THE DEMOCRATIC BOUNDARY PROBLEM RECONSIDERED

Who should have a right to take part in which decisions in democratic decision making? This “boundary problem” is a central issue for democracy and is of both practical and theoretical import. If nothing else, all different notions of democracy have one thing in common: a reference to a community of individuals, “a people”, who takes decision in a democratic fashion. However, that a decision is made with a democratic decision method by a certain group of people doesn’t suffice for making the decision democratic or satisfactory from a democratic perspective. The group also has to be the right one. But what makes a group the right one? The criteria by which to identify the members of the people entitled to participate in collective decisions have been surprisingly difficult to pin down. In this paper, I shall revisit some of the problems discussed in my 2005 paper in light of some recent criticism and discussion of my position in the literature, and address a number of new issues. Resumo. Quem deve ter direito a participar em que decisões no processo democrático? Este “problema da delimitação” é uma questão central para a democracia, e tem importância tanto teórica como prática. Todas as diferentes noções de democracia têm, pelo menos, uma coisa em comum: uma referência a uma comunidade de indivíduos, “um povo”, que toma decisões de forma democrática. No entanto, que uma decisão seja tomada de acordo com um método democrático por um determinado grupo de pessoas não é suficiente para que a decisão seja democrática ou satisfatória numa perspectiva democrática. O grupo tem também de ser o grupo certo. Mas o que é que faz com que um grupo seja o grupo certo? Tem sido surpreendentemente difícil determinar os critérios com os quais devemos identificar os membros do povo que têm o direito de participar em decisões colectivas. Neste artigo, irei revisitar alguns dos problemas discutidos no meu artigo de 2005 à luz de críticas recentes e da discussão da minha posição na literatura, e abordarei algumas questões novas. * Institute for Futures Studies; http://www.iffs.se/forskning/forskare/gustaf-arrhenius. Ethics, Politics & Society


Introduction
Who should have a right to take part in which decisions in democratic decision making?This "boundary problem" is a central issue for democracy and is of both practical and theoretical import. 1 If nothing else, all different notions of democracy have one thing in common: a reference to a community of individuals, "a people", who takes decision in a democratic fashion.However, that a decision is made with a democratic decision method by a certain group of people doesn't suffice for making the decision democratic or satisfactory from a democratic perspective.The group also has to be the right one.But what makes a group the right one?The criteria by which to identify the members of the people entitled to participate in collective decisions have been surprisingly difficult to pin down.
Resolving the boundary problem raises a number of theoretical problems, some, it seems, quite intractable.
Although the boundary problem is a fundamental issue in democratic theory, surprisingly little attention has been given to it in the classical canonical treatises on democracy.As Robert Dahl put it in the seventies, "how to decide who legitimately make up 'the people' (…) and hence are entitled to govern themselves (…) is a problem almost totally neglected by all the great political philosophers who write about democracy" (Dahl 1970, 60).Cf. (Dahl 1989, 119ff), which is rather surprising.Actually, very little had been written about this topic until Dahl's own work on it and Frederick G. Whelan pioneering paper in the eighties.
In the last ten years or so, however, there has been a significant and welcome improvement and there is now a bourgeoning literature in the area.In this paper, I shall revisit some of the problems discussed in my (Arrhenius 2005) paper in light of some criticism and discussion of my position in the new literature on the boundary problem and address a number of new issues. 1(Robert Dahl 1989) refers to this problem as "the problem of the unit " (p. 193), "the problem of inclusion" (p.119), and sometimes as the "boundary problem" (pp. 146-7).(Robert Goodin 2007) calls it "the problem of 'constituting the demos'" and others have referred to it as the "demos problem".Frederick G. Whelan calls it "the boundary problem" in his (1983) pioneering article on the subject, and so shall I. (Goodin 2007 fn.1), thinks that "calling it 'the boundary problem' makes the issue seem more a matter of geography than it necessarily is".As should be clear below, I don't conceive of the boundary problem as a matter of geography and I use the term "boundary" in its general sense, like, for example (Frege 1970, 159), when he talks about concepts having "a sharp boundary" or as it is defined in Oxford English Dictionary: "That which serves to indicate the bounds or limits of anything whether material or immaterial; also the limit itself".

Practical and Theoretical Boundary Problems
In difference from many philosophical problems, the boundary problem is clearly also a pressing practical political problem.For example, what is the relevant constituency for a democratic solution to the Northern Ireland conflict? 2 Should a treaty be approved by the citizens (or their representatives) of Northern Ireland alone or should it involve those of the United Kingdom and the Irish Republic as well?The latest treaty -"the Good Friday Agreement" -was subject to a referendum in Northern Ireland and the Republic of Ireland whereas the citizens of Great Britain were represented by their government.It is hardly a solution acceptable for an old-style Unionist, since she would prefer a referendum in United Kingdom of Great Britain and Northern Ireland or perhaps only in Northern Ireland. 3 Yet, such a referendum would not impress an Irish nationalist who would consider these boundaries arbitrary and illegitimate, nothing more than a kind of international gerrymandering.Still, both the Unionist and the Irish nationalist could be dedicated democrats in the sense that they think that a fair solution should be based on a democratic referendum. 4  The proposed secession of Quebec from Canada, Scotland from UK, UK from EU, Catalonia from Spain, Kosovo from Serbia-Montenegro, and the like, raise similar problems. 5The boundary problem also arises in the context of migration.For instance, consider the dilemmas confronted by host states.What voting rights, if any, should non-citizens have?Should they enjoy these rights only at certain levels-say, local rather than national elections-or only over certain issues (Beckman 2006)?Analogous questions arise from the perspective of migrants' countries of origin (Grace 2003;Bauböck 2006Bauböck , 2007;;Rubio-Marín 2006;López-Guerra 2005, 2014).Do long-term emigrant diasporas in Europe, 2 (Whelan 1983, 23) discusses this example at length. 3 It is telling that in the referendum about the treaty, an estimated 96% of the Catholics supported it whereas only 52% of the Protestants gave it its blessing.See (Encyclopaedia Britannica 2014) 4 And what of the millions of Irish who left the island, in part because of the conflict? 5It is interesting albeit worrisome to note how rapidly the number of practical instances of the boundary problem have increased in the short time span since my (Arrhenius 2005) paper.
for example Turkish and Kurdish communities in Germany, have a right to participate in democratic decision-making in their homelands?
