Jump to ContentJump to Main Navigation
The Liberal ArchipelagoA Theory of Diversity and Freedom$

Chandran Kukathas

Print publication date: 2003

Print ISBN-13: 9780199257546

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/019925754X.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 19 January 2019

Freedom of Association and Liberty of Conscience

Freedom of Association and Liberty of Conscience

(p.74) 3 Freedom of Association and Liberty of Conscience
The Liberal Archipelago

Chandran Kukathas (Contributor Webpage)

Oxford University Press

Abstract and Keywords

This chapter discusses the principles of a free society. It advances six proposals on the principles of a free society, and discusses a theory of free association. It argues that free association lies at the heart of the idea of a free society, and that the protection it offers to liberty of conscience is important.

Keywords:   free society, freedom of association, culture, liberty of conscience

  • Compel them to come in, that my House may be full.

Luke XIV. 23

‘Tis pleasant enough, and very glorious to the Christian Name, to compare the Griefs of the Orthodox, and their Complaints against the Pagan and Arian Persecutions, with the Apologys for persecuting the Donatists. When one reflects on all this impartially, he'l find it amount to this rare Principle; I have the Truth on my side, therefore my violences are good Works: Such a one is in an Error, therefore his Violences are criminal.

(Pierre Bayle1)

The human world is marked by such a diversity of cultures, traditions, and ways of life—of moralities—as to prompt one to wonder whether there is anything constant in human nature or human moral experience. The previous chapter argued that there was a constant human nature, and that what was most important for human beings was right conduct. While understandings of right conduct vary, human beings are, in the end, profoundly attached to the idea of acting according to the dictates of their understanding. Another way of putting this is to say that they are governed by conscience.

The question is, how, or by what principles, should human society be ordered given that acting in accordance with conscience is prized, and that the world is marked by diversity? The answer offered here, in a nutshell, is that these are the principles of a free society. This raises its own question about what are the principles of a free society. That, broadly, is the question addressed here. The answer offered in this chapter goes to the centre of the theory advanced in this book. That theory can be summarized in the following propositions.

  1. 1. A free society is an open society and, therefore, the principles which describe its nature must be principles which admit the variability of human (p.75) arrangements rather than fix or establish or uphold a determinate set of institutions within a closed order.

  2. 2. Such principles should take as given only the existence of individuals and their propensity to associate; they need not and should not assume the salience of any particular individuals or of any particular historical associations.

  3. 3. Granted this, the fundamental principle describing a free society is the principle of freedom of association. A first corollary of this is the principle of freedom of dissocation. A second corollary is the principle of mutual toleration of associations. Indeed, a society is free to the extent that it is prepared to tolerate in its midst associations which differ or dissent from its standards or practices.

  4. 4. An implication of these principles is that political society is also no more than one among other associations; its basis is the willingness of its members to continue to associate under the terms which define it. While it is an ‘association of associations’, it is not the only such association; it does not subsume all other associations.

  5. 5. The principles of a free society describe not a hierarchy of superior and subordinate authorities but an archipelago of competing and overlapping jurisdictions.

  6. 6. A free society is sustained to the extent that laws honour these principles and authorities operate within such laws.

The thought these abstract propositions seek to capture can be expressed in a more commonsensical way. In a world in which people live in groups or communities or traditions governed by different beliefs, and in which they are often strongly committed to their own moral beliefs in particular, a problem arises to the extent that these groups or communities or traditions find themselves at odds with one another. Given that this could lead to a conflict which might be destructive of (some of) the particular groups, or even of society more generally, how should society be ordered? One answer is that it ought to be ordered according to the right moral principles: the principles of justice. A second (which is really a variant of the first) is that it ought to be ordered according to the principles of justice, which are themselves principles which protect each particular group's interest in surviving (perhaps because this is considered important for the well‐being of its members). A third answer is that society should be ordered not according to principles of justice but according to norms of mutual tolerance or civility, under which people accept that different groups or communities live by different moral beliefs, but also recognize that no group has the right to compel anyone to become, or to remain, a member. It is the third answer this work defends: the good society—or at least, one that is feasible for human beings taken as they are—is not a stable social unity governed (p.76) by a shared conception of justice but a regime of toleration in which disparate and conflicting standards of morality and justice co‐exist. Such a society would be one in which individuals were free to associate with one another to live by the moral standards they can in good conscience accept, and thus also free to refuse to live among those whose moral standards they cannot abide. A good society is thus one in which individuals are free to associate with, and dissociate from whomever they wish, since dissent from the views of the majority or the powerful is tolerated, and conformity is not compelled.

One important conviction underlying the view presented here is that it would be a profound mistake to think that the primary question political philosophy ought to pursue in addressing the problem of group conflict is the question of justice. If we accept that acting rightly is fundamentally important to human beings, that this concern to act rightly should be respected, and that humans differ on the question of what right conduct, including justice, requires, the task of a political philosophy is to explain how we should deal with the fact of different or conflicting understandings of right conduct. What cannot count as a good answer to this question, however, is one which presents a particular theory of justice as a moral conception which should command universal assent (or around which a consensus could emerge); for, ex hypothesi, it is the absence of consensus on moral fundamentals which is the problem. Indeed, an answer of this kind will, in the end, compel many people to live by standards they cannot accept, and, so, fail to respect the desire to live rightly.2

The good society, then, is a free society; and a free society is one that upholds freedom of association—which is important, ultimately, because it upholds liberty of conscience. This is a society in which difference and dissent are tolerated.

Several things, however, are in need of further explanation: what freedom of association amounts to; how it protects liberty of conscience; and why this makes for a free society. Indeed, why are freedom of association and liberty of conscience so important? Such questions are of particular salience in the light of the fact that many contemporary liberal theorists, Will Kymlicka most prominent among them, have argued that the critical value the right political principles should uphold is ‘choice’. Freedom of association and liberty of conscience, on this view, are important values, but subordinate nonetheless to choice, which is of fundamental importance because it reflects a proper appreciation of the (p.77) crucial value of autonomy. A part of the answer to this contention was offered in the previous chapter; but the centrality of freedom of association and liberty of conscience needs more careful exposition. A fuller answer is also required to meet the other aspect of Kymlicka's challenge, which comes in the form of a theory explaining why protecting autonomy requires not just upholding freedom of association but also granting special rights to particular cultural groups, and guaranteeing certain universal rights to persons within all such groups. For an important part of the view put here in this chapter, and in this work more generally, is that cultural groups or communities are not entitled to special rights or protections—save those that derive from the individual's freedom to associate with others.

To this end, the rest of this chapter offers a theory of free association and of the free society. The first task, however, is to explain what is a society, and what is the place of cultural groups within it. Once this is accomplished, the chapter will elaborate the idea of free association, showing why it lies at the heart of the idea of a free society, and why the protection it offers to liberty of conscience is of especial importance.

Society and Culture

What, then, is a society? One answer is that a society is a form of association made up of people who belong to different communities or associations which are geographically contiguous. The boundaries of a society are not always easy to specify, since the contiguity of societies makes it hard to say why one society has been left and another entered. Nonetheless, distinctions or boundaries can be drawn. Since all societies are governed by law, the move from one legal jurisdiction is, to some extent, a move from one society to another. This understanding has to be qualified, however, by the recognition that law is not always confined by geographical boundaries. For one thing, people moving from one region to another may still find themselves subject to laws whose long arms reach even into other countries. Tourists, businessmen, and ‘visiting scholars’ remain subject to the laws of their home countries—especially to their tax laws. In the Middle Ages, the merchant law established codes of conduct and mechanisms of dispute resolution which bound traders who wandered across Europe—almost wherever they might be. And for another thing, an important dimension of law deals precisely with the fact that people cross boundaries into different legal jurisdictions all the time; much of law is inter‐jurisdictional.

Yet this fact itself may help to get us a little closer to an account of what is a society. For a society surely exists when there is some established set of customs or conventions or legal arrangements specifying how the laws apply to (p.78) persons whether they stay put or move from one jurisdiction to another within the greater realm. On this understanding, there was not (as much of) a society among the different highland peoples of New Guinea in the nineteenth century since they lived in legal isolation from one another, even if they were aware of one another's existence. There was, however, a society in Medieval Spain, in which Jews, Muslims, and Christians coexisted under elaborate legal arrangements specifying the rights and obligations individuals had within their own religious communities, and as outsiders within the others.

It may be unwise to seek any greater precision than this in accounting for what is a society. For the moment at least, then, a society will be taken to be a region of contiguous jurisdictions related by law. Societies can be distinguished from one another by jurisdictional separateness. This in itself may be a matter of degree, since some borders or boundaries are more porous than others. One particularly clear way in which societies may be distinguished is by their political separation. Thus we might talk of America and Mexico, or France and Germany as different societies. Yet the distinction cannot be drawn equally sharply, since France and Germany belong to the European Union, whose laws permitting the free movement of people across borders have lessened (though not ended) the significance of the political borders in distinguishing the two societies. The United States and Mexico are, perhaps, more clearly distinguishable as separate societies—although the North America Free Trade Agreement may, eventually, have a profound impact on the nature of this separateness.

This account of the notion of a society is not an especially comprehensive one. In particular, it says nothing about the cultural dimension of society. This, clearly, would add some important complications to the picture. For one thing, many political borders cut through regions in which peoples immediately on either side of the (new) boundaries have more in common with their neighbours than with their countrymen. The Kurds may feel more in common with each other than with their fellow Iraqis or with their fellow Turks. And along the much‐shifted borders of France and Germany live peoples who once saw themselves not as members of French or German society but as peoples of particular local regions.3 But two points of especial importance should be recognized. First, a society is not coextensive with a culture; nor, a fortiori, is a political society or a state. Second, a society is not necessarily a particularly stable or enduring association, since its boundaries shift, and membership waxes and wanes to some degree. Indeed, all human associations, including cultural groups, are not fixed but highly mutable things which change with economic, legal, and political circumstances. What laws prevail in any society (p.79) may depend to some degree on cultural norms; but what cultural groups exist is no less dependent upon the laws, the rights people have, and the tendency of the political and economic climate to favour some arrangements or values rather than others.

