powers. Democratic elections and the division of power between branches of government are devices that have been designed to minimize such abuse of government power and build on the anti-paternalistic idea of democratic legitimacy through popular self-government.

Enabling minor children and cognitively impaired citizens to shape their own lives and have a say over the laws that regulate them is very important. But these inputs into the political process are, in my view, fully covered through the principles of including affected interests and those subjected to coercion. Under the latter, individuals who are not included in the demos enjoy not only equal protection under the law but also rights of contestation. Moreover, because of the transgenerational nature of political communities, minor children and cognitively disabled persons are also included in the citizenry. They are not, however, members of what Owen calls the authorial demos. Age thresholds for voting can and have been lowered, and cognitively impaired persons should retain voting rights by default. But none of this does away with the need for linguistic and cognitive capacity conditions for membership in the demos.

What we need to consider instead is vicarious representation in legislative institutions. Owen suggests a sophisticated model of indirect special representation for minor children based on second votes for enfranchised adult citizens (pp. 186–187). I find this an attractive proposal. A lot of details would have to be worked out. Should the representatives of citizens who are not members of the demos have veto power or rather the power of legislative proposal, or both? Would the powers of such special representatives be limited to certain policy areas that are specifically relevant for children? If we consider children's interests as continuous with those of future generations of citizens, then no policy area could be excluded.

Should all citizens without voting rights be represented indirectly in this way? This depends on the reasons why they are not enfranchised and their relation to the wider citizenry. Owen's proposal would hardly work for marginalized minorities, such as mentally handicapped citizens who should therefore be better generally enfranchised and additionally represented by specially appointed ombudspersons rather than by separately elected representatives.

Let me now address the second challenge raised by Donaldson and Kymlicka. If citizenship status does not depend on cognitive capacities why shouldn't domesticated animals be included as citizens? I propose three arguments why not. The first is that they are already sufficiently included under the affected interests and subjection to coercion principles. I cannot see which kind of injustice is committed towards domesticated animals if they are considered members of the private households of their care-givers and as denizens of the polity that owes them protection under the law and has to take into account the needs that they articulate. Once this has been accepted, what would be gained by calling them citizens?

My second objection refers back to the circumstances of democracy. These include not only internal diversity (which could very well cover a diversity of animal species) but also the condition of a pluralism of distinct and bounded polities. We may recognize that social animal species form distinct and bounded communities of their own and, as Donaldson and Kymlicka (2011) have suggested, we can then conceive of our relations to them, in analogy with those between states, as relations between sovereign and bounded communities of different species. Further progress in the study of animal behaviour might also lead to the discovery of norms and institutions in some animal societies that are functional equivalents of citizenship in human polities. But what Donaldson and Kymlicka propose is something quite different: cross-species citizenship through which humans and other animals share membership in political communities. Over the course of their history, humans have always formed political communities, membership in which has been determined through relations with other similarly organized human communities and through the imperative of maintaining transgenerational continuity of their particular community over time. These two features make it difficult to conceive of a cross-species citizenship that includes domesticated animals even if these have been bred to serve human needs and to live inside human societies in close relations with human beings.

A third objection builds on my interpretation of the stakeholder principle. We should see domesticated animals as stakeholders in the policies that affect and regulate their lives and also as having a stake in a government that protects them and enables them to articulate their needs. But, as I have argued, none of this is sufficient for citizenship as membership in a self-governing political community. Human beings have a stake in citizenship because their autonomy and well-being depend to a large extent on membership in a democratic polity. I am not convinced that domesticated animals have a similarly fundamental interest in recognition of their political membership and in the democratic constitution of the polity.

Transnational voting and multilevel citizenship

Unlike most contemporary political theorists, David Owen is not merely interested in defending general principles. He engages with comparative studies of democracy and offers a basket of concrete policy proposals. I share Owen's attitude. Normative theories become richer and more relevant when they address the puzzles and dilemmas that citizens and policy-makers face in real-world democracies. Comparing the variety of answers at different times and in different places cannot tell us what is the right solution. But it provides us with a broader view of options and constraints and focuses our minds on dilemmas that serve to test the robustness of our intuitions and principles.