The boundary problems also appear in more humdrum small-scale cases.
Consider for instance the local referendum about congestion charges in Stockholm.Was it right that only people living in the city of Stockholm had a vote?One might claim, as many did, that the inhabitants in the surrounding suburbs who on a regular basis commute to central Stockholm also should have had the vote.
It is perhaps not equally clear that the boundary problem is an issue in the justification and legitimacy of democracy but consider how questions like the above should be decided democratically.One suggestion could be to have a referendum about who should have say in these questions.But who should be allowed to take part in such a referendum?And so on without any end, we seem to end up in an infinite regress.This chain of reasoning has led some to draw quite gloomy conclusions regarding both the ability of democratic theory to solve the boundary problem in a satisfactory manner and the scope and the legitimacy of democratic decision-making.In his pioneering paper on the boundary problem, Frederick G. Whelan, concluded that: (…) democratic theory cannot itself provide any solution to disputes that mayand historically do -arise concerning boundaries.(….)It may not be surprising that democracy, which is a method for group decision-making or self-governance, cannot be brought to bear on the logically prior matter of the constitution of the group, the existence of which it presupposes.Nevertheless, strong claims are frequently made for democracy, both by its philosophical advocates and by ideologues and activists of the modern world; democracy is commonly put forward as the sole foundation of legitimate government, and as the sole legitimate method to make binding public decisions of all sorts.(….)The boundary problem does, however, reveal one of the limits of the applicability of democracy, and acknowledgement of this may have the beneficial effect of moderating the sometimes excessive claims that are made in its name. 6  Likewise, Dahl stresses that "we cannot solve the problem of the proper scope and domain of democratic units from within democratic theory" and that "in solving this particular problem [the problem of constituting the people] democratic theory cannot take us very far.Democratic ideas, as I have said, do 6 (Whelan 1983, 40,42 my emphasis).(Näsström 2003(Näsström , 2004) ) repeats the same claim.not yield a definitive answer" and it cannot be solved even by "reasoned inferences from democratic principles and practices". 7 Although I don't agree with Whelan's and Dahl's gloomy conclusions, as I shall discuss below, I indeed agree that the boundary problem reveals a problem at the heart of the very idea of democracy.That a decision is made with a democratic decision method by a certain group of people (or by an elected assembly that represents the group) doesn't suffice for making the decision democratic or satisfactory from a democratic perspective.The group also has to be the right one.But what makes a group the right one?
Actually, the problem I have referred to above as "the" boundary problem is just one among a number of such analogous problems.The perhaps most discussed boundary problem concerns people's capabilities as political agents, their political competence.In order to effectively further one's interests through democratic processes one must, arguably, possess a certain degree of knowledge and rationality.The question then becomes how we should decide the relevant political competence for membership in the political community.In practice, this minimal competence is often approximated by age, but one could ask which age is the appropriate one, and whether the same age limit is appropriate for all democratic processes, e.g., parliamentary elections versus local elections.Moreover, we could not only ask who should have a right to vote but also who should have a right to run for office, and whether these rights go together (historically, this has often not been the case).Another boundary problem concerns beings that lack the capacity to take part in the democratic process but who are going to be affected by policies adopted and that could be represented by proxies, for example minors, future generations, and animals.8Members of the two first groups are likely to have the same political competence as present people, but they cannot take part in the democratic process since they are simply not around.
7 (Dahl 1989, 207, 209) In Dahl's terminology, the "scope" of a democratic unit is the set of matters that are to be decided by it, and the "domain" is the set of persons who comprise it.Cf. (Barry 1991).
I shall not discuss these problems here, but my suggested solution to Whelan's boundary problem ---Which people, given that they are politically competent and can take part, ought to have a right to take part in which decisionmaking processes?----has clear implications for how we should approach these boundary problems too. 9

Schumpeter and Ross: A Historical Detour
Before turning to the discussion of a general answer to the boundary problem, it will be useful to look at two classical definitions of democracy.As mentioned by Dahl in the quote above, the boundary problem has been almost ignored by the great theorist of democracy in the past.A case in point is Joseph Schumpeter's revisionist definition of democracy: The democratic method is that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote.(Schumpeter 1976, 269).
Schumpeter is here discussing democratic governance of states and developed his definition in analogy with firms in a capitalist market: Firms compete with others to sell products at a profitable prize and only the most competitive firms survive.As he writes elsewhere "(…) we have restricted the kind of competition for leadership which is to define democracy, to free competition for a free vote.---Free, that is, in the same sense in which everyone is free to start another textile mill" (Schumpeter 1976, 271 and 272, fn. 6).He developed his definition by looking at states that people called "democratic" and extracted what he thought they had in common.Hence, his definition is based on a denotation (extension) analysis of the term "democracy" as it was used in his time and environment and as such, I surmise, quite accurate.10 9For reasons that will become clearer below, I also think that the answer to problem of what kind of majority ---simple, qualified, or unanimity ---that should be required to pass, abolish, or amend certain laws, and of what kind of voting system that should be used ---direct democracy, representational democracy, plurality voting, Borda count etc ---is analogous to the answer I'm going to give below to the boundary problem.