Even ethnicity is not fixed but variable: ethnic identity is not static but changes with the environment.4 The importance of this point cannot be too strongly emphasized. Those who have insisted on ‘the “naturalness” of ethnicity’, and criticize scholarship which starts ‘from the premiss that nations and nationalism are peculiarly modern phenomena, and that there is nothing “natural” or inborn about national loyalties and characteristics’,5 are quite mistaken. There is an ‘interactive quality’ to the variables related to group identity: culture, boundaries, conflict, and the policy outcomes of conflict,6 and ethnic identity has a contextual character, since group boundaries ‘tend to shift with the political context’.7

Many ethnic groups were, in fact, products of subgroup amalgamation in the colonial period in Asia and Africa. The Malays in Malaysia, for example, emerged as a ‘distinct’ group only after colonizers created specific territories out of loose clusters of villages and regions; and much the same can be said for the Ibo in Nigeria, and the Moro in the Philippines. On the other hand, some groups emerged not as the result of amalgamation but disaggregation in response to political change. In the former Indian state of Madras cleavages within the Telugu population held little significance until a Telugu‐speaking state was created—at which point Telugu subgroups quickly appeared as political entities. Indeed many groups ‘were “artificial” creations of colonial authorities and missionaries, who catalyzed the slow merger of related peoples into coherent ethnic entities. They did this by the way they categorized those they encountered and by the incentives they established to consummate the amalgamation.’8 Of course, it was not only colonialism which shaped these identities. The Malays, for instance, despite the fact that their numbers were drawn from island peoples as far away as Sumatra, Sulawesi, Borneo, and Java, as well as Malaya, developed their highly cohesive identity partly because of the appearance of Chinese immigrants.9 But the important point remains: group formation is the product of environmental influences; and among these environmental factors are political institutions.

This is not to say that culture is unimportant; but it is not fundamental, even for the constitution of group identity. Legal rights can themselves be important determinants. In the late 1960s in Assam, Bengali Muslims found it advantageous to declare Assamese to be their language in part to become eligible for land reserved for indigenes.10 ‘Culture is important in the making of (p.80) ethnic groups, but it is more important for providing post facto content to group identity than it is for providing some ineluctable prerequisite for an identity to come into being.’11

This fact ought to be emphasized in part because the idea that society and culture are mutable, and indeed, highly unstable, stands at odds with Kymlicka's view, which divides the world more definitely into ‘societal cultures’. As was noted earlier in the Introduction to this work, Kymlicka's societal cultures are cultures that tend to exist in territorial concentrations, based on a shared language, and which provide their members with meaningful ways of life. They tend, he says, to be national cultures (and nations are almost always societal cultures); and cultures which are not societal cultures are not likely to survive given the pressure within each country towards the creation of a single common culture. In this view, immigrants invariably lack a societal culture, while national minorities (such as indigenous peoples) and nation‐states typically do possess one.

The distinction which lies at the core of Kymlicka's theory is the distinction between national minorities and (immigrant) ethnic groups, and any assessment of that theory's plausibility must consider whether this distinction is workable, or capable of bearing the considerable weight that is placed upon it. Despite its intuitive plausibility, however, it is not going to be up to the task. In the end, it masks rather than illuminates the complexity and fluidity of cultural diversity in the modern world, and offers an unduly rigid, static set of categories through which to assess the various claims and concerns of cultural communities, and of the individuals who comprise them.

At its simplest, Kymlicka's distinction supplies a contrast between ethnic groups who are voluntary immigrants in a polyethnic society, and national minorities who are involuntarily incorporated communities in a multinational society. Yet, as Kymlicka himself recognizes, matters are not always clear cut; and the fact that they are not is of greater importance than he concedes. Using the basic categories implicit in his theory we can identify at least four different kinds of groups or categories of people, which are distinguished in the matrix below (Figure 3.1). First, there are those who are voluntary members of minority ethnic groups. Many immigrants come into this category, since they acquire their minority status as a result of a deliberate decision to move to a new society in which they will belong to, or be identified as a part of, a smaller ethnic group. Not all migrants, however, come into this category. Many migrants—possibility the majority of them—are not voluntary migrants or voluntarily members of a minority (though much, here, turns on what is understood by ‘voluntary’). Some migrants are obviously not voluntary settlers. The convicts who settled Australia in the eighteenth century (p.81)

                   Freedom of Association and Liberty of Conscience

Ethnic group

National minority


Adult economic migrants

Some adult Aborginals




Migrant children

Remote Aboriginals


Aboriginal children

Persecuted migrants

Individual Equality

Individual Inequality

Intergroup Equality



Complete Equality

Equality of Inegalitarian


Intergroup Inequality



Inequality of

General Inequality

Egalitarian Groups

Figure 3.1

came involuntarily, as did the Africans who were brought to America as slaves. Similarly, refugees are involuntary migrants who become ethnic minorities not because they wish to acquire that status but because they are fleeing war zones or trying to escape persecution. More controversially perhaps, some members of migrant families are not voluntary migrants: children (almost invariably) and spouses, usually wives, (often) migrate because they have no choice but to accompany the decision‐maker. And, of course, if we ask how many migrants emigrated reluctantly because driven to do so by economic necessity, a good deal more would be classified as involuntary settlers. Some migrants are involuntarily members of ethnic minorities simply because they are regarded as foreigners: Turkish guestworkers, Fijian Indians, and Jews in Europe for much of their history. Had they known how they, or their children, were going to be treated they might not have emigrated in the first place.

Yet while many migrants are involuntary migrants—and involuntarily members of ethnic minorities—not all ‘national minorities’ are involuntarily in their position. Some indigenous peoples are members of national minorities by choice. In some cases this is because they can exit their communities at low cost and low risk to live as (cosmopolitan) members of the wider society. This is true for many (though by no means all) people of mixed descent. It is also true of many members of national minorities who have become urbanized, and whose identities have been shaped by a greater variety of influences than those of their more remote (and less assimilated) fellows. Similarly, many Quebecois are in a position to live either as French Canadians or as Canadians, but choose voluntarily to hold on to a treasured heritage. (Though in some cases, they are people who have chosen to acquire this heritage by immigrating to French Canada; indeed, as Kymlicka points out (MC 23), Quebec actively seeks francophone immigrants.)

More interestingly, there are many groups who are in the position of the Ngarrindjeri people of South Australia, who are forced to make a choice between adopting traditional ways and assimilating into mainstream society. In the case of the Ngarrindjeri, all surviving members of the people are of (p.82) mixed (i.e. European and Aboriginal) descent, and more is understood of the group's traditions by white anthropologists than by the members. The task for the Ngarrindjeri who want to live by their culture is to learn it, and to discard the traditions by which they have largely been raised: those of Christianity and Australian capitalism.

For some, of course, membership of national minorities is not a matter of choice. Many who are raised within their particular cultural communities will not find it so easy to leave, and to cease being members of particular national minorities. Children, most obviously, are involuntary members. But so too are those adults who have lived in communities which have been remote from the life of the wider society, or who have learnt to live by traditions which leave them ill‐equipped for life elsewhere. Like the Hutterites of North America (who qualify in Kymlicka's terms as an immigrant ethnic group), the Orang Asli of Malaysia, and some Aboriginal people in Australia are national minorities who are in this position. They cannot easily take up the cosmopolitan alternative—though, as we have seen, many people can, do, and, sometimes, must.

Already, then, it should be clear that national minorities and ethnic groups are not easily distinguishable—particularly if voluntariness of membership in the community or wider society is the yardstick. The effluxion of time increases the difficulty, as groups mingle, grow, or contract; and as migrant generations find their ancestry more in their land of birth than in the home‐lands of their grandparents. The ‘Indians’ of Fiji and the West Indies cannot return to their ancestral homeland, which is now largely foreign to them. Malaysians of Chinese and Indian descent cannot return to China or India; nor have the Nyonya people of Malacca anywhere else to go. And those of mixed descent, such as the offspring of Malays and their sixteenth‐century Portuguese colonizers, are, if not classifiable as ‘national minorities’, simply immigrants from nowhere.

In many cases the differences within national minorities may turn out to be greater than the differences between some members of national minorities and those of immigrant ethnic minorities. Urban Aborigines in Australia can find themselves with little in common with Aborigines in remote rural areas—even though there may be much they share with immigrant cultural minorities who are neither fully assimilated into, nor entirely independent of, the mainstream society. Equally importantly, national minorities (and, for that matter, ethnic groups) may turn out to be united less by cultural similarity than by political imperatives which create particular groups. In Australia, for example, Aboriginal interests are addressed as if there were a single, homogeneous, Aboriginal society; and Aborigines have constituted themselves as a minority group with a common interest. Yet this Aboriginal identity masks not only the important cultural differences among the various Aboriginal societies, but also (p.83) the conflicts among them. Aborigines have a common cause, but not a common culture.12

The more general point to which all this leads is that group identity is a political (because a legal and institutional) construct rather than simply a cultural one—when it is cultural at all. Ethnic groups tend to shed some of their cultural peculiarities in urban environments, where ethnic identity—which is often expanded to make them more competitive politically—is perpetuated through common residence and common political interests.13 But even when membership is not confined to urban centres, groups may redefine themselves, or be constructed anew, because of political interests held in common.14

All this makes it difficult to distinguish national minorities and ethnic groups, since many national minorities are internally diverse and turn out to be political alliances rather than cultural communities—and often alliances shaped by elites whose perceptions differ significantly from those of the masses. It is also difficult to distinguish them by appealing to Kymlicka's notion of a ‘societal culture’—which national minorities enjoy and ethnic group lack. The Chinese of Malaysia, for example, have much more of a ‘societal culture’ than the Ngarrindjeri people of South Australia. The Chinese here have a culture which is embodied in schools,15 in their print and broadcast media, and in their economic organizations. The Ngarrindjeri, by contrast, have shared memories and values, but little deep understanding of their cultural heritage and traditions, and no institutional embodiment of them. As an immigrant people, the Chinese must be regarded as an ethnic group, while the ‘indigenous’ Ngarrindjeri are a national minority. Yet, if possession of a societal culture is the measure, it is the immigrant Chinese who are the national minority.

Even if a rough distinction may be drawn between ‘national peoples’ and ‘minority peoples’, as is suggested by Ted Robert Gurr—to whom Kymlicka appeals [MC25, 201n.18]—suggests, that distinction has its limitations. Aside from the fact that group identification is political more than cultural, the distinction is a rough one also because members of minority peoples shift strategies and (p.84) change objectives depending on opportunities and circumstances. As Gurr points out, although national peoples generally seek separation or autonomy from the states that rule them, while minority peoples seek greater access or control, sometimes minority peoples who are denied equal access and protection shift strategies and try to exit (as did Soviet Jews); just as, at other times, national peoples may decide to seize power at the centre rather than seek autonomy or secession.16 In Australia, Aborigines (a national minority) have used a number of strategies to pursue a range of different goals, varying from autonomy and self‐determination to ‘polyethnic rights’ involving special entitlements and exemptions. And the same is true of (predominantly Chinese) Singapore, before it eventually seceded from Malaysia in 1965.