My own thinking about the citizenship stakeholder principle did not start from the philosophical paradoxes of the democratic boundary problem. It was instead inspired by a puzzle about migrants’ access to citizenship status and voting rights that I have studied comparatively for many years. Liberal theorists like Walzer, Carens and Kymlicka have provided us with rich accounts of immigrants’ claims to membership and multicultural rights based on the idea that the rule of citizens over permanent strangers is a form of tyranny (Walzer 1983), that immigrants become over time members of society and democracies must be inclusive for all their members (Carens 1989), or that liberal democracies need to integrate immigrants into a shared societal culture that provides its citizens with a meaningful repertoire of options (Kymlicka 1995). These first generation theories ignored almost completely the well-known fact that international migrants retain their citizenship of origin. What has until recently been much less known is the strong global trend among source countries to let emigrants keep their membership and voting rights in home country elections even after becoming citizens of host countries.

Political theorists not only neglected the migrants’ perspective, which requires combining an immigration with an emigration view, but their theories of political community and citizenship were also overwhelmingly statist. The second set of puzzles that emerged from my comparative research interests was about freedom of movement between the member states of the European Union, which seemed to challenge assumptions about immigration control as a requirement for state sovereignty, and about local voting rights that many of these countries grant not only to all EU citizens residing in the municipality but also to third country nationals.

My first discussions of the stakeholder principle saw it as an alternative to immigration-centric views of social membership: migrants are simultaneous stakeholders in a country of origin and settlement, which supports their claims to multiple citizenship and voting rights. However, this did not resolve the puzzle about local citizenship and voting rights, where the most progressive practices instead follow a principle of automatic inclusion of all residents and exclusion of non-residents. This latter democratic practice seemed to be much more in tune with ASC than with ACS. From a normative perspective it is obviously not possible to accept that one democratic inclusion principle applies to states and another to local polities, since these principles aim to spell out basic moral ideas about democracy and membership. The solution I found consisted of two moves.

The first was to restate the stakeholder principle as entailing a relational correspondence between individual inclusion and collective self-government claims. This made it possible to explain why the same stakeholder principle supports different rules for membership determination and voting rights depending on the nature of the polity and its relations to other polities as conditions under which it can be self-governing (see chapter 1, section 4). This explains why birthright citizenship is an appropriate interpretation of the stakeholder principle for independent states, while ius domicilii is for local municipalities. The second move was to consider AAI, ASC and ACS not as rival principles responding to the same question about membership in the demos and the citizenry, but as complementary ones that cover inclusion in the different domains and stages of the democratic process (see chapter 1, section 3).

I want to use this general approach now to explain why I am sceptical about Owen's further differentiation of the authorial demos into an executive, legislative and constitutional one, which aims to bring back ASC as the relevant principle for determining membership in the demos in specific contexts. Owen agrees broadly with my view that enfranchising first generation emigrant citizens in national elections is generally permissible but not required. As Owen correctly clarifies, I think that a stakeholder principle supports presumptive inclusion so that justification needs to be offered for denying non-resident citizens the franchise. Owen wants to go further by identifying instances in which emigrants have a justice-based claim to voting rights. I have myself argued that this is the case when emigrants have been unjustly driven out of the country. In section 3.2 of chapter 1 I claim that, insofar as the situation of exiled citizens is comprehensively marked by past subjection, a democratic successor state may have duties to provide them with ongoing protection of their rights according to the ASC principle – an argument that Miller finds mysterious (p. 133) because he misreads it as a claim to political participation rather than protection. The reason why coerced emigrants also have claims to membership in the demos is, however, a quite different one that does not rely on ASC: denying them voting rights would amount to democratically ratifying the results of political purges or ethnic cleansing (Bauböck 2007).