Notice, however, that nothing is said about the boundary problem, that is, who has a vote or for whose vote one can compete to acquire power.Schumpeter's denotation analysis and definition are remarkably incomplete in this respect but also, I'm afraid, representative of the definitions that have been proposed in the literature. 11  One might think that Alf Ross' definition is an exception.He defines an "ideal type" of democracy with three dimensions that can be fulfilled to varying degrees: 1. [I]ntensity, that is with respect to the size of the population of people who are allowed to take part in referenda and elections.---2.[E]fficiency, that is, with respect to the effectiveness of the popular will in deciding issues.---3.[E]xtensity, that is, with respect to the scope of popular influence and control over the different branches of government. 12  Here, one might think that the clause regarding Intensity answers the boundary problem: The more people that are allowed to take part in a vote, the greater the degree of democracy.However, this is not exactly what Ross had in mind since he has presupposed a constitution of a "people" and Intensity is the percentage of the "people" who are (legally) allowed to take part in a vote.As he writes just before the passage quoted above: "(…) the people's influence on the exercise of public authority, can vary with respect to(...)". 13  So at least with respect to Schumpeter and Ross, Dahl is right that the boundary problem has been ignored in the canons of democratic theory.Let's 11 The same holds for (Schumpeter 1976, 250) statement of the "classical" definition of democracy: "The democratic method is that institutional arrangement for arriving at political decisions which realizes the common good by making the people itself decide issues through the election of individuals who are to assemble in order to carry out its [the people's] will".Again, nothing is said about what constitutes the people. 12See (Ross 1968, 101-2), my translation, italics in original.
13 (Ross 1968, 101), my translation.Still, Ross definition seems to have something to say about the boundary problem inside a demos and thus it might have interesting implications regarding, for example, economic democracy.For a discussion, see (Arrhenius 2012(Arrhenius , 2017)).
now turn a possible answer to the boundary problem and its implication for how we should conceptualise democracy.

Democracy as a Normative Ideal or Decision Method
It will be useful to consider an important distinction between two ways of understanding democracy which unfortunately hasn't been observed sufficiently in the discussion.In general, we should distinguish between normative ideals, on the one hand, and practical decision methods or rules for regulating social interactions (e.g., social norms, laws, institutions), on the other hand. 14Roughly, a normative ideal states the ultimate goal that that we strive towards, such as the just or good society (i.e., the considerations that ultimately make actions, policies, institutions etc, right, just, or fair), whereas a decision method is a strategy for decision-making which we use to achieve the goal specified by the ideal.We use the normative ideal, in conjunction with empirical considerations (e.g., economical and psychological facts), to evaluate and rank alternative decision methods, social norms, laws, institutions, etc., for different situations and contexts, in respect to how well they would promote the ideal.In that sense, the application of a certain decision method is justified by our normative ideal whereas the ideal is justified by being in accordance with our considered normative judgments and by satisfying other relevant epistemological and methodological criteria. 15Take utilitarianism as an example.According to utilitarian ideal, we should maximise people's well-being, or expected well-being.Now, a common complaint against utilitarianism contends that it is self-defeating since in many instances it is practically impossible to calculate the value of the outcomes of the alternative actions available to a person.Consequently, if we try to apply the utilitarian principle in every single case, we are likely to choose the wrong action since our calculations are bound to be wrong.This is, however, no argument against 14 For a related distinction between criterions of rightness and decision methods, see (Bales 1971) and (Danielsson 1974, 28-29).
utilitarianism as a normative ideal but an argument against utilitarianism as a practical decision method.One can still accept utilitarianism as a normative ideal but hold that in practical deliberation, we have to rely on "rules-of-thumb" and approximations: help those that are in distress, be honest, do not break promises, obey the law, etc., on the individual level, and rule of law, democracy, independent mass media, and so forth, on the institutional level.Whether we should accept these practical decision methods, on the other hand, have to be judged against the utilitarian ideal in a "cool hour" when we have enough time and resources to evaluate the consequences of the general application and implementation of these decision methods relative to the goal specified by the ideal.Likewise for an advocate of equality of resources, well-being, power, etc, and other normative ideals.
There are two important lessons to draw from this example.Firstly, one can reject a theory as a decision method but still accept it as a normative ideal and vice versa.Secondly, even if we don't find a particular decision method satisfactory in regard to some case, it doesn't follow that it isn't useful in other cases.The utilitarian decision method, for example, might be a good one for some governmental bodies.Again, this has to be decided by evaluating the decision method against the ideal.A normative ideal, on the other hand, we expect to be applicable to any actions, rules, or institutions, that falls under its domain without exceptions.
What does this distinction have to do with democracy?A theory of democracy can also be taken either as a normative ideal or as a practical decision method.As R. J. Pennock puts it succinctly in a discussion of Wollheim's paradox: "One must distinguish at the outset between democracy as an ideal and democracy as a practical device for approximating the ideal" (Pennock 1974, 88).
If one wants to defend democracy as a normative ideal, for example as an idea about fair distribution of power, then one needs to show that it is in a reflective equilibrium with our considered judgments about democracy and power.Such a normative ideal need not be directly applicable to choice-situations in the real world.Rather, we use the normative ideal, in conjunction with empirical considerations, to evaluate and rank alternative practical decision methods for different situations and contexts.
For those who study how democracy works in practice, it is probably more common to view democracy as a kind of decision method, as a matter of institutional arrangements.Schumpeter is a case in point.Implicit in much reasoning about democracy, however, is also the idea that democracy is a kind of normative ideal.For example, it is presumed in many lofty political declarations, often expressed in terms of justice and equality. 16Although I shall not dwell much on the details of such a theory in this paper, I think that the most promising approach is to take democracy as a normative ideal concerning fair distribution of power. 17 Let me here take the opportunity to point out that there is an ambiguity in the discussion of the boundary problem. 18As we have formulated the problem, it concerns who should have a right to take part in different decisions.This can be interpreted in at least three ways.On one interpretation, it is about who should have a right to take part, all things considered, that is, when we have taken into account all relevant moral and political aspects (efficiency, prosperity, freedom, equality, etc.).Complete normative ideals such as utilitarianism and Rawlsian liberalism answer this question.On a second interpretation, it is about who should have a right to take part given that other important normative ideals are not compromised too much.Here, the answer to the boundary problem yields a normative pro tanto reason for including and excluding people in a democratic decision process.Both of the above interpretations take the boundary problem as a normative problem.On a third interpretation, the boundary problem concerns the conceptual question regarding who should have a right to take part for a decision procedure to be democratic or more democratic than another procedure. 19Ross' theory discussed above is an example of a partial answer (inside a given demos) to this interpretation of the boundary problem.Here we are discussing 16 See (Naess, Christophersen, and Kvalø 1956) for a list of such slogans. 17The best developed version of this kind of ideal in the literature is (Brighouse and Fleurbaey 2006).Two other examples are Danielsson's (1974) suggestion to take problems of preference aggregation, such as Arrow's impossibility theorem, as problems of just distribution of influence, and Christiano's (1996Christiano's ( , 2002) ) theory of democracy as an ideal of equal chances to affect the outcome. 18Including my (Arrhenius 2005). 19For similar distinctions, see (Beckman 2009).
democracy not as a normative ideal but in terms of a non-normative ideal type of democracy (just as we can give ideal type definition of a circle).In itself, an answer to this problem has no normative implications since it says nothing about who ought to be given a say, all things considered.