Society, in the end, is a kind of union of associations. (Tempting though it may be to call it an ‘association of associations’, this would not be correct, for reasons which will be presented below.) But it is not a union of stable or immutable associations; and indeed, its own boundaries are unstable. People come and go; they change their habits; they re‐form into new associations; and they re‐define themselves under the influence of changing circumstances. Of course, it would not do to overstate this. It is not the case that all societies are so unstable and subject to transformation that we cannot recognize something constant that endures from one moment (or epoch) to the next. Yet it is important to recognize the fact of social mutability. For one thing, international political boundaries shift constantly, and many others are seriously contested. And for another, the internal character of many political societies has changed dramatically even when boundaries have remained stable. Australia went from a predominantly British descendant society to a highly multicultural one in less than two generations; and the complexion of many European societies has changed in similarly striking fashion. Migrations, both within and between states (and both forced and unforced), becoming increasingly common also suggests that this pattern of continuous social transformation is unlikely to cease.

Society is a variable thing. And for this reason it will not do to begin reflection on the nature of the good society by assuming that it is not, or to minimize the extent to which it is mutable. So, for example, it will not do to begin with assumptions of the sort John Rawls makes about the nature of the well‐ordered society: assumptions about its stable character and about its being closed off from other influences.17 Rawls suggests, in A Theory of Justice, that a well‐ordered society is a ‘social union of social unions’.18 In such a society, he argues, ‘the successful carrying out of just institutions is the shared final end of all the members of society, and these institutional forms are prized as good in themselves’.19 It is a union which, while capable of accommodating (p.85) a diversity of social unions, with varying and even conflicting ends, nonetheless evinces a kind of unity which, stemming ultimately from everyone's possession of an effective sense of justice, reveals a common moral consciousness. In such a society, Rawls says, ‘Each citizen wants everyone (including himself) to act from principles to which all would agree in an initial situation of equality. This desire is regulative, as the condition of finality on moral principles requires; and when everyone acts justly, all find satisfaction in the very same thing.’20

On the view advanced in this chapter, and this work more generally, however, this is not how one should view society. It is not, in the end, something which can, or need, exhibit the level of moral cohesion Rawls's image suggests. Indeed, one should be wary of the idea that this vision of the good society as one which coheres around a single understanding of social justice is one that should guide our reflections on the good society and on the most desirable political institutions. For what is characteristic of society, and even more so of modern societies, is that people disagree about questions of morality, and thus also about questions of social justice. In short, we need a different starting point, and an altogether different theory.

The Starting Point for a Theory of the Good Society

The appropriate starting point for a theory of the good society (or, for that matter, for any general political theory) cannot be the interests or the good of some particular society, or group or community or association within society. Nor can it be an assumption that the good society is one marked by fundamental agreement on principles of social justice. The fact of the mutability of human societies tells against the first assumption, which privileges existing social formations against others that might be formed in the future. The divided and unstable character of societies tells against the second assumption, which aspires to eliminate the problem—disagreement—that is an ineradicable feature of the human condition. The nature of these difficulties needs to be considered more closely so that a more appropriate starting point might be advanced.

First, then, why is it not appropriate to develop a theory of the good society by beginning with or appealing to the interests of existing societies or groups? In fact, those interests exist, or take their particular shape, only because of certain historical circumstances or because particular political institutions prevail and not because they are a part of some natural order. There is no more reason to see particular interests as fixed than there is to see particular political arrangements as immutable. Political theory should take as its starting point the existence of a plurality of interests—often competing, if (p.86) not in actual conflict—and ask how or by what principle a political order might adjudicate between or accommodate competing claims. But recognizing that many interests, cultural or otherwise, might have well‐founded claims, political theory should look at the problem of divining political rules from a standpoint which owes its allegiance to no particular interest—past, current, or prospective.

For this reason, the appropriate starting point is the perspective of the individual rather than the group or culture or community. Such collectives matter only because they are essential for the well‐being of the individual. If the condition of the community or the culture made no difference to the life of any individual then the condition of the collective would not matter.21 None of this implies that there is such a thing as ‘the individual’ in the abstract. Individuals do not exist in the abstract any more than interests do. But interests matter only because individuals, and individual lives, do. Thus while groups or cultures or communities may have a character or nature which is not reducible to the nature of the individuals who inhabit them, their moral claims have weight only to the extent that this bears on the lives of actual individuals, now or in the future.22 While the interests given expression within groups, cultural communities, or other such collectives, do matter, they matter ultimately only to the extent that they affect actual individuals.

So groups or communities have no special moral primacy in virtue of some natural priority. They are mutable historical formations—associations of individuals—whose claims are open to ethical evaluation. And any ethical evaluation must, ultimately, consider how actual individuals have been or might be affected, rather than the interests of the group in the abstract. It is not acceptable to evaluate or choose political institutions, or to establish legal rights on the basis of the claims or interests of cultural communities because those very institutions or rights will profoundly affect the kinds of cultural communities individuals decide to perpetuate or to form. Groups may generate entitlements; but entitlements can also create groups. Historical priority does not confer upon a community the right to continued existence.

There is one obvious objection, however, that needs to be considered. If institutions or legal rights are to be established, why not choose conservatively and protect existing cultural communities? Granted that the choice of laws and institutions can indeed alter the composition of groups, is there not a case for establishing rights which recognize—and protect—the actual societies or cultural communities upon which individuals depend? After all, the breakdown or (p.87) disintegration of such communities, bringing social dislocation and anomie, is scarcely a good—for group or individual. So there appears to be good reason to recognize the right of groups and societies to guard themselves against the intrusions of the outside world and to determine their own destiny.

Yet this case is not as straightforward as it appears, for reasons which have much to do with the mutable nature of cultural communities. And indeed, this brings us to the second untenable starting assumption in theories of the good society: the assumption of general agreement on ethical principles, and on social justice in particular. In reality, it is important to note, not only does the composition of society change over time but also most societies, and even groups within them, are not homogeneous at any given moment. Within all communities there are often important differences and conflicts of interest. Internal divisions can take two forms: divisions between subgroups within the larger community; and divisions between elites and masses, which may have quite different interests.

Differences of interest between subgroups might be observed, for example, in the experiences of groups such as the Yoruba of Nigeria, the Lozi of Zambia, and the Bakongo of Zaire, Angola, and Congo (Brazzaville). In each of these cases, the group was formed in response to internal differentiation among subgroups, many of whom fought each other. It was only in opposition to colonialism that their leaders sought to minimize subgroup cultural differences, standardize language, and take other measures to assimilate the many interests into a united association with political strength. While many of these movements of assimilation met with great success, subgroup identities have remained, and in some cases subgroup conflict persists.23

The more important conflict of interest within societies and groups, however, is that between the masses and elites. This is particularly starkly revealed within ethnic cultural communities confronted by ‘modernization’. Under these circumstances elites have ‘distinctive interests that relate to modernity: good jobs, urban amenities, access to schools, travel, prestige’.24 In some cases there is no doubt that elites use their advantages to further their personal ends, in some cases manipulating ethnic sentiment in pursuit of their career aspirations.25 In others, however, matters are more complex. Aboriginal representatives on the National Aboriginal Consultative Committee established by the Australian Commonwealth government were often suspected by their people of succumbing to ‘white’ patronage, even when they were innocent.26 To some extent this was the product of ignorance: in many cases Aborigines did not understand agreements entered into on their behalf by their ‘representatives’. And it is not always easy for those uninitiated into the ways of (p.88) bureaucracies to understand how difficult it is to avoid being ‘swallowed up’.27 Yet these cases also reveal the real gap that sometimes exists between the interests of the elite and the interests of the mass of group members.

This poses a particular dilemma for cultural minorities seeking self‐determination within the larger society and wishing to preserve their cultural integrity. To be self‐determining in the larger society requires a measure of political power, and this means becoming involved in the political processes of the nation. Elites from minority cultures must invariably mix with the educated elites from other minorities and from the dominant society. But in this process the interests of the minority elite become further removed from those of their cultural community. If their cultural community itself undergoes changes, however, the prospect of preserving cultural integrity diminishes.

The cultural community and its elite may, of course, share a common interest in the symbolic standing of the group as a whole. If both gain from the growth of collective self‐esteem then the masses might welcome the prestige derived from the success of wealthier or higher‐status group members.28 Yet while it may indeed be the case that ‘the distribution of prestige is a real and rational object of conflict’29 among ethnic groups, securing this goal can serve to heighten the divisions within the community. Indeed, it could be argued that often the mass may be more interested in jobs and economic progress while the elites, who already enjoy these material benefits, have a greater interest in symbolic traditionalism.30

The divided nature of society, and of the communities within it, strengthens the case for not beginning with a view that the interests of society are fixed or necessarily cohere. Societies and groups are not undifferentiated wholes but associations of individuals with interests that differ to varying extents. Within societies are to be found minorities; and within such minorities are to be found other, smaller, minorities. To regard the wider group as the bearer of privileged interests is to affirm the existing structures and therefore to favour existing majorities. Minorities within a cultural community, which might over time have formed quite different coalitions with other interests, may find that their interests are to a significant degree subject to control by the larger rights‐bearing (p.89) community. More importantly, it restricts the opportunity of minorities within the group to reshape the cultural community, whether directly or through its interaction with those outside the group. Yet the mutability of human associations, and the diversity of human interests and moral views suggests that we should be more concerned to avoid entrenching majorities or creating permanent minorities.