Owen proposes a second set of circumstances under which expatriates would have a justice-based claim to inclusion in a constitutional demos. It is not entirely clear to me for what kind of decisions a constitutional demos would have to be specified. Owen identifies a narrow class of constitutional decisions that “directly [concern] [non-residents’] very status as citizens” and that “specify the entitlements and obligations of citizens – such as, for example, laws on nationality and expatriate voting rights” (p. 193; see also Owen 2010). Yet most constitutional laws that “fundamentally concern the nature of the political association” (p. 192) are not of this kind. They often concern procedural aspects of democracy or entrench fundamental human rights. Should the demos for such constitutional changes be different from the legislative one? If not, is there really a need and justification for a special demos deciding on laws that concern the status and rights of citizens? Even decisions about nationality law are ambiguous in this respect. As Owen points out, the Irish ius soli referendum of 2004 was about changing a fundamental aspect of nationality law but its impact was on future generations rather than today's citizens. Assuming for the sake of argument that the current disenfranchisement of Irish non-resident citizens is permissible: was there a good reason to extend voting rights to them only in this particular referendum?

The strongest prima facie case for a specific constitutional franchise exists if the decision would change the citizenship status and rights of those who are not included in the current legislative demos. Consider first the case of the 2014 Scottish independence referendum. Would it have been required to enfranchise persons who might have become Scottish citizens in case of a yes vote? No, because this inclusion would have illegitimately pre-empted a possible future composition of an independent Scottish citizenry that was still to be decided depending on the outcome of the very same referendum. It was therefore right to stick to the current legislative demos for Scottish Parliament elections (Ziegler, Shaw and Bauböck 2014). Consider now as a contrasting case the Brexit referendum. Was it also right to use then the current Westminster franchise? No, because that franchise excludes British citizens who have resided for fifteen years outside the UK, which conflicts with their EU citizenship right to free movement. It is perverse to disenfranchise people in member state elections because they exercise their rights as citizens of a union formed by these states. The British Parliament should have exempted British citizens residing elsewhere in the EU from the fifteen years limit for retaining the franchise. This argument applies, however, to the legislative as much as to the constitutional demos. There is again no need to distinguish these two. Finally, consider a hypothetical case of a country that has the kind of franchise and citizenship law for non-residents that I advocate: native-born expats remain enfranchised but their children born abroad do not obtain the franchise at majority unless they have taken up residence at some point during their lives. Imagine there is a referendum on whether second generation emigrants that fail to meet this condition should also be allowed to cast ballots from abroad in national elections. Since this is an important constitutional change and since their rights and status might change, should they be included in the constitutional demos for that particular vote?

In my view, the relevant difference is not between legislative and constitutional decisions, but between two types of decisions that fundamentally change the composition of the demos: those that are permissible and those for which a specific outcome is normatively required. Holding an independence referendum in Scotland and a Brexit referendum in the UK may have been politically irresponsible, but it was democratically permissible to let a referendum determine the outcome. In such a case, the legitimacy of the decision is purely procedural. And if the legitimacy is procedural, then the demos for such a decision should be composed in the same way as the legislative demos for the polity whose future status is to be decided.

A second type of decision is where a specific outcome is normatively required. The inclusion of categories of citizens that were unjustly excluded in the past, such as women or African Americans, is of that type. Consider the plebiscite that was held in Switzerland in 1971 and that extended voting rights to women. Was the decision illegitimate because it was taken only by the then enfranchised male citizens? Would a negative outcome of a referendum on enfranchising women become legitimate if women had been enfranchised for that particular vote?7 I suggest the answer to both questions is no. Including unjustly excluded categories in an ad hoc constitutional demos does not alter in any way the injustice of their exclusion from the legislative demos.