A connection between the two interpretations is often presumed, however, by an implicit normative premise according to which a decision ought to be taken as democratic as possible given that other important values would not be too compromised.Democracy is thus understood as a partial normative ideal that must be weighed against other partial normative ideals to yield an answer to the problem of who should have a right to take part in a decision, all things considered.
Unless otherwise indicated, I shall take the boundary problem in the latter way below.The answers to this problem will thus specify who should have a right to take part in a certain decision in order to make it more democratic, but also who ought to have a say given that other important normative ideals are not compromised too much.For reason of space, I have to leave the interesting question of how to weigh the democratic ideal against other ideals to another time.

Democracy as Decision Method and the Boundary Problem
Does Whelan discuss democracy as a decision method or as a normative ideal?Whelan's position is unclear on this issue.He sometimes writes like he has a normative ideal in mind, for example, when he talks about democracy as "the sole foundation of legitimate government" (Whelan 1983, 40).Someone who takes democracy as a practical decision method justified by a normative ideal doesn't hold that democracy is the foundation of legitimate government but that its legitimacy derives from the normative ideal.For utilitarians, for example, democracy (of some kind) is justified if and only if it maximises people's wellbeing as compared to alternative decision methods.For Rawlsian liberals, to take another example, democracy is justified (roughly) if it is the best decision procedure for the safeguarding of basic civil liberties, equal opportunity and the well-being of the worst-off.For Nozickian libertarians, democracy is justified insofar it respects people's property-rights, and so forth for other normative ideals.
On the other hand, much of Whelan's writing seems to concern democracy as a decision method.For instance, he spends quite a lot of space on discussing boundary principles based on territory, nationality, culture, or geography.These principles are quite obviously poor candidates for a boundary principle seen as part of a democratic ideal.
The territorial state principle is an illustrative example.According to this principle, we should just take existing territorial states as a given and include every person residing or born in a particular state in the democratic process governing that state.As Whelan himself points out, the territorial state principle cannot give any guidance in cases where borders of territorial states are in question.It falsely assumes that we can take boundaries between territorial states as something already fixed and undisputed.Needless to say, territorial disputes are frequent in human history and some of the most tragic and bloody conflicts in the present revolve around the issue of establishing boundaries where no entrenched territorial boundaries exist.
Secondly, the territorial state principle has a very limited scope.Arguably, any social union, from the world community to the family, is part of the domain of democracy, i.e., are candidates for being democratically organised. 20The territorial state principle only addresses one particular boundary problem and leaves open the question of how to delimit participation in decision making procedures in other social unions.
Thirdly, even if we lived in the best of possible worlds where all territorial boundaries were settled, these boundaries would still be irrelevant and ad-hoc from a normative point of view.Suppose the U.S. Government decides to resume atmospheric nuclear tests and predicts that fallout would cause several deaths and injuries.The test would either be performed above the Nevada desert, where the fallout will only affect U.S. citizens, or next to the Mexican border where, because of wind conditions, it would only affect Mexican citizens.As good democrats, the U.S. Government arranges a referendum in the U.S. where, not 20 See (Cunningham 1987, 51), for the same view.
surprisingly, the vast majority of U.S. citizens votes for the Mexican border alternative.According to the territorial state principle, this would be democratically impeccable decision relative to the boundary problem.
The above example is of course just fictional 21 but it is easy to find analogous real cases.One example is the nuclear plant Barsebäck on the south coast of Sweden just across from Copenhagen.The choice of location for Barsebäck was under the democratic control of the Swedish people through their elected representatives in the parliament whereas the Danes didn't have hardly any democratic influence over the choice. 22The same holds for the regulations regarding the maintenance of the plant and the eventual decision to close it down.
Yet, the Danes would, arguably, have had to carry most of the burden had a serious accident happened. 23Given the territorial state principle, this is no problem from the perspective of democracy, a view few Danes would agree with, I surmise.
As these examples show, the main flaw of the territorial state criterion is that it is completely insensitive to who is affected by a decision.This will not suffice in a world where pollution, goods, and capital move more or less freely over state borders.
The three other boundary principles mentioned above which Whelan discusses ---nationality, culture and geography ---share the same flaws as the territorial state principle.Since many of these problems are rather obvious, it is hard to believe that anyone has seriously suggested them as a boundary principle for a democracy as a normative ideal and Whelan himself doesn't state clearly what he has in mind.However, Whelan's discussion of this topic becomes more interesting if we see it as a criticism of certain rules of thumbs for who should take part in which decision, that is, as partial boundary principles for democracy as a decision method.We can then see his criticism as an effort to point out when 21 When this paper was written.Given the recent political development in US foreign politics toward Mexico, one might fear that the example will not stay fictional forever, or at least that some similar cases will appear. 22Admittedly, one can argue that they had some influence since the Danes had the opportunity to lobby the Swedes by taking advantage of Swedish freedom of speech laws and the like.This highlights the important question for democracy of how to regulate and facilitate lobbying.I'm grateful to Shlomi Segall for pressing this issue. 23Within a zone of 40 kilometres from Barsebäck, all of Copenhagen is covered (and also the Swedish cities Helsingborg, Landskrona, Lund, and Malmö).
these rules are useful and when they are not.Moreover, I don't doubt Whelan's claim that these principles are commonly discussed among political scientists, geographers, and others, but it is quite likely that what these theorists had in mind was a useful rule of thumb for certain cases, not a general principle applicable in all cases.