There are two objections that will undoubtedly be raised here, questioning this starting point for political theory. The first questions the individualism evident in this approach. One of the reasons why liberalism's critics often invoke the idea of community when taking it to task is that they wish to reject what they see as the liberal understanding of persons as separate, and self‐contained, atoms, sharing certain formal rights—pre‐eminent among these, the right to be left alone. Michael Sandel, for example, criticizes Rawls for advancing a political theory giving primacy to justice, and presupposing that justice is a virtue of a society peopled by separate individuals whose goals and values are as independent of society as they are. This typically liberal stance, for Sandel, is mistaken because it does not appreciate the extent to which individual selves—and their desires—are not separate from but constituted by the community into which they are born. The idea of a separate self is not merely unreal; it is incoherent.31 Similarly, Benjamin Barber objects to liberal theory's constructions because they presume that people are little more than consumers, pursuing their private desires, and incapable of valuing society (and politics) as intrinsic goods. Its assumptions about the atomistic character of persons thus lead to conclusions which are insensitive to the values of community, and which seem to affirm little other than the pursuit of self‐interest.32

The flaw in liberalism, according to such views, is to be found not in the ideas it advocates but in its social ontology. The problem stems not from its vision of the way the world should be (though this in itself is troubling), but from its understanding of the way the world is to begin with.33

This argument is presented very forcefully by Iris Young, for whom the kind of starting point taken here reflects a view that rests on a flawed social ontology. ‘Liberal individualism denies difference by positing the self as a solid, self‐sufficient unity, not defined by anything or anyone other than itself. Its formalistic ethic of rights also denies difference by bringing all such separated individuals under a common measure of rights.’34 Indeed, this individualist (p.90) position, Young suggests, is essentially assimilationist in spirit; and its granting of rights offers little protection to minorities, who find themselves oppressed less by the deliberate denial of their rights than by the more subtle denial of the worth of their aspirations in the construction of a dominant cultural and political structure. Upon these ontological foundations nothing secure can be built.

But Young's view must be rejected. If we are to theorize about the good for human beings (or about anything, for that matter), something must be kept constant. The suggestion in this work is that that constant is the individual, as the entity with whose good we must ultimately be concerned. Yet while we must assume at the outset that it is the individual who is the subject of our concern, we cannot assume that the identity of the individual is fixed or stable: that the world is made up of certain fixed kinds of individual, identified by their membership of entities—groups, societies, peoples—which are immutable, natural, or original. The most seductive and dangerous move in politics is that move which asserts identity to be not political but, somehow, natural or original. But identity is not natural, or original, or permanent, or even necessarily particularly enduring. It is fluid, ever‐changing (to varying degrees), and inescapably political. A social ontology which begins with the assumption that the basic units of social reality are groups or communities or societies is fundamentally disabled—or at least disables any political theory which is dependent upon it. It is disabled because it cannot account for the politics of social life, which is the story not only of the interactions among groups but also of the creation of groups out of units more fundamental still.

This is why the individualist standpoint is to be preferred. It is not because individuals are somehow natural or original. It is rather because it recognizes that the identity of groups—no less than the identity of individuals themselves—is the product of the interaction of individuals. Now, to assert this is not to accuse the critics of individualism of reifying collectivities—Young, for one, is quite clear that groups ‘are real not as substances, but as forms of social relations’.35 It is to say that to understand groups as forms of social relations it is necessary to acknowledge that those relations are relations among individuals. This does not require wholly denying that a ‘person's particular sense of history, affinity, and separateness, even the person's mode of reasoning, evaluating, and expressing feeling, are constituted partly by her or his group affinities’.36 It does, however, require denying that groups or communities ‘constitute individuals’.37 Groups affect the nature of individuals; but they do so because in groups individuals affect one another, whether directly through personal contact, or indirectly through the influence of the (p.91) forms—the language, the art, the religion, the physical and moral landscape—they have created or sustained. Whether and how they continue to do so depends very much upon the actions of individuals, both within and beyond the group.

A group, in other words, is an association. It is an association of individuals. And society is a union of such associations. This, however, brings us to the second objection to the individualist standpoint adopted here: it appears to assume, implausibly, that individuals can exist or be imagined as apart from or independent of society. This is an allegation Iris Young presents in rejecting the idea that society should be seen as composed of associations, even though it should, in her view, be understood as composed of groups. In Young's account, a ‘group’ is different not only from an ‘aggregate’ but also from an ‘association’. An association, for her, is something which is constituted by individuals, who come together as already formed persons; and the relationship of persons to associations is usually voluntary.38 The trouble with this model, she argues, is that it relies on a conception of the subject as something of autonomous origin, and of consciousness as something ‘outside of and prior to language and the context of social interaction’.39 It may account for formally organized institutions such as clubs and churches—but not for groups.

But the trouble with this objection is that it depends upon a non sequitur: the claim that to see a group as something constituted by already formed persons is to see it as constituted by a subject of autonomous origin, and existing outside the context of social interaction. Yet there is no need for an individualist to make any such assumption, even while agreeing that an association is indeed constituted of already formed persons. The individualist need not deny that the subject was itself formed by its (earlier) social context. But to understand any present, and current, form of association we need to understand it in terms of the individuals whose interactions make it what it is, and has become.

The critical assumption made by the individualist here is not that the individuals who make up the group‐as‐association are autonomous, or presocial, but rather that they are actors. The social landscape, in the social ontology in which this is a crucial assumption, is made up of actors. To the extent that there are other features on the social landscape they are either aggregates of actors or else epiphenomena. The sum of red‐haired people in society is an aggregate; the class of red‐haired people (should someone assert the existence of such a thing) is epiphenomenal. An association of individuals is not merely an aggregate of persons because the association is understandable only in terms of the relations among the actors who comprise it. A morgueful of musicians is not an association, let alone an orchestra. Of course, individuals are not the only actors in society. Associations can themselves be (p.92) actors. ‘An actor is a locus of decision and action, where the decision is in some sense a consequence of the actor's decisions.’40 But associations can be made up only of actors. What distinguishes an association from an individual is that in an association individual actors are governed by or operate under the rule of some authority. Authority is a relation between agents; and such a relation is only to be found within associations, including associations of associations.

Society is, in the end, made up only of actors. In this regard, there are no groups in society, unless they are associations. For if they are not associations, they cannot act; and if they cannot act, they have no place in any plausible social ontology, except as epiphenomena. There are, in the end, only actors; and this means that there are only individuals, or associations of individuals. Any reference to other kinds of collectivities as actors ‘would be allegorical at best, presenting us with a complex story in the guise of something simpler, and at worst thoroughly misleading’.41 (This does not mean we have to deny the operation of things like crowds or mobs, for example; we have only to maintain that, since they lack loci of decision they are not actors, and that their behaviour has to be accounted for in terms of individual interaction.)

Society, on this understanding, is not itself an association, for it is not an actor. It may be comprised of a multiplicity of associations, but it is not itself one. Unless—and this is an important caveat—it is constituted as one. And it may be constituted as one if a political society is created. To put it more succinctly, society is not an association, but a polity is. Californian society is not an association, but the state of California is. In precivil war America, the southern states were a society, since they amounted to a union of groups and communities living under common laws—some of which sharply distinguished it from the North—but they did not form a single (political) association until they constituted themselves as the Confederacy.42

All this matters for politics because understanding the nature of groups and society should make us less inclined to reify them and also less willing to accord them a standing they do not warrant. It should make us less likely to see them in the ways that many have tried to describe some of them in the past—as entities which are not only purer, or nobler, or more worthy, but also more stable and less divided than they really are.

The most important strength of this starting point in approaching politics, however, is something that its critics see as a weakness. According to Iris Young, this ‘atomistic conception generates a political theory that presumes conflict and competition as characteristic modes of interaction’.43 Whether or not it is ‘atomistic’, liberal political theory certainly does make this presumption. The (p.93) human condition is one of conflicts. There is nothing in history that suggests that it has ever been otherwise; and there is no plausible social theory that explains how it might ever be different in the future. It is from this assumption that political theory should begin.

The Good Society as the Free Society

Once we grant the variability and mutability of the various forms of human association, the diversity of cultures and traditions and moralities, and the strength of people's attachments to their own ways and understandings of right conduct (and disdain for those of others), what kinds of terms should govern the relations of people in political society? The answer which best appreciates the importance of these circumstances is one that asserts that the terms are those of a free society; and that such terms give expression to an idea of freedom of association.

Given what we know of human nature, and of the nature of society, the good society must be one in which no particular social formations are protected or privileged, but human beings are left free to associate: to form, reform, or transform the associations under whose authority they would live. Such a society would be one in which none would be required to live in groups or in associations (under terms) they reject. They would be free to dissociate from those who would live according to terms they could not abide. Such a society would also be one in which each group would tolerate those others whose ways they disagree with or even find (morally) disagreeable. This is a society that tolerates dissenters, however weak they might be; even if it cannot be one which encourages or offers especial support to those who want to go their own way.

But why should freedom of association, and, more particularly, freedom of dissociation, be regarded as so important, or indeed as fundamental to a free society? And what does this freedom of association amount to? Before we can address these questions fully, certain features of the associations that make up society ought to be made clear. First, it should be noted that not all of the associations to be found in society value freedom. A monastery, for example, is notable—if anything—for the restriction placed on its denizens, who are bound by comprehensive codes of conduct. Members of the Unification Church founded by the Reverend Moon, even if not confined to a monastery, are bound by codes that severely restrict their choice of marriage partner, their choice of career, and even their choice of a place—or country—to live. Indeed, they have no choice in these matters. In these, and other cases, freedom is disvalued and forsaken in the pursuit of other, higher, purposes.

Second, some associations, while they may value freedom, do not practise freedom. A People's Liberation Army, for example, may be dedicated to the preservation of ‘the people's’freedom. In some cases, we know they are perhaps (p.94) even too dedicated to that freedom, and to the freedom of other, as yet unliberated, ‘peoples’. Nonetheless, the army itself will not be an association in which there is much freedom. The actions and movements of soldiers are always governed by a strict obedience to authority, which operates along a deliberately delineated chain of command. While this is most clearly visible when the soldier is on duty, it remains the case (to varying extents) even in private life, since non‐combatants too may be transferred at short notice to places where the army has needs to be met. The army may value freedom intensely; but its members do not enjoy much freedom.

A slightly different example of the same phenomenon is the political party. A party may value freedom so highly that it even puts the name freedom or liberty into its self‐description. A ‘Freedom Party’ or a ‘Liberal Party’ always claims to value freedom; almost invariably, it claims to value political freedom. Yet it would be hard to find a Freedom Party which would allow its members political freedom within the party. While the party's cause may be to promote freedom, it will generally expect its members to be loyal not only to its goals but also to its methods; and it will not tolerate dissent. Just like an army, a party needs discipline if it is to win; for that reason it will not, in its behaviour, exemplify or honour the values it seeks to promote.