Consider now who should vote in a decision about extending voting rights to first generation emigrants. Imagine that Ireland, where they are currently disenfranchised, held a referendum on this question. There are many reasonable arguments on either side in this debate. But none of them strikes me as making a case that they must be excluded or must be included in the demos as a matter of justice. If their inclusion in the demos is permissible rather than required, then it can be legitimately decided by the current legislative demos, even if the decision concerns a fundamental democratic right. And if it were a matter of justice and for some reason must be decided by referendum rather than a constitutional court, then procedural legitimacy for the required change could still be better provided by the same legislative demos rather than an ad hoc constitutional demos whose composition pre-empts a desired outcome.

If it is generally permissible to decide democratically whether first generation emigrants should be enfranchised, then it follows that excluding foreign-born citizens by descent, whose ties to their parents’ country of origin is weaker, from the legislative demos cannot be unjust. This holds not only for the legislative demos but also for a constitutional demos that decides on their status and rights. A stakeholder principle provides similar substantive reasons for other decisions about expatriates that could be taken by a constitutional demos. If the decision at stake were to abolish ius sanguinis for the first generation born abroad, I would consider this unjust, independently of whether those affected are enfranchised; if the decision were about enfranchising second and later generations born and residing permanently abroad, I would consider such a proposal over-inclusive and would therefore a fortiori resist the ad hoc inclusion of these categories in a plebiscite on that very question.

I am more sympathetic to Owen's proposal to include foreign-born adult citizens if they cast their votes in the territory of the home state. My preferred condition for including such second generations is their return. A single trip “back home” to cast a vote is not the same thing as taking up residence, which turns the foreign-born into resident citizens. However, these are minor quibbles over practical criteria. Generally, I do not see why there would be any difference between returning for legislative elections or for a vote in a constitutional referendum.

Let me now consider Peter Spiro's objections to my proposal to phase out membership transmission with the first generation born abroad and not to enfranchise them automatically at the age of majority. There is a certain tension between Spiro's comments on external citizenship and his general diagnosis that citizenship is in decline because citizens are ever less interested in collective self-government. He sees diasporas as having a strong stake in their countries of origin.

There are some misunderstandings about what my proposal entails. Spiro quotes correctly my qualifier that source country citizenship should be phased out for third generations of immigrant origin “unless their parents have themselves renewed their links to this country through taking up residence there” (p. 69; p. 189). If second generations return to their parents’ country of origin, they become first generation emigrants in relation to their own country of origin and their children born abroad should be dual citizens by birth and for life. So there is no assumption of “linear flows” in my account. I do, however, insist that holding a stake in collective self-government is not a matter of strength of subjective preferences for a certain citizenship because such preferences are quite naturally influenced by instrumental motives. Instead, objective facts such as prior periods of residence or dependency on citizen parents or partners who have previously resided in the country provide better proxies for a presumption of stakeholding.

What I find surprising about Spiro's account of diaspora is that he seems to think that being outside a state territory enhances one's interest in the flourishing of the political community whereas being inside diminishes it. Spiro says that “it is not clear that they [i.e. diasporas] have a self-governance interest in the community [of residence] defined in terms of the state, much less its ‘flourishing’ ” (p. 208). At the same time, he sees diasporas as “the paradigm case in which individuals will feel exactly the sort of stake that Bauböck describes” (p. 217), citing the intense demand for multiple citizenship and voting rights. This contradicts somewhat Spiro's later claim that “the average level of interest and participation will be lower in external communities, reflecting the aggregately lower, more highly variable self-governance interest” (p. 220). I find the latter statement plausible and it is supported by evidence about low turnout among external voters. As Brubaker has reminded us, we should not attribute diasporic identities to all individuals of emigrant origin (Brubaker 2005). Diaspora formation happens in specific contexts, mostly in response to violent conflict in the home country or segregation in the host country. In any case, even if communities mobilized as diasporas do not care much about citizenship in their country of residence and care a lot about politics in the country of origin, they should still be seen as having a stake in both.