There is, however, a problem for Whelan if his discussion is supposed to be about democracy as a decision method: his main charge against democracy loses its force.If we take democracy as a practical decision method and again raise the question of who should take part in which decision procedures, it seems clear that this has to be answered by the normative ideal that motivated the choice of democracy in the first place, in conjunction with empirical considerations. 24  Again, if our normative ideal is utilitarianism, then the allotment of voting rights and the scope of democratic decision-making should be devised such that the total welfare is maximised, and likewise, mutatis mutandis, for other possible ideals.In other words, it is a misplaced criticism of democracy as a decision method that it "cannot itself provide any solution to disputes that mayand historically do -arise concerning boundaries" since it never was supposed to do it by itself.(Whelan 1983, 40).Consequently, for Whelan's challenge to have any bite, we have to take it to be about democracy as a normative ideal.

Democracy as a Normative Ideal and the All Affected Principle
An intuitively attractive boundary criterion for a democracy as a normative ideal is the All Affected Principle: The people that are relevantly affected by a decision ought to have, in some sense and to varying degrees depending on how much they are affected by it, influence over the decision. 25I think it is fair to say that it is implicit in much reasoning in the democratic tradition and many (most?) the contemporary democratic theorists who explicitly discuss the boundary problem endorse some version of this principle: "Everyone who is affected by the 24 Cf.(Dahl 1989), (Barry 1991).
25 I'm here assuming that the All Affected Principle will answer both the question of whom should have influence and how much influence they should have.One could take the principle in a less expansive way to just answer the former question and leave it to another principle to answer the latter one.As we shall see below, however, much criticism of the All Affected Principle assumes the more expansive version.Thanks to Krister Bykvist for pressing this issue.decisions of a government should have the right to participate in that government" (Dahl); "In a perfect democracy all who are thus affected [by a decision] play some part" (Cohen); "[A]ll affected interests should have a say" (Goodin); "Power in any decision-making process should be proportional to individual stakes" (Brighouse & Fleurbaey). 26  It is easy to garner intuitive support for All Affected Principle.We don't think that the curriculum imposed by the School board of Waco, Texas, is any business of Icelanders since they are not relevantly affected by this decision.
Likewise, people in Luleå (far up north in Sweden) should not, in most cases, have much of a say on how the public transportation is organized in Stockholm, e.g., whether to increase the number of buses to a certain suburb.However, what kind of hair spray the teachers use in Waco might the business of Icelanders too, i.e., if the hair spray used destroys the ozone-layer.Similarly, whether state tax revenue should be used to subsidise the public transportation system in Stockholm is arguably an issue that the people in Luleå, qua taxpayers, should have some form of influence over.
According to the All Affected Principle, how much power you ought to have over an issue depends on how much your interests are at stake.In actual democratic practices we approximate this standard by having different issues handled on different levels: councils, provinces, regions, states, European, and so forth.The subsidiarity principle, frequently invoked in the discussion of decision making in the European Union, is in one of its popular interpretations ---"decisions should be taken as closely as possible to the citizen" ---very much along the lines of All Affected Principle.The general prescription of this principle is that an issue should be handled by the democratically run body that represents Principle as "all those people who are affected by a particular law, policy, or decision ought to have a voice in making it" (Whelan 1983, 16).Ian Shapiro suggests that "[e]veryone affected by the operation of a particular domain of civil society should be presumed to have a say in its governance"; Lars Bergström claims that "the all-affected principle … says that every individual who is affected by a given decision should have a vote"; and Torbjörn Tännsjö renders the All Affected Principle as "[e]veryone who is affected by a decision should be allowed to take part in it" (Shapiro 1996, 232); (Bergström 2009, 1); (Tännsjö 2007, 5).See also (Cunningham 1994, 147) and (Cunningham 1987, 25-26).My emphasis in all the quotes.the social union that best approximates the set of relevantly affected people relative to the type of issue.Just as in the discussion of welfare and equality, we need to develop a measure or index of what should count as being relevantly affected by a decision by consulting our considered judgements about which effects on people's lives are of such significance that they should have a say in a decision, and how much influence they should be assigned.The idea is that one is relevantly affected by a decision if and only if one is made better or worse off relative to the measure by the different possible outcomes of the choice situation, that is, the difference between the values of the outcomes for the individual given the index.

One reason why many people agree with the
Such a theory would in many respects be similar to the theories of welfare that have been suggested in the discussion of utilitarianism and to the theories regarding "the currency of egalitarian justice" suggested in the discussion of how a just society would look like. 27One might also think that one could just import a currency from these areas, such as Rawls' "primary goods" or Sen's "capabilities", as an explication of "relevantly affected".This is suggested by Brighouse and Fleurbaey and an advantage with this approach is that it might bring democratic decision making more in line with what is good from the perspective of justice and morality. 28However, our judgment about when people are affected by a decision in such a way that they should have some influence over it may be different in many respects from our judgment about when people's well-being is 27 See e.g., (Rawls 1971); (J.Cohen 1989), (Dworkin 1981b, 1981c, 2002); (Sen 1985(Sen , 1992)). 28(Brighouse and Fleurbaey 2010, 15).Roughly, if people vote in accordance with what is good for them from the perspective of the metric of social justice, then the winning alternative will also be the one that maximises social justice.
affected, or about the relevant goods for the state to distribute in an egalitarian fashion.The example of "nosy preferences" is a case in point: Even if I am so disgusted by the lewd literature that you read, or by your choice of bedroom activities, that my well-being is seriously at stake, it still seems that I shouldn't have any power over you in regards to such activities.Rather, you (and your partner if one is needed) should have all the power to decide such issues. 29Similar considerations are likely to appear for other currencies developed for a different context, I surmise.
A quite popular suggestion, usually presented as an alternative to the All Affected Principle, is that those who are legally bound by the laws should have the right to take part in making the laws. 30This might very well be a better exegesis of the common phrase "government by the governed" or, as Lincoln once expressed it, "A government of the people by the same people". 31It might also be more in line with how we historically have thought about democratic governance.
The scope of the "Legally Bound" or "All Subjected Principle" is quite unclear, however.A person who spends a fortnight in South Africa every year is arguably legally bound by the laws of South Africa, at least during the time she is in the country.Does that mean that she should have some kind of influence on the South African elections according to the All Subjected Principle?