Third, there are many associations that are indifferent to the question of freedom—or at least, maintain a silence on the issue—but which, nevertheless, place important restrictions on the freedom of members. A league of sports teams, for example, may place important restrictions on members, affecting the rules by which they play, and even whom they may play with. During the years of South Africa's international isolation, cricketers in Australian and English leagues faced expulsion if they wished to play in South Africa. In the United States, in 1994, the National Baseball League was faced with a strike by players who, ultimately, wanted free agency so that they could compete in an open market for the rents to be gained in the professional game.

In all of these cases, individuals who find their liberty restricted suffer these restrictions because they have voluntarily accepted membership. It ought, however, to be noted that this is not always so. In some cases membership might be compulsory: soldiers, for example, are often victims of the draft. In others cases, people may have been born into associations that restrict their liberty: Christians may be born members of that faith and of a particular denomination and church; the Amish and Hutterites are born into these particular and unusual communities. The children of Mafia families take up an offer of membership they could not refuse. And, of course, people are born the subjects of particular authorities or citizens in particular countries, and so are subject to restrictions imposed by governments.

In some of these cases of involuntary membership, the individuals are not unfree to leave the association. While drafted soldiers may not resign (at least, (p.95) not immediately), members of the Amish community may move to the city; Christians may become Buddhists, and disgruntled Mafiosi may enter the Federal Witness Protection Program. And citizens in many countries may emigrate. Needless to say, the opportunity‐cost of switching memberships varies enormously.

In the cases just mentioned, the members belong to associations which do not place any special emphasis on the value of liberty—or at least, not on the liberty of the member within the organization. We should note, however, that some kinds of associations may both promote freedom and allow members a great deal of freedom, yet not be voluntary associations members are free to leave. A school, for example, may hold up the ideal of freedom to teach students its value and importance; and it may grant students a substantial say in the running of the school, as well as the free choice of course of study; but at the same time, schooling may be compulsory, and the students unwilling participants in their freedom. A trade union may enhance its members' opportunities for work and promotion and not interfere with their personal lives, but deny them the freedom to leave the union.44

With these considerations in mind, let us move to the main question: what makes for freedom of association, and why is it important? A broad answer to these questions might be useful here, to present a general idea of the concerns that govern the theory to be advanced. But a more detailed explanation must follow. In general terms, freedom of association exists when individuals are free to leave the group or community or enterprise of which they are a part. This is important because, ultimately, what matters is that people not be required to live in or be a part of ways they think wrong, or to participate in practices which (morally) they cannot abide. People should be free to live as conscience dictates; and not be required to violate conscience. Yet since those members of the community or group from which particular individuals dissent may also find alternative ways or practices unattractive and, in the end, morally unacceptable, it cannot be a solution to the conflict to say that one side must conform to the wishes of the other when no compromise is to be found. The principle of freedom of association recognizes that, in such circumstances, either party should be free to disengage from the other, for neither is entitled to compel the other to remain in association on its terms. And neither can appeal to the rightness of its beliefs or standards to justify compelling the other to remain, since it is the rightness of those standards which is the occasion for the dispute. Not even social justice can provide that standard; for it may equally be in contention among the disputing parties. A free (p.96) society is one which honours freedom of association by allowing those who disagree to dissent.

All of this requires fuller elaboration, for there is a great deal that is implicit in this outlook which needs to be spelt out. A free society, then, is made up of communities or groups which are not necessarily voluntary associations, and whose practices may differ to a considerable degree. These associations have authority over their members, for as long as those members recognize as legitimate the terms of their associations, and the authority that upholds them. All that is necessary as evidence of such recognition is that members elect not to leave. Recognition in these austere terms would, of course, be meaningless without the individual having one important right against the community: the right to leave the community. If there are any fundamental rights, this has to be that right. It is an inalienable right, and one which holds regardless of whether the community recognizes it as such. It would also be the individual's only fundamental right, all other rights being either derivative of this right, or rights granted by the community.45

This view of the rights of the individual gives a great deal of authority to the community or association of which the individual is a member. It imposes no requirement upon these associations to be associations of any particular kind. It does not even require that they be associations that value or honour individual freedom, or themselves tolerate dissent: they may indeed be quite illiberal. But if members of such communities or associations wish to continue to live by the terms of the association, the outside community has no right to intervene to prevent those members acting within their rights.

Yet at the same time, this view does not give the community any fundamental right. The basis of any association's or community's authority is not any right of the group to perpetuation, or even existence, but the acquiescence of its members. Those members have the inalienable right to leave—to renounce membership of—the community. Such communities are not entitled to force individuals to remain, even if the consequence of those individuals leaving might be the destruction of the association. Nor can they claim any right to bind individuals on the basis of kinship or culture. Individuals within any community are free to leave together or in association with others, and to reconstitute the community under modified terms of association. Communities without the broad support or commitment of their members will thus wither; yet communities within which there are only isolated pockets of discontent with its practices might well prevail.

(p.97) The right of exit is, in fact, nothing more or less than the right to repudiate authority. It arises out of what might be called the ‘no‐right’ of any authority to coerce people into becoming or remaining members of a community or association. No authority has the right to prevent anyone from dissociating from the community and seeking to leave it.

This way of viewing matters strikes a particular balance between the claims of the individual and the interests of the community. It recognizes the existence of many kinds of association, including cultural groups, but regards them simply as associations of individuals, drawn together, perhaps, by history and circumstance. While they may have certain acquired interests, these are in no way equivalent to the interests of all their members. The mutability of such communities reflects their nature as associations of individuals with different interests. The interests of the community as a whole and the interests of particular (groups of) individuals within may well conflict. The liberal individualist view outlined here, by regarding the group as having its moral basis in the acquiescence of individuals in its ways, including its cultural norms, rejects the idea that the group as such has any right to self‐preservation or perpetuation. Nonetheless, by seeing the right of association as fundamental, it gives considerable power to the group, denying others the right to intervene in its practices—whether in the name of freedom or any other moral ideal.

What this view does not give groups or communities, however, is recognition as agents with fundamental rights. It does not recognize the group as having any authority over and above that deriving from the acceptance of that authority by the group's members. So, for example, it views the Amish community as having authority over its members only to the extent that it concedes that those members are free to live under the authority of that community if they so wish. But the Amish community has no right to deny members the freedom to leave; nor does it have any right to demand from others recognition of its laws, or support in their enforcement.

On the other hand, neither has an individual any fundamental claim upon others to ensure that he has the capacity to join a particular group or community, to remain within that group, or to leave it. Different associations of people take different views about what can rightly be expected from others, and about what persons and communities are owed. (Although it is important not to assume that a group or community position reflects unanimity, or even majority agreement, since any stance taken by those in authority may be the product of a range of circumstances, including majority apathy, minority power, or complex constitutional rules governing policy formation.) Some will devote considerable resources to recruiting new members; others might concentrate on maintaining incentives for members not to exit; and yet others might consider it important to ensure that all their members are fully equipped to leave the community and enter another if they so choose. This (p.98) diversity of views or practices of the various groups has to be accepted. What does not have to be accepted is the claim of any group or community to deny its members the right to leave.

Now, what would the good society look like if it were indeed one governed by these principles of freedom of association? Given what we know of human beings, their propensity to associate and to dissociate, and the diversity of their moral outlooks, it would be a society of many associations, governed by many authorities. And these associations would be ones whose members were, above all, free to leave. To be sure, this freedom could not be all there is to the good life; but in a world in which people differ over what the good life amounts to this freedom is of crucial importance, for it upholds the idea that people who differ should be able to live in different ways.

Yet if this idea of freedom of association is indeed honoured, the society it would make for would not be one in which every association valued or exemplified freedom. If people were free to form associations of any kind, some (or many) might elect to form associations which were unfree. A free society is therefore not a society of free societies or free associations. A free society is a society of many associations,46 not all of which need be free—indeed, none of which need be free.

One way of imagining what such a society might look like would be by considering a contrast between two kinds of society. The first is the land of Panoptica, which comprises a number of associations that place no restrictions on individuals beyond the minimum necessary to preserve peace and security. In these associations individuals face no impediments to action beyond those intended to prevent them injuring or harming others. Indeed, they may live almost however they wish within their associations; they may dissent from the prevailing norms by expressing their disagreement; but they may not form subcommunities or internal associations which restrict their own freedom; and they may not leave the association. The other society is the land of Mytopia, which comprises a number of monastic communities, each of which places different but extremely severe restrictions on liberty. In this society, individuals in each of these associations may leave their association if they wish (perhaps to set up another one with others of like mind), even if no one has, as yet, considered doing so. Perhaps those who have thought about leaving regard it as too risky or expensive. In fact, the only ones who have left are ones who have been taken away by outsiders and forced to live in Panoptica. Those who remain acquiesce in the ways of their monastic communities, though no one is bound to stay.

(p.99) In the real world, the differences between societies are not always so stark (even if the ethical gulf separating America from North Korea is much greater than the Pacific). But if the difference between Panoptica and Mytopia can be imagined, it is the latter which is the free society, and the society described by the theory offered here, for its inhabitants are not bound to stay. Unlike their cousins in Panoptica, the people of Mytopia are not forced to be free.

The second society is, in fact, much more of a liberal society than the first, for at the core of liberalism (according to the theory advanced in this work) is an appreciation of the sanctity of disagreement. Liberal doctrines—and practices—arose out of a growing awareness that disagreement could not be eliminated, least of all by compelling those who were supposedly in error to change their behaviour (let alone their thinking). The solution to the problem was to accommodate disagreement rather than to abolish it, to some degree by tolerating dissent, and ultimately by allowing those whose dissent was intolerable to depart to live elsewhere by different beliefs. This solution has always had its critics. When the Peace of Westphalia (1648) ended the possibility of an undivided Christendom, the Pope declared the accord reached by the European powers ‘null, void, invalid, iniquitous, unjust, damnable, reprobate, inane, and devoid of meaning for all time’.47 But this is the solution that lies at the heart of the emergence of liberal democracy over the past 350 years.

At the same time, however, it needs to be recognized that modern states are inclined to make the same error as that committed by Innocent X, regarding the idea of a divided polity as damnable, if not altogether devoid of meaning. And political philosophers, even if cognizant of the virtues of a federal constitution, tend to think of the good society as one united by a commitment to norms of social justice. Among liberal political philosophers, some consider agreement on certain fundamentals to be vital; and this is perhaps most evident in the social contract tradition, as interpreted and revived by John Rawls.