Spiro advocates bespoke arrangements that reflect political pressure exercised by diasporas. But combined with domestic fears of excessive diaspora influence, these arrangements are likely to produce deviations from the ideal of equal individual citizenship. Extraterritorial electoral constituencies with reserved seats in parliament often under-represent external citizens numerically by weighting down their votes compared with domestic constituencies, while at the same time enhancing their substantive representation as a special interest group. Both features are problematic, although the former goal is possibly defensible where diasporas are very large compared with domestic populations (Bauböck 2007). From a stakeholder perspective, external citizens should have a vote only if and as long as they can be seen as sharing a common interest in the flourishing of the polity with those who reside in the territory. Their special interests do not warrant special representation in legislation any more than those of other special interest groups inside or outside the country who may be affected by its legislation. The MPs elected in extraterritorial constituencies in the Colombian, Croatian, Ecuadoran, French, Italian, Portuguese and Romanian parliaments do not merely vote on issues concerning expats; they participate fully in agenda-setting and decisions concerning all citizens.

Spiro is, however, right that setting and maintaining standards of equality among citizens become more difficult once we consider individuals’ multiple ties to different polities across levels or international borders. This problem is exacerbated by transnational migration but it already exists within democratic states which are subdivided into self-governing territories. Most obviously, in democratic federations there is a tension between treating all citizens of the federation equally and treating the federal entities (provinces, cantons, regions, states) equally as constituent polities of the federation. This tension is institutionalized rather than resolved in bicameral legislatures. In the U.S. the votes of individual Californians and Rhode Islanders count equally for representation in the House, but unequally for the Senate, where each state has two delegates independently of its size. Unlike federal entities, external citizens do not form distinct constituent polities with a claim to collective representation – although they are sometimes labelled the (n+1)^th province. Yet migration does create unequal stakeholdership between sedentary citizens with a singular affiliation to their state of residence, transnational migrants with a claim to dual citizenship and highly mobile persons who may not satisfy reasonable residence requirements for citizenship in any host country. Working out what equality entails in these contexts is a difficult task. However, it does not seem to me an impossible one as long as we keep the relational nature of stakeholder citizenship firmly in mind.

After discussing external citizenship and voting extensively, let me now consider more briefly voting rights for non-citizen immigrants, where it seems that Owen and I have a deeper disagreement over principle rather than policy. As we have seen above, with regard to emigrant citizens Owen wants to defend the stakeholder principle against my proposal to restrict voting rights for second generations who have not “touched base” with their country of citizenship. In his view, “ACS not only demarcates those entitled to citizenship, but it also simultaneously identifies those who are entitled to authorial membership of the constitutional demos” (p. 194). When discussing the claims of immigrants, however, he relies on Dahl's principle of full inclusion of non-transient residents with the necessary cognitive capacities and claims that ASC is the right norm for determining membership in the legislative demos. In policy terms, our disagreement seems a minor one: should immigrants who decline a fair naturalization offer still be entitled to vote in national legislative elections?

Let me first clarify again what may seem like an inconsistency: I have argued in chapter 1, section 4 that non-citizen residents should be included in the local demos. Does this mean that I apply ACS at the national level while accepting ASC at the local one, as Miller suggests in his comments? No, because foreign residents are non-citizens only in their relation with the national polity, whereas they should be regarded as full citizens at the local level where citizenship itself is derived from residence. At the national level, this is not the case and non-citizen residents are offered a choice to naturalize as a condition for acquiring voting rights. Owen defends instead that non-citizens who are not merely transients should be automatically enfranchised.

My objection is that this move would undermine the representative relation between citizenry and authorial demos that requires that the latter should be a subset of the former (as explained in the previous section). The members of the demos are those who exercise the core right of participation in collective self-rule of the political community. Only citizens should be able to rule over citizens and to authorize the rulers on behalf of all citizens.