It is sometimes suggested that the All Subjected Principle will keep voting rights and other democratic influence roughly along the lines of current democratic practices, or at least extend it less widely and counterintuitively as compared to the All Affected Principle. 32As the above example indicates, this is not clear but depends on how we spell out "legally bound".Actually, on a natural reading of what it means to be bound by or subjected to a law, the All Subjected Principle entails that we should include everyone.On this reading, you're bound by a law if you are liable to prosecution were you to violate the law.For example, I'm bound by the law in Sweden to wear seat belts whenever traveling in a car 29 See e.g., (Sen 1970);(Dworkin 1981a(Dworkin , 2000)). 30See (Miller 2009); (Beckman 2006(Beckman , 2009(Beckman , 2014)); (Tännsjö 1992); (Owen 2012); (López-Guerra 2005);and (Dahl 1989).For an extensive discussion of the All Subjected Principle, see (Goodin 2016). 31Lincoln in Message to Congress, 1861, quoted from (Naess, Christophersen, andKvalø 1956, 285).
even if I actually never go by car since if I were to go by car and not wear a seat belt, I would be liable to prosecution.Hence, all of us, irrespective of where we live, are bound by the laws in South Africa since were we to violate them, by going to South Africa and doing something against the law, we are indeed liable to prosecution.
Moreover, the circle of people subject to legal duties doesn't always correspond to the territorial jurisdiction of the state, as illustrated where the law includes provisions of "universal jurisdiction". 33In addition, there is a distinction to be drawn between being subject to legal duties and being subject to coercive institutions enforcing the law.These don't always coincide as is illustrated by cases where people are beyond the reach of public authorities and yet subject to the law. 34  The point here is that the All Subjected Principle also needs an explication of relevantly affected although in terms of relevantly legally affected. 35Rather than taking the All Subjected Principle as an alternative to the All Affected Principle, I suggest that it is more fruitful to see it as a version of the latter but with a specific currency, namely being relevantly legally affected (to different degrees).This is analogous to the different version of other normative principles.
Take, for example, Utilitarianism.The same formal principle ("An action is right if and only if it maximises welfare") can be combined with different conceptions of welfare to yield different versions of Utilitarianism: Hedonistic Utilitarianism, Preference Utilitarianism, etc.And just as there are different versions of hedonism, which yield even more versions of Utilitarianism, there will be many different versions of legally affected, yielding different versions of the All Subjected Principle.
Most importantly, the right currency of the All Affected Principle might in the final analysis turn out to be a quite complicated combination of different 33 See (Goodin 2016) for an extensive discussion of this issue. 34A recent suggestion is that the All Subjected Principle should be interpreted as including both requirements; i.e., a person is subjected in the relevant sense if and only if the person is both subject to legal duties and coercive institutions (Beckman 2014). 35Such an explication would also consider whether people can me more or less legally affected by different laws.
aspects including being affected in certain legal and non-legal ways (e.g., affected wellbeing).This important possibility is obscured if the take the All Subjected Principle as an alternative competitor to the All Affected Principle. 36  Another fundamental question for the All Affected Principle is when can we say that a person has had correct influence over a decision?How to analyse "influence over a decision" is a tricky question which needs its own essay.A starting point, however, could be to analyse it in terms of whether an individual's preferences ordering could determine the collective ordering in some possible situations, what I have called "potential influence" and which the standard measures of voting power measures.You have potential influence on a decision if there is a possible situation (i.e., a possible set of individual preference orderings or voting patterns of the involved people) where you are decisive, that is, where your preference or vote will determine the outcome. 37In addition, we should also consider an individual's probable and actual influence.An individual's probable influence in a situation is the probability of her being decisive whereas an individual's actual influence given a number of issues is the number of times she is decisive divided with the number of decided issues (Arrhenius 2008a(Arrhenius , 2018)).
Such an analysis has to be supplemented with an analysis of an individual's influence on other peoples' preferences and beliefs, and her influence on the agenda.For example, a person may have great influence on a decision by just being the kind of person that many people trust (e.g., an expert or a charismatic leader), or by having influence over what issues that are discussed in the mass media, or by having control over which alternatives that are on the voting agenda. 38  We would then have to consider what kind and degree of influence that should be given to an individual depending on how she is relevantly affected.This 36 There might be interpretations of "legally bound" which would be somewhat counterintuitive to subsume under "relevantly affected", for example being bound by an unenforced or even unenforceable law (I'm grateful to Bob Goodin for suggesting this possibility).However, my guess is that any reasonable explication of "relevantly legally affected" would rule out such "effects" as irrelevant for giving people a say in a decision. 37Measures of potential influence were first proposed by (Penrose 1946), (Shapley and Shubik 1954), and (Banzhaf 1965, 1966, 1968) (see (Felsenthal and Machover 1998) for an overview).See also (Danielsson 1974), (Goldman 1974), (Morriss 1987).I discuss these kinds of measures further in my (2008,2018) papers. 38See (Arrhenius 2008a(Arrhenius , 2018)).Research on the increasing influence of lobbyists and so-called "policy professionals" in Western democracies are highly relevant here.See e.g., (Svallfors 2016;Svallfors and Tyllström 2018).
decision in the sense that they may present their case, send flowers and poems, etc., as a corollary of some form of freedom of speech.And the woman in Nozick's case is of course also free to just ignore such courting and can demand and demand that the suitors stop with their efforts, and so forth.
In other words, to think that Nozick's example is a decisive argument against the All Affected Principle is to make a triple mistake.Firstly, the All Affected Principle is flexible since it can be coupled with different notions of "relevantly affected", and, secondly, "having a say" ("having influence") need not to be equated with voting rights, and, thirdly, one can give people different degrees and kinds of influence relative to how they are affected by a decision.I would say the same thing about the alleged counterexamples presented by Lars Bergström that evidently is meant to show that the All Affected Principle "is not very plausible" (Bergström 2006, 7).For example, he writes that this can be seen if one tries to apply it to individual agents.Everyone makes a lot of decisions that affect the interests of other people.But hardly anyone would maintain that all these decisions should instead be taken collectively by all the people that are affected by them.This is not only impossible in practice.It is not even a desirable ideal that one should try to realize as far as possible.It does not seem to leave much room for personal decisionsexcept perhaps for very trivial ones.It would be a threat to individual freedom (Bergström 2006, 7-8).