Here a distinction might usefully be drawn between two understandings of agreement: agreeing to and agreeing that.48 In the first case, the agreement reached is about what the agreeing parties will do (or refrain from doing). We agree to meet, or to marry, or to visit a marriage‐guidance counsellor, or to see each other in court, or to avoid contesting custody of the children, or to let bygones be bygones. In the second case, the agreement reached is about the truth of a proposition. We agree that meeting would be fun, or that we were made for each other, or that Dr Gatens is the best marriage‐guidance counsellor going, or that meeting was a big mistake, or that the children's interests (p.100) are paramount, or that it's worth giving the whole thing another try. The two kinds of agreement are quite different, and the agreeing parties might agree in one sense but not in another. We both agree to meet but only one of us thinks it a good idea; we both agree to consult Dr Gatens, though only one of us thinks she is sane or that any counsellor is worth consulting; we both agree to give the whole thing another try, but one thinks it is worth trying anything for the sake of the children while the other thinks that there is a relationship worth preserving.

Often it is easier to ‘agree to’ when we ‘agree that’. Other things being equal, we are more likely to agree to a course of action if we agree that these will be the outcomes, or that the probable outcomes are desirable. But this is not necessary. Indeed, in some instances ‘disagreeing that’ is essential for us to ‘agree to’. In market exchange, the parties must disagree about the relative values of the goods forsaken or exchange would be pointless: each must hold the good he is forsaking to be less valuable than the good he is acquiring. Disagreement about value is the basis of agreement to exchange. But even if we leave aside the case of market exchange, it is not necessary to agree on the truth of some proposition to come to an agreement about a course of action. I want ‘Iron‐bar Maley’ dead because he lied to the police and you want him dead because he did not lie, so we agree to kill him. The important point, however, is that ‘agreeing that’ is not necessary for us to ‘agree to’.

One reason why we might agree to certain things is that it is difficult to agree on the truth of some propositions in dispute. Unable to reach agreement on the truth of the matter, we agree, perhaps, to one deferring to the other with known expertise, or to seeking an independent arbitrator, or to letting the matter drop. Or we may even agree to disagree; or to go our separate ways.

Political society, if it is a free society, embodies agreement. Yet this is not agreement on substantive truths about matters of justice, but, rather, agreement to abide by norms which tolerate disagreement. Among human beings, agreement on the truth of matters of great ethical significance—on how one should live, for example—is difficult to secure, even if the number of persons involved is small. In these circumstances, the best solution to the problem created by disagreement about the truth may be to put to one side the issue whose truth is in dispute: to agree to disagree; or to agree to part company. The solution to be sought, in the end, is not epistemic agreement but practical agreement. We seek not a modus credendi but a modus vivendi.

The idea of freedom of association gives expression to the conviction that people differ in their assessments of what is good and true, that such differences cannot always be reconciled, that we ought to try to live with this by tolerating those who hold dissenting views about how to live, and that, if we cannot abide the ways of others we should either dissociate from them or let them dissociate from us. Freedom of association reveals a recognition that we cannot agree to agree; but we can, and should, agree to disagree.

(p.101) Implicit in all this is the assumption that force should not be used to try to secure agreement in either sense of the term. It should not be used to secure epistemic agreement because this would constitute a direct violation of conscience. But it ought not to be used to secure agreement in the second sense either. For even to try to coerce agreement to a compromise would amount to a violation of conscience. A good society is one in which agreement is not compelled; for it is recognized that people disagree, and it is accepted that those who cannot be persuaded to think and behave differently should be tolerated or allowed to go their separate ways.

Mytopia is such a society. Of course, it may not be the most attractive of possible free societies, since its members are in important ways unfree. Here the extreme case is used to illustrate the point that freedom of association is the relevant freedom; and to concede that people may well use their freedom to restrict their freedom—or fail to use their freedom to free themselves. Panoptica, however, is not a good society. While it is not a place without its attractions—not the least of which are the many freedoms its inhabitants enjoy—it remains, in the end, a society whose inhabitants are unwilling participants in their freedom. The people of Panoptica are not free—for they are forbideen—to live as they wish.49

This view of the nature of the good society is quite different from that offered by writers such as Kymlicka. It begins with the relatively innocuous, shared assumption that moral evaluation is individualistic in the sense that what counts, ultimately, is how the lives of actual individuals are affected. ‘It is individual, sentient beings whose lives go better or worse, who suffer or flourish, and so it is their welfare that is the subject matter of morality.’50 But unlike some other liberal views, including Kymlicka's, it does not go on to restrict what counts as (a legitimate form of) human flourishing. It does not suggest that human flourishing requires that the individual be capable of autonomy, or have the capacity to choose his way of life on the basis of critical reflection upon a range of options. Rather, it contends that what matters most when assessing whether a way of life is legitimate is whether the individuals taking part in it are prepared to acquiesce in it.

These premises are somewhat austere. They may be more austere even than those upon which Loren Lomasky chooses to defend his own conception of liberal basic rights: the idea of individuals as project pursuers.51 Lomasky is critical of the idea of grounding liberal rights in the ideal of individual autonomy, and for good reason: with the defence of autonomy often comes a disregard (p.102) for actual practices and ways of life.52 For this reason, he argues that what is most important, and requires recognition, is that individuals are project pursuers. Projects may not be chosen: ‘A person's commitments may be unarticulated and not at all the product of conscious deliberation culminating in a moment of supreme decision. They may rather be something that he has gradually and imperceptibly come to assume over time in much the same way that one takes on distinctive vocal inflections or the cast of one's face.’53 Nonetheless, project pursuit is ‘partial’. ‘To be committed to a long‐term design, to order one's activities in light of it, to judge one's success or failure as a person by reference to its fate: these are inconceivable apart from a frankly partial attachment to one's most cherished ends.’Thus, Lomasky maintains, an ‘individual's projects provide him with a personal—an intimately personal—standard of value to choose his actions by. His central and enduring ends provide him reasons for action that are recognized as his own.’54

But even to take personal project pursuit as fundamental to our nature excludes a part of human practice, since some cultures are not able to accept the idea that individual projects can provide any sort of standard of value. Consider, for example, Maddock's portrait of Australian Aboriginal society.

If we take human culture to be humanly created, then we are forced to the conclusion that there is among Aborigines a profound resistance to crediting themselves with their own cultural achievements. Their plan of life is held to have been laid down during The Dreaming by the powers and occasionally to have been modified since by the intervention of these powers, as when one appears to a man in a dream and communicates a new song or rite. Aborigines claim credit only for fidelity to tradition or, as they put it, for ‘following up The Dreaming’. It is powers alone who are conceived as creative, men being passive recipients of unmotivated gifts. As men deny the creativity which is truly theirs, they can account for their culture only by positing that to create is to be other than human. To be human is to reproduce forms.55

In such a society, it would seem, individuals are not project pursuers; although they might be said to display commitments, they do not regard themselves as possessing personal goals.56 Nonetheless, there may be enough reason to respect that way of life into which they have been inducted, and which is the only life they know.

The theory advanced here recognizes as legitimate communities which do not in their own practices conform to individualist norms, or recognize the (p.103) validity of personal projects. Yet at the same time it is a liberal theory inasmuch as it does not sanction the forcible induction into or imprisoning of any individual in a community. No one can be required to accept a particular way of life. Thus if, as has often happened, some members of a particular culture on making contact with the wider society wished to forsake their old ways, they would be free to do so and the objections of their native community would not be recognized. In this respect minorities within cultural minorities receive some protection. On the other hand, if those members wished not to leave their community but to assert rights recognized by the wider society but not by their particular community or culture, they receive no recognition. What is given recognition first and foremost is individual freedom of association (and dissociation).

Objections to the Exit Principle

There are two major objections that will be levelled against this account of the good society which ought to be considered at this point. The first is that leaving freedom of association to determine the shape and composition of society simply gives power to the dominant group or culture or tradition to determine the form society will take. Minority communities and minority views will dwindle. The second objection is that the principle of freedom of association, with its emphasis on freedom of exit, leaves unprotected those individuals whose communities systematically deny them any liberty, and in the process make it more difficult (if not impossible) for them to exit. Freedom of association, on both these counts, is an empty freedom, which benefits neither individual nor community. These objections should be elaborated and then considered in turn.

There are at least two reasons to think that the principle of freedom of association alone will not give enough protection to groups or communities. The first is that a view of this sort does not recognize group claims to self‐determination; and the second is that, without special entitlements such groups will not be able to take action to protect their identity. Vernon Van Dyke, for example, suggests that self‐determination is vitally important and ‘is essentially the moral right of a group’.57 The initial defence he offers for this view, that ‘the existence of needs implies a right to act (within limits) to meet them, or that a conception of the good has a corresponding implication’,58 is not especially robust since little is said to explain how needs imply rights. But Van Dyke offers two other considerations: first, that sometimes ‘an interest of (p.104) individuals can be best served, or only served, by allocating the related right to a group, and this is the case with self‐determination’.59 In the face of ‘free and open individualistic competition with those who are more advanced’, history shows us, groups such as indigenous peoples are not capable of defending their interests.60 Second, he argues that, as a matter of empirical fact many societies, including international society through such agencies as UNESCO and the UN, have chosen to recognize the right of ‘peoples’ to self‐determination or to preserve their culture.61 This only makes sense in the context of a conception of group or collective rights of self‐determination. Unless we think in terms of group rights, a large part of modern practice cannot be justified.

In response to this a number of things need to be said. First, and most generally, it has to be recognized that there are considerable limits to the extent to which collective self‐determination is possible. Once cultures come into contact with others and trade and other forms of social intercourse develop, it is very difficult for the community to preserve its ways.62 As Richard Mulgan has pointed out, once there is the possibility of the individual leaving the group with impunity the balance changes, and some practices can never be recovered. While it is true to say that many cultural minorities find it difficult to preserve their ways in the midst of an alien culture, there is little reason to think that giving rights to the group will change the fundamental nature of the problem.

Second, even if self‐determination is possible, groups are not always united. Often, as was pointed out earlier in this chapter, communities form and acquire an identity only after cultural invasion (whether by immigrants or colonizers). The desire for collective self‐determination is in many instances forged by elites seeking to unify disparate groups which did not always claim a collective identity. Thus the Maori, for example, did not exist as a single people in the pre‐colonial period, consisting of different subtribes grouped in larger ‘iwi’ or tribes.63 They began to perceive a common identity after European settlement. Yet even so there is considerable ambiguity in the matter of Maori identity because of generations of intermarriage, with many undecided about which group they belong to. With the Maori, as with many other cultures, self‐determination is problematic because there is considerable internal disagreement about the direction it should take.