Owen might point out that in some empirical cases non-citizens who meet certain residence requirements can vote (although not be elected) in national elections. I have endorsed the view that it is legitimate for democratic states to extend the vote to long-term residents who have not opted to become national citizens (Bauböck 2015).

In the empirical cases we find two reasons why national voting rights have been granted to non-citizens: one is where voting rights were not seen as an alternative to naturalization but as a pathway to it;8 the second is where historical voting privileges for special nationalities were preserved or – in the unique case of New Zealand – extended to all long-term residents (Barker and McMillan 2014). None of these empirical cases lends itself to inductive generalization that liberal democracies endorse a national franchise for non-citizens who do not intend to become citizens.

Of course, a lack of empirical support is not sufficient to knock down a normative argument. Owen makes a pragmatic case that granting naturalization refusers voting rights “supports the conditions of genuine consent by weakening an instrumental reason for resident non-citizens to naturalize” (p. 202).
However, if immigrants who enjoy secure residence and all other rights desire to participate fully in the political life of the community by voting or running as candidates, then they have a non-instrumental motive for naturalization. They want to become members of the demos and it is not clear why they could reasonably reject to become also citizens in terms of their legal status.

Finally, there may be good reasons for democratic states not to enfranchise those who refuse to become citizens but still want to vote in national elections. Imagine a country (call it Luxembourg) where half the population consists of resident citizens and the other half of non-citizen residents who enjoy more or less the same rights apart from the right to vote in national elections.9 Assume further that the non-citizen long-term residents could at any time decide to become citizens while keeping their citizenship of origin.10 The reason why they do not want to naturalize is then presumably that they are not sure that they will stay in the country for much longer. Being aware that national citizenship is generally a lifetime status and lacking instrumental motives for acquiring it, they do not want to take such a step. This country faces a democratic dilemma. On the one hand, there is certainly a deficit of legitimacy if only 50 per cent of the resident population is eligible to vote in national elections. On the other hand, is it really a requirement of justice on grounds of ASC to include them in the legislative demos? Wouldn't it rather undermine trust among the citizens if a large number of those who participate in elections do not commit themselves to be members of the political community for a long-term future? Couldn't the citizens point out that this amounts to being ruled by non-citizens who – by rejecting the naturalization offer – have demonstrated a lack of commitment towards the political community? Note that this second horn of the dilemma comes into sight only from a citizenship stakeholder perspective whereas it vanishes from sight if we cling to a principle of including as equal members all subjected to the same government.

I find it hard to resolve this dilemma either way, which is why I consider it permissible but not required to introduce non-citizen voting rights at the national level. Behind this dilemma lurks my deeper worry about the future of democratic citizenship in a hypermobile world where majorities of citizens are non-residents and majorities of residents are non-citizens.

In our present world, which is certainly not hypermobile, ius domicilii regimes at the local level are embedded within the territorial jurisdictions of states and their birthright citizenship regimes. Local citizenship helps democracies to “digest” hypermobility and superdiversity without endangering the transgenerational continuity of national citizenries. Contrary to Peter Spiro's reading of my view, I do not regard local citizenship as the “stepchild” of national citizenship. I see them as a symbiotic pair, so that the strengths of each compensate to some extent for the defects of the other. But this symbiosis might be threatened in a hypermobile world, which could be imagined as consisting only of provinces and municipalities under a world government without independent states. Ius domicilii would then remain as the only relevant rule for determining membership in the citizenry and the demos of territorial polities, while exclusionary birthright and naturalization regimes would probably proliferate within non-territorial associations and communities (Bauböck 2011). This is probably a fair description of the world towards which Peter Spiro thinks we are heading.

The future of citizenship

Peter Spiro's disagreement with me seems to emerge to a significant extent from arguing in a different register. While my theory is explicitly normative, his approach is more predictive. Spiro is convinced that citizenship is in decline or even already moribund. I am not convinced, but if he were right then I would be interested in how democracies should react to such a decline.