Again, since individual freedom is quite a basic interest, a proponent of the All Affected Principle can happily agree that the individual should have most of the influence in many decision and even be a dictator about certain decisions, for example such decisions that concerns her human rights.

Whelan's Objection
What is then the Whelan's original problem with the All Affected Principle?He worries that it "would require a different constituency of voters or participants for every decision" (Whelan 1983, 19).Similarly, Dahl writes that the "logic of the [all affected principle] (…) is that for every different set of persons affected there be a different association or decision-making unit" (Dahl 1970, 64).In other words, the All Affected Principle demands what is practically impossible.This is surely true about the All Affected Principle taken as part of a practical decision method but misses the target if we take it as part of a normative ideal.As with utilitarianism, the All Affected Principle might not be possible to use as an everyday decision method but it might still be correct as part of a democratic ideal.As such, it is part of an ideal that we use to evaluate the practical procedures that we implement in the real world in respect of how well they approximate the ideal.We will never be able to create a perfect democratic system but that is not an argument against trying to approximate it.To take an analogue example: It would be silly to criticise a criterion of "tall person" on the grounds that we cannot in practice measure length exactly.Although there is always going to be borderline cases, there are clear examples of procedures that are better and worse according to the All Affected Principle, as the examples used in this paper illustrates.
Whelan raises another problem that at first sight looks more damning: The deeper problem is that before a democratic decision could be made on a particular issue (by those affected), a prior decision would have to be made, in each case, as to who is affected and therefore entitled to vote on the substantive issue… And how is this decision, which will be determinative of the ensuing substantive decision, to be made?It too should presumably be made democratically ---that is, by those affected ---but now we encounter a regression from which no procedural escape is possible.(…) Thus to say that those who will be affected by a given decision are the ones who should participate in making it is to attempt to bypass the crucial question, and to propose what is logical as well as a procedural impossibility (Whelan 1983, 19 last emphasis mine).
These are harsh words but Whelan's reasoning begs the question.Why should we determine who is relevantly affected by certain decision by a prior democratic decision? 40Why shouldn't it, as I suggested above, be determined by a theory of the currency of relevantly affected and an analysis of the consequences of different courses of action, policies, and institutional structures on people's interests?
When we try to figure out which theory of democracy is the best one, and which conception of relevant effects is correct, we have to weigh the evidence for and against different theoriesit is an epistemic question, not a practical political 40 One might think that there is an obvious end to Whelan's regress: when everybody is included.This is, however, not true, since there is nothing that guarantees an expansion of the number of involved people for each step of the regress (I'm grateful to Wlodek Rabinowicz for pressing this issue).However, one might consider having a vote on what counts as being "relevantly affected" since this is bound to be an issue that reasonable people will disagree over.We could then use this notion of relevantly affected for any future decision.Intuitively, this is a vote that should involve everyone.Of course, this is impossible in practice but can still be an ideal that we should try to approximate, quite similar to the hypothetical consensus used in the contractarian tradition in political philosophy (e.g., Hobbes, Locke, Kant, Rousseau, and Rawls).
one.I don't think any democratic theorist has thought that whether or not their theory is correct depends on people's opinions about it.Compare again with utilitarianism according to which an act or policy is right if it maximises people's well-being.No utilitarian, however, has ever made the absurd suggestion that what constitutes well-being should be determined by the utilitarian principle. 41

Another Worry
Here's another but related objection to the All Affected Principle that I at first found more troubling.Assume that you're the pivotal voter for the choice between two alternatives A and B, and that the choice will affect my relevant interests.It follows from the All Affected Principle that I should have some kind of influence over the choice, which is fine.However, your choice how to vote will also affect my interest, since your vote will determine whether A or B is chosen.
Hence, it looks like I should have an influence on how you vote, according to the All Affected Principle, which might seem counterintuitive.Moreover, should we then vote on how you should vote in the first vote?Again, the outcome of that vote might also affect me.Thus, should we also vote on how you should vote in the second vote?We seem to get into an infinite regress.
A first answer to this objection is that it can be avoided by a reasonable theory of "relevantly affected".Arguably, it is one of my fundamental interests, like freedom of speech, to be decisive when it comes to which alternative I vote for.Thus, I should be a decisive on such issues.Moreover, it seems reasonably in a democratic society that people should have the right to influence other people's voting behaviour by arguments and discussion, and that they in that sense should have some influence over how other people vote.Whether there will be a regress or not depends on which theory of "relevantly affected" the All Affected Principle is combined with, and there seems to be promising candidates that stops the regress in the first step.Of course, the devil is in the detail and to properly show this we need to work out our theory of relevantly affected. 41It might be that Whelan himself believes in a normative theory according to which a principle can only be justifiably implemented in a constitution via some kind of democratic decision.This is an untenable idea since it would imply that all decisions to implement democracy in undemocratic countries would be unjustified.
More importantly, I think there is only a spurious infinite regress here.
Consider a case with two outcome A and B in which three people, p1, p2, and p3 are involved: A: (p1, 1), (p2, 1), (p3, 0) B: (p1, 0), (p2, 0), (p3, 1) For each person, there is equally as much as stake in the choice between A and B, namely one unit (we are assuming that the numbers in the case is measuring relevantly affectedness on at least an interpersonally comparable ordinal scale).It follows, according to the All Affected Principle, that each person should have the same influence over the decision which could be secured by giving every person one vote and using the majority principle.Assume that, unsurprisingly, p2 will vote for A and p3 for B. Then p1 is the pivotal voter for the choice between A and B and the choice will affect p2's and p3's relevant interests since they get either 0 or 1 unit depending on p1's choice.