Third, it ought always to be borne in mind that to give any community the right of self‐determination is never a matter of giving it the power of determining its own destiny alone. ‘To encourage some groups to determine their (p.105) own future may also mean allowing them to determine the future of others.’64 If power is to be devolved in such ways, it ought to be done with great care, partly because it can adversely affect the peace and stability of the larger society, but partly because it will also have serious repercussions within the smaller cultural community which is acquiring the so‐called power of self‐determination. This latter concern is particularly important when some of the self‐proclaimed indigenous minorities are involved. The devolution of political power tends to push conflict away from the major centres and into the subgroups.65 The result is the exacerbation of subethnic divisions. While this may benefit the larger society which will no longer bear the brunt of ethnic confrontation, it may be highly destructive within some, relatively fragile, ethnic communities. The case of the Salish fishing communities provides a striking—but tragic—example of this.66

This brings us to our fourth point: Van Dyke's objection that unless we think in terms of group rights, much of modern practice cannot be justified. Here he has in mind the fact that many polities and international practices give recognition to groups. Lebanon, for example, from the National Pact of 1943 to the civil war that began in 1975–76, had an electoral system which required interethnic coalitions and ‘prevented the crystallization of allegiances around the overarching affiliations of Muslim and Christian’.67 This was accomplished by recognizing and institutionalizing ethnic claims. For example, all major offices were reserved: the presidency for the Maronites, the prime ministership for the Sunnis, the speakership for the Shiites, and so on. On the liberal view, Van Dyke would object, such practices would not be acceptable.

One initial observation must be made: such attempts to regulate and contain ethnic conflict do not always succeed; Lebanon's institutions failed to (p.106) prevent civil war. Nonetheless, there may sometimes be good reason to design political institutions to take into account the ethnic or cultural composition of the society. Yet there is no reason to see this as inconsistent with the theory advanced here. At least since Montesquieu, liberal theory has recognized the importance of the institutions conforming to the nature of the social order. While the guiding principle of respect for individual rights or liberties has to be upheld, the question of what institutional mechanisms are necessary to protect individual rights and provide for the stability of the social order is to some extent an independent one. The choice of electoral systems, for example, will vary according to any of a number of factors, ranging from the structure of the wider political system to the geographical concentration of populations to the country's political history.

One of the factors which has to be considered when constructing—or reconstructing—institutional mechanisms is the power of significant groups within the polity. A political structure which ignored the power of a significant minority could run the risk of minority disaffection developing into separatist demands and breakdown into civil war; here the case of Sri Lanka springs to mind. Equally, a structure which was unmindful of the power of the majority group could bring about similarly unpleasant results. There is no need, however, to appeal to the rights of groups to justify designing mechanisms to cope with, and temper the exercise of, political power.

Indeed, it is often vitally important that group rights play no part in the justification of the mechanisms that uphold the modus vivendi. For such mechanisms to succeed it is crucial that they be sensitive to the changing group composition of the political order, and capable of adjusting the formal powers devolved towards particular groups in accordance with their actual power. (One of the reasons for the collapse of the Lebanese state was that the changing demographic structure produced by differential birth‐rates led to changes in the balance of power which some groups felt were not reflected in their formal standing.68) If groups are recognized as having rights as groups, it is much more difficult to justify mechanisms that vary their political entitlements with their size and influence. Far better then to maintain an emphasis on the rights and liberties of the individual, while conceding that institutions have to be designed with a view to protecting those liberties by accommodating (and guarding against) the vagaries of group power.69

(p.107) Freedom of association protects groups and communities to the extent that those who wish to remain separate from other parts of society, or to break away and form their own associations of like‐minded people, are left undisturbed: free to go their own way. But it does not provide any group with any assurance of success. And it cannot do so without requiring others not only to condone the actions of those who prefer to be separated but also to give them succour. Yet there is no reason why people should be required to support, or help to prevail, ways of life they have no wish to share, and see no reason to value.

The second objection, however, is not to the failure of the principle of freedom of association to protect groups or communities but to its failure to protect individuals. A society governed by this principle would not be a good society or, for that matter, a free one; it would simply be one in which the wealthier or the more powerful would be secure while the more vulnerable are left with little choice but to put up with their lot.

There are a number of reasons why this is thought to be so. First, freedom of association, underpinned by freedom of exit, does not make for a free society because, in itself, it says nothing about the cost of exit. Individuals would be unfree, even with freedom of exit from their associations, if exit were extremely costly—as it often is. For example, it seems unreasonable to say that Amish teenagers are free because they are at liberty to exit the Amish community, leaving behind family, friends, and property. They are not.

The reply to this objection, however, is not to deny that exit can be extremely costly. It is simply to acknowledge that exit may, indeed, be costly; but the individual may still be free to decide whether or not to bear the cost. The magnitude of the cost does not affect the freedom. The individual may decide to forsake the freedoms of the society beyond his community for other goods—for wealth, or security, or love; on the other hand, he may abandon all these goods for other liberties he might enjoy in leaving. In neither case is he unfree if he may decide to take one option or the other. Cost may have a large bearing on the decision taken; but it has no bearing on the individual's freedom to take it. One might see this by considering an imaginary example. I am offered a billion dollars not to leave my position as CEO to become a professor. Exit is thus rendered extremely costly. Yet it would be odd to say that I am no longer free to leave. The cost of leaving has risen; but I am as free as I ever was. This example puts in a dramatic (or melodramatic) way a point that can also be illustrated more prosaically. All costs are, in the end, opportunity costs. There is a sense in which the costs of leaving the Amish community are highest for those who are most powerful in the group for they have the most (p.108) to lose. Indeed, in most societies, those who are most likely to exit are not the well‐to‐do or the powerful but the poor and the powerless. Amish elders do not commonly defect. Nor are the world's refugee camps filled with members of political elites. They have too much to lose to be likely to leave, though they are certainly free to do so.

A different argument, however, suggests that this example does not do the trick because exit may be costly in a different sense. A member of the Hutterite community, for example, may have to bear enormous costs in attempting to leave not only because of what he leaves behind but also because of the difficulties he will face in trying to move into a society for which his own community made him unprepared. He probably will not survive outside his own society since he does not possess the skills to do so. Here, once again, the reply to the objection is to distinguish between cost of exit and freedom of exit. In this case, to be more precise, we need to distinguish between the freedom to exit and the risks of exit. The fact that exit may be risky does not alter the individual's freedom to take the risk.

Consider another example. If I leave my present career to become a professional boxer I run the risk of failure in my new endeavour. This risk rises to the extent that the new path I have chosen is in a field in which competition is severe, and failure common. And it rises again when a well‐known champion is released from prison to reenter the arena. Yet this increase in risk does not diminish my freedom to take the risk—even if it makes it more likely that I decide not to make the attempt.

And, once again, in reality it is those with the least to lose who are most likely to take the risk of leaving their communities. Often immigrants leave their home countries for places where their opportunities are uncertain because their prospects if they remain put are not good. Many who are highly qualified professionals leave their poorer societies to become low‐paid workers in western countries where their qualifications are not recognized. Moves of this sort are always risky, for migrants almost invariably take a sharp drop in their capital stocks when they move. What makes the risk worthwhile is the poverty of their prospects at home and the opportunity offered by the new society. This is not, of course, to deny that some groups face particular difficulties. Hutterites must leave with nothing if they leave their society at all since private property is not recognized. Many countries with exchange controls will not allow their citizens to emigrate with the proceeds of the sale of their private property. Indeed, this may be quite unjust. The principle of freedom of association, with its emphasis on the importance of freedom of exit, will not eliminate these difficulties or injustice, but it does provide something that is of most value to the vulnerable.

A third argument, however, suggests that there is a more fundamental difficulty still in attempting to judge the freedom of a society by the freedom of (p.109) individuals to exit the associations to which they belong. The problem is that, particularly in associations into which people are born, individuals are conditioned (or socialized) into accepting their lot. It makes it very difficult for individuals to contemplate leaving a community when they are raised to believe that its ways are right and others wicked—or even if they are simply raised to prefer a particular (limited) range of options. They are unfree to leave because they have no taste—or hunger—for exit. Indeed, they may be unfree precisely because they have no yearning for freedom. The principle of freedom of association does not protect individuals from being conditioned (or educated) to have preferences which make them little more than accomplices in their subordination to the norms of their societies.

A reply to this objection must concede that preference formation is, generally, endogenous. We learn to prefer what we are used to; and we could learn to prefer unfreedom. Is a person who prefers unfreedom unfree? The answer implicit in the theory offered here is that, in the relevant sense, no: she is not. A person's preferences have no bearing on whether or not he is free.70 People generally have different preferences and different preference orderings; and they have preferences about their preferences. Their freedom is a matter not of what preferences they have but of whether they may act in accordance with them. Freedom matters—or is an issue—precisely because people have different preferences. If everyone wanted to take the same path, the freedom to deviate would be irrelevant. If we are concerned about freedom, then, we should be concerned not with the nature of people's preferences but with whether or not they can act on them.

This entire analysis is still open, however, to further objection. In general, it could be argued, the position defended here is simply illiberal, since it takes so minimalist a view of the right of exit. That right is honoured insofar as the association of which an individual is a part does not prevent that individual from leaving, and insofar as that association is not recognized by other associations (such as the state or other states) as having any right to prevent its members from leaving. And raising the cost of exit does not count as prevention. Yet surely a more nuanced view is possible—and necessary, since only the guarantee of a more robust right of exit would make such a right meaningful, and ward off the charge that the view defended here is illiberal. One author who has put this view is Brian Barry, and it would be worth considering his argument carefully to see whether it damages the position defended here.