Spiro says he is “not proposing an alternative theory of citizenship” (p. 218). This is a pity, since I would be interested in knowing what follows normatively from his diagnosis of the demise of citizenship. A normative theory can best be refuted by a rival normative theory. Yet since “ought implies can”, an applied normative theory, such as the one that I propose, can also be challenged on empirical grounds. If it were no longer possible to apply norms of inclusion and equality of citizenship to bounded and self-governing political communities, then my argument would be pointless. In fact, I acknowledged myself that what I called the “contexts of democracy” – territorial jurisdiction and relative sedentariness – are historically contingent. Elsewhere I reflected on the scenario of a hypermobile world in which these contexts would have vanished (Bauböck 2011). My conclusion then was that this is a dystopia in which not only citizenship but democracy itself would be much more difficult to realize. Spiro ends his comment by calling on theorists to “turn their sights towards carrying citizenship values forward to novel institutional arrangements” after the death of citizenship itself (p. 222). I would be keen to follow this call, but rather as an exercise in counterfactual reflection, since I am not at all convinced of Spiro's diagnosis of mortal illness.

In contrast with Spiro, I see the proliferation of citizenship statuses and rights across international borders and across levels of democratic polities not as a decline but as a liberation of citizenship from the straitjacket of state-centred nationalism – a liberation that creates new potentials for realizing democratic self-government. The new opportunities create also new challenges. Spiro correctly identifies the most fundamental of these: how to reconcile a transnational and multilevel conception of citizenship with the core value of equality.

Before I address this problem, let me say that I am not naively optimistic about the prospects for citizenship, because I also see a serious trend towards decline. This is not the toleration of multiple citizenship or the emergence of sub- and supranational citizenship, all of which demonstrate the strength of the concept and its ongoing role in structuring the contemporary political world. It is instead the trend towards instrumental uses and abuses of citizenship both by states and by individuals. More precisely, it is the instrumental use of citizenship by states that creates new opportunities of instrumental abuse by individuals. Citizenship is not by nature a market commodity. States and other polities have the exclusive monopoly of producing it. If they start to sell their citizenship in a global market, then the value of this commodity depends entirely on the recognition of the status by other states. Investor citizenship thus epitomizes the over-inclusive side of this trend (Shachar and Bauböck 2014; Dzankic 2015; Parker 2017), but it also has an exclusionary side that is generally ignored by Spiro. One example is when governments restrict access to ius soli or to naturalization for refugees or family members of citizens in order to curb undesired migration inflows. Spiro himself mentions another instance where restrictions on citizenship create incentives for instrumental abuse by individuals: long-term residents who decide to naturalize shortly before returning to their country of origin in order to secure re-entry rights and – I would add – diplomatic protection in their country of origin. Isn't the problem here that the country of origin's non-toleration of dual citizenship and lack of protection of human rights (China is the best-known case) create perverse incentives to use U.S. or Canadian citizenship instrumentally? And would toleration of dual citizenship and respect for fundamental rights not remove this incentive?

This broader point can be illustrated by low naturalization rates of EU citizens in other member states because of the absence of incentives for instrumental naturalization. Free movement rights, non-discrimination and respect for fundamental rights remove most instrumental reasons for changing citizenship between member states. Of course, they greatly enhance the instrumental value of EU citizenship for third country nationals, but for EU citizens they reduce the value of national membership to its intrinsic and identity components. My conclusion is that in a world with more freedom of movement between states and better protection of rights within them, citizenship would have little instrumental value but would not lose its intrinsic value as membership in a self-governing polity. Instrumental (ab)uses of citizenship result from a combination of enhanced interconnectedness and mobility between states and persistent disparities between them. From a normative perspective, we should try to reduce the disparities instead of abandoning citizenship.