Assume that it then follows from the All Affected Principle that p2 and p3 should have influence over p1's vote in proportion to how they are affected and that there is a vote on how p1 should vote in the first choice.What is then at stake in this second vote?Well, either it will result in p1 voting for A and then we get outcome (p1, 1), (p2, 1), (p3, 0), that is, A. Alternatively, it will result in p1 voting for B and then we get outcome (p1, 0), (p2, 0), (p3, 1), that is, B. Likewise for the other possible voting patterns in this case.Thus, if we give people proportional influence over other people's votes, then we just reproduce the original choice situation and the distribution of influence will be the same.So no infinite regress.

Miller's Objection
David Miller argues that the All Affected Principle has a problem of circularity since to know who to include in a decision according to principle, we already need to know which outcome the decision will produce: [I]f we take "affected interests" in its simplest sense to mean being made significantly better or worse off by the policies a demos adopts, then for any question that has to be decided the domain is likely to vary with the outcome, creating a problem of circularity.That is, if group X has to make a decision between two policies, one of which is detrimental to group Y, but the other of which has no impact on that group, then to know whether the members of group Y should be included in the demos alongside the members of group X, we have already to know which outcome will occur.(Miller 2009, 215) This argument, however, is based on an uncharitable analysis of what proponents of the All Affected Principle mean with being affected by a decision.
You are affected, I and other have suggested, only if something is at stake for you, that is, if the two outcomes are of different value for you, given the currency of the All Affected Principle.Consider the following representation of Miller's case above: Policy 1: (X, 10), (Y, -10) Policy 2: (X, 5), (Y, 0) One can say that the choice of policy 2 have no impact on group Y if one, like Miller, "assumes that a person's interests are affected by a decision when it makes him or her better or worse off by comparison to the status quo ante".But that is not what the proponents of the All Affected Principle have had in mind (at least not me).As we discussed above, the idea is that one is relevantly affected by a decision if and only if one is made better or worse off depending on the possible outcomes of the choice situation, that is, the difference between the values of the outcomes for you given the currency of the All Affected Principle.Group Y is affected in the above case since 10 units are at stake for them in the choice between policy 1 and 2. So the people affected doesn't vary with the outcome that actually comes about but only with the choice situation.
Miller seem to suggest, however, that with this approach, we face another devastating problem: If to avoid that problem we say that everyone who may possibly be affected by a decision should be included in the constituency that makes it, then the demos will expand in all directions, depending on which possibilities are contemplated in the deliberation leading up to the decision.The upshot is that if we adopt the affected interests principle, we can only be sure of avoiding the circularity problem by making democracy's domain universal, that is to say by including every (…) human being in the demos.(Miller 2009, 215).Cf. (Goodin 2007, 52-53).This is, however, a false dichotomy.There is a third alternative, namely that we include those people who interest are actually at stake in a certain decision, like the Y-and X-people in the case above.We should not include those people whose interests might possibly be at stake, or would have been at stake if we were in another choice situation.Alternatively, we can take probabilities into account but then we should of course adjust people influence to match the probability that their interests are at stake.This means that for most decisions, the vast majority of people will have no or close to no influence at all.In both cases, it doesn't follow that everybody should be included.

Summary
I hope to have showed that the seemingly intractable theoretical problems raised by the democratic boundary problem are solvable and that the All Affected Principle is still very much a live candidate as a boundary principle for democracy as a partial normative ideal.Moreover, I hope to have contributed to a fruitful way of approaching the boundary problem so that future work can move ahead on the important question of developing a currency for the All Affected Principle and what notion of influence with which the principle should be coupled.
To forestall some possible objections, however, let me point out that I haven't given a full-fledged defence of the All Affected Principle and of democracy as a normative ideal but have focused on clarifying how to understand these two ideas and the boundary problem, and a rebuttal of the criticism that has been directed towards the principle.We might, in the end, reach the conclusion that democracy is not an attractive normative ideal as compared to other such ideals further inquiry has to determine this.However, the reason will not be that the All Affected principle "propose [s] what is logical as well as a procedural impossibility" or is "a threat to individual freedom" or that it makes "democracy's domain universal", etc., but that another normative ideal is more in reflective equilibrium with our considered judgements.Recall also that democracy need not be our only ideal, we might have other ideals that the democratic ideal has to be weighed against in reaching a final theory of how a just or good society should look like.
Secondly, the All Affected Principle fits well with democratic ideals that concern distribution of influence or power.It might not fit other conceptions very well, however.The epistemic democrat is a case in point.According to the epistemic conception of democracy, what makes democracy valuable is that it generates better decisions, that is, it is a better "truth-tracker" than alternative decision-making procedures. 42The All Affected Principle doesn't fit well with this conception of democracy.Rather, one could argue that inclusions and exclusions should be based on whether it is likely that the voter in question is better than chance at getting the right answer. 43If there is good reason to believe that the average voter will be better than chance at getting the correct answer, then we have a prima facie argument for including as many people as possible in a decision making process.If not, the argument works the other way around, that is, as a reason for excluding people from the decision.In other words, the criterion of inclusion and exclusions for the epistemic democrat has to do with what will generate the most accurate decisions, not with who is relevantly affected by the decision.
I find it more plausible to take epistemic democracy as practical decision method that is justified by an appeal to some normative ideal, such as the utilitarian value of reaching the right decisions in many contexts.Some might not agree, however.If so, then we have a normative democratic ideal for which the All Affected Principle is not a suitable boundary criterion but for which there is another promising candidate: competence at getting the answers right. 44 26(Dahl 1970, 64);Cohen (1971 p. 8); (Goodin 2007, 50); (Brighouse and Fleurbaey 2010, 2).To the best of my knowledge, the first formulations of the All Affected Principle are by Robert Dahl and Carl Cohen.It has been formulated in different ways, both by its advocates and by those opposing it.Frederick G. Whelan, in his influential paper on the boundary problem, defines the All Affected All Affected Principle is, of course, that it is quite imprecisely formulated and thus open to many interpretations and precisfications.As others and I have stated it, it doesn't say anything about what amounts to being relevantly affected, or what it means to have influence over a decision, and to what degree one should have influence given that one is affected in a certain way.To forestall some possible misunderstandings, let me just hint at what I think an analysis of these concepts would and would not look like.