Instrumental uses of citizenship diminish its intrinsic value as membership in a self-governing polity. They do not, however, change the background context within which this value makes normative sense and could be defended by the right kind of policies. Spiro's and my diagnosis of current threats to the value of citizenship differ in this regard and mine does not commit me to stop arguing for the therapy.11

The decline of citizenship is, in Spiro's view, just a symptom of a more fundamental change, which is the demise of the agent issuing citizenship certificates and passports: the state. The redistributive capacities of the state may be waning, the security imperative in relation to hostile competitor states has dissipated. “States increasingly serve as an administrative servant of the global system” (p. 212). The protection of individual autonomy is “now protected by substantive, exogenously imposed human rights more than procedural internally generated self-governance” (p. 208) and even states’ “own membership practices are increasingly set by exogenous agents” (p. 213).

Individuals still have an interest in public order but “it is not clear that they have a self-governance interest in the community defined in terms of the state, much less its ‘flourishing’.” (p. 208) So individuals’ interests in autonomy and protection are increasingly realized in other associations and communities than the state. Some of these (Spiro cites the Mormon Church) have greater capacities and enjoy better compliance of their members with associative duties. Under these circumstances, for most citizens “politics is a waste of associative energy” (p. 210).

Here are a few questions that I have about Spiro's account. First, who governs the global system of which states are now administrative servants? How are human rights maintained and secured except by states subscribing to them and committing to upholding them? Who are the exogenous agents that are independent of states? Consider the UN, the UNHCR and the Council of Europe, that is, those organizations that have tried – and to a degree successfully so – to constrain state practices in matters of nationality law through international legal norms. Aren't these still international organizations made up of states which agree on their mandate and which reserve for themselves the right to accept or reject the norms produced by these IOs?

Second, voluntary associations have often been regarded as schools of citizenship, but Spiro sees them instead (as did Rousseau) as its grave diggers. I remain agnostic on this point. My main objection is that polities are entirely different from voluntary associations and civil society communities and there is thus neither essential rivalry nor harmony between them. First, as I argue in chapter 1 section 4, none of the three basic membership rules in self-governing polities (birthright, residence, derivation) transforms polities into voluntary associations. Second, polities have coercive power over anybody within their jurisdiction to enforce laws emerging from an open legislative agenda. Self-government in voluntary associations may satisfy important psychological needs but it cannot substitute for the task of authorizing and taming government power through subjecting it to the votes of citizens.

Finally, I do not see any empirical evidence that citizens are merely interested in public order but no longer in self-government or the flourishing of their states. To the contrary, the rise of right- and left-wing populist parties in the U.S. and Europe seems to me – in a perverse way – proof that citizens expect much more from the state than maintenance of public order. We are witnessing a rather strong counter-movement against the thinning out of democracy that reclaims it in the name of popular self-government. Populists’ appeal results from their promise to restore the power of the people against the elite, to purify the people by purging them from alien elements, and to restore the independence of the state from transnational corporations and supranational bodies like the EU. This is illiberal democracy running amok. But it is certainly not proof that citizens have lost their self-governance interest in the community defined in terms of the state.

Continuing the conversation

I am deeply grateful to my interlocutors, not only for their praise but especially for their frank critiques. They have made me acutely aware of the gaps in my argument. And my thoughts about democratic inclusion have already changed in some respects due to their feedback. I would love to know what they have to say about my self-revisions and rebuttals. Yet books do not only have their own fates (I wish I could say that about this one!), but they also have their own formats. Unlike the open-ended exchange of arguments in public spheres, a book is a container with limited space for ideas. The dialogue closes once the last page has been turned.

My hope is that the conversation about principles of democratic inclusion can be carried on in other forums and not just among theorists. I started my rejoinder by claiming that normative political theory ought to be informed by ideals that can guide reforms in the current world. If this is correct, then the most important test of the theory I have proposed is whether it can be accepted by reasonable and engaged citizens.