(p.110) In Culture and Equality, Brian Barry argues that liberalism requires that ‘groups should have the utmost freedom to handle their affairs in accordance with the wishes of their members’.71 Freedom of association is of fundamental importance, and individuals should be free to associate as they wish, provided they respect the laws that protect those outside their own association. But there are two important qualifications to be made. First, all participants of the association should be sane adults, and, second, their participation should be voluntary. This means that individuals should be free to associate with one another as they please, and even enter relations of submission and domination.72 However, since many forms of association are unchosen, the important question is, can individuals exit from them? Here, Barry argues, it is ‘a legitimate object of public policy to ensure as far as possible that members of associations have real exit options open to them’.73 This means that they must be able to leave without incurring what Barry calls ‘excessive costs’.74

The critical problem here is how to account for ‘excessive costs’ of exit. If costs were excessive, in Barry's view, the state would be justified in intervening in an association, since such costs would undermine the voluntariness of the association. The state, in his view, is not justified in intervening when the costs of exit are ‘intrinsic costs’: costs of the kind that it cannot, in the nature of things, prevent or ameliorate. Nor should it do so when exit costs are ‘associative costs’: costs that come with individuals exercising rights recognized by the state. It should only intervene when exit costs are ‘external costs’: costs that may not legitimately be imposed by the group.75 The problem, however, is that saying that any ‘external’ cost imposed by the group on its members compromises voluntariness would make it impossible for groups to raise the cost of exit even in the slightest degree—as Barry himself recognizes. Indeed, one could easily conclude that any group, any of whose members decided the costs of leaving outweighed the costs of staying, must be guilty of imposing excessive costs on those members—otherwise they surely would leave. Groucho Marx could not voluntarily be a member of a group that would have him as a member, even if he did have a mind to. More than this, as Barry explains it, ‘this solution would “make it conceptually impossible that a group might be (p.111) acting within its rights in imposing certain costs of exit and that at the same time these costs might be so grave as to undermine the voluntariness of membership” ’.76 Yet it is very important, he continues, to ‘leave open the possibility that a case might arise in which the costs imposed were legitimate but were nevertheless such as to make adherence to the group non‐voluntary’.77

So how are we to determine what costs are excessive? This is a problem that Barry never really solves.78 And it is a difficult problem to solve if one wishes to hold on to any serious principle of freedom of association. If the account of an ‘excessive’ cost identifies that cost in terms of its propensity to affect individual decision‐making, it is hard to see what would not count as an excessive cost. If, on the other hand, the account of an ‘excessive’ cost identifies cost in terms of an independent set of desiderata, those desiderata would then do the work of determining what kinds of associations individuals could and could not maintain or form: their willingness to be a part of such associations would become of secondary importance. So, for example, if the independent desiderata included the requirement that individuals be ‘fully informed’ about all alternatives (before, say, deciding to enter or leave an association), this would mean that there would have to be some authoritative standard specifying what counts as ‘fully informed’. And, by definition, there could be no practical dissent from this standard. No group could argue, for example, that the teachings of its religion have provided its members with all the information they could need if the authoritative standard specifies that they need to be informed differently. The danger in going down this path is plain. If choices are regarded as voluntary only if they are sound—because they are well‐informed, or rational, or not likely to lead to other ‘known’ unattractive consequences—then the way is open for all kinds of individual wishes to be denied, and the willingness of persons to follow their preferred paths will count for little, if for anything at all.

This is clearly not the path down which Barry wants to travel, since he is firmly convinced that, in a liberal society, the individual's wishes must count for something, or indeed be of primary importance. It is also the path resisted by this work. But the theory offered here departs from Barry's because it assumes that if the wishes of individuals are what matter most, then they cannot be subordinated to some conception of what sound choices or acceptable wishes would be.

Now, it might be suggested that, even if all this is accepted, there is still a case for ensuring that the costs of exit are not excessive by at least ensuring that those wishing to exit from an association are enabled to exit once they have decided to leave. How they should be enabled would depend on the circumstances, but it would involve relieving them of burdens that the group (p.112) might impose upon them and, so, make them unable to exit. Here again, Barry provides an instructive example. Some groups make a practise of shunning former members, sometimes to the extent that it makes it difficult for ex‐members to attract sufficient clients to run a business. The threat of boycott, he argues, ‘renders involuntary the continued membership in the group of any member whose departure would lead to the imposition of such a cost’.79 While it would be unduly invasive to compel the group to maintain professional relations with renegades, however, it would be quite appropriate to require the group to compensate the victims—those who have exited the group. The reason Barry gives is ‘that the liberal principle underlying freedom not to associate has to be formulated so as to guarantee that its exercise should be costless’.80 In Barry's opinion, provided ‘that the costs are shifted from the victim to the organization instigating the boycott, membership in the group can count as voluntary while at the same time the remaining members of the group can exercise their freedom not to associate with the ex‐member’.81

Several things need to be said about this argument. First, it is not at all clear why freedom of association or dissociation must be costless. Second, and more importantly, Barry's argument that the ‘victim’ has his freedom compromised by being made to bear the costs of dissociation suggests that associations (and their members) from which the ‘victims’ have dissociated do not have their freedom compromised by being made to bear the costs. Yet surely if cost‐bearing reduces freedom, it does so for anyone who does so. And the costs to the group of having to pay compensation to shunned renegades could be quite significant. Apart from the money, the price might be encouraging others to threaten to defect to try to get their own way within the community, which might in turn lead to the disintegration of the group. In short, the price of giving individuals greater opportunity to dissociate voluntarily might be the creation of circumstances under which others are led to dissociate with one another involuntarily. The point of the argument here, it should be emphasized, is not to suggest that dissociation should be discouraged or group integrity promoted. It is simply to point out that there is no case for trying to shift costs one way or the other—not least because there is no way of rendering actions costless.

All this said, however, it should be made clear that the theory advanced here, unlike Barry's, does not place any great weight on choice, and emphasizes not so much the importance of decisions (to associate or dissociate) being voluntary as the value of their not being forced. Many people are a part of associations they have never chosen to enter, but have never chosen to leave. In some cases they are unwilling to leave because they are not willing (p.113) to bear the costs of exit; in other cases they are not so much unwilling to leave as lacking in any will to leave, since life beyond the group or its traditions is something they have never contemplated. For writers like Barry, such people are unfree and a regime that tolerates such arrangements is illiberal. In the account offered here, however, these people are free and the regime is not illiberal.

It may be worth looking at this point through a practical example. Consider the case of Fatima, the wife of a Malay fisherman living in the state of Kelantan on Peninsular Malaysia. She is a Muslim, a mother, and a wife; and her life is very much shaped by these aspects of her identity, and also by her membership of the village community, which reinforces the view—her view—that her life should be governed by her religion and by her duties as wife and mother. She has no desire to live elsewhere or otherwise. If she did wish to live in some other way she probably would have to live elsewhere, since it is unlikely that the village would tolerate—let alone welcome—any deviation. Is Fatima free?

According to the view defended here, Fatima is free if she may leave the community she is associated with—and the life she now leads—if she so prefers. By implication, she is thus free if she does not have any wish to leave—even if she is ignorant of the possibility of leaving or living differently—and simply continues to live her life. A society of villages such as Fatima's is, on the view offered here, a free society—whatever else may be said about it.

But why is Fatima free? And why is hers a free society? We should answer these questions, first, by discounting two possible answers. The answer cannot be that Fatima is free because she is content or happy. The reason is simply that happiness is not freedom and we may exercise our freedom only to find unhappiness. A free society may be a society of unhappy people, even if there is no reason to think it need be one. Nor can the answer be that Fatima is free because she is autonomous—because she has ‘chosen’ the course she has taken. For this is not true: she has not ‘chosen’ it; she has simply not rejected it. She has acquiesced in a life she has been raised to lead; but she has not embraced it. Being free, she may reject it if she so chooses; but she may not choose at all and remain free. Her action—or inaction—may not be the result of any kind of reflection or deliberation or ‘choice’ that suggests autonomy or self‐direction. It may be the result, instead, of whimsy, or caprice, or simply inertia—and still remain free. A free society, according to this account, is not a society of choosers.

Fatima is free because she may live a life she has not rejected and is not forced to live a life she cannot accept. She is, in a sense, free because she enjoys a certain ‘inner freedom’; however, that inner freedom is not autonomy or self‐direction. It is liberty of conscience.

A society is free to the extent that its members do not have to live lives they cannot, in good conscience, accept. If this is so, then a society will not be free (p.114) simply because there are many options open to individuals. It can only be free if individuals can dissociate themselves from options they cannot abide. It is this sense of freedom that the principle freedom of association protects.

Freedom of Association and Liberty of Conscience

A free society is one whose laws uphold freedom of association and liberty of conscience. More than this, however, one cannot be had without the other. To see this more clearly we need first to return to the question of what is liberty of conscience. If we follow the New Testament, we should take conscience to be the light and guide of our moral conduct. It is the source of what is also sometimes called the ‘inner life’ of the individual.82 Liberty of conscience is enjoyed when the individual can indeed live his life under the guidance of conscience (which identifies right and wrong conduct) and is not impeded by others from doing so.83 Usually, we associate liberty of conscience with the idea of religious freedom, since the idea has been pressed most vigorously in defence of religious toleration. Yet one does not have to hold to any religious beliefs to possess a conscience. An atheist may have a conscience, for he may regard certain kinds of action as morally wrong—and performing them as unconscionable. And both atheists and believers alike may object ‘conscientiously’ to actions and viewpoints they regard as morally obnoxious. They might, for example, find membership of the Ku Klux Klan or the Nazi Party morally obnoxious and regard membership of or employment in organizations which support such groups as unconscionable. If the professional association to which they belong decided to send money to such groups, they would probably decide that they could not, in good conscience, continue their own membership in that association. They would probably also regard being forced to continue membership of an association which supported (however weakly) the Ku Klux Klan as a violation of liberty of conscience.

The most troubling issue confronting the notion that conscience is of fundamental importance is the problem of what to do about the erroneous conscience. This issue has exercised Christians from St Augustine to Peter Abelard to Aquinas, and cannot be addressed in all its complexity here.84 Nonetheless, the argument here accepts two conclusions. The first is that an (p.115) erroneous conscience has binding force. Thus, one does wrong in acting against conscience. Second, one should not do violence to other consciences—even if they are in error.85

If one regards conscience as fundamentally important in this way, the only way in which differing consciences can act is to go their separate ways. Persecuting those with different conscientious beliefs (whether for heresy or immorality) is wrong because it fails to respect conscience—by demanding of people that they act against conscience. In this regard, respecting liberty of conscience requires freedom of association, for it requires letting people dissociate from those with whom they could not, in good conscience, associate. Freedom of association has to be understood not as the freedom to enter into association with others (since those others may be unwilling to associate) but as the freedom to dissociate from those one does not wish to be with.

Equally, respecting freedom of association must mean respecting liberty of conscience. It would be hard to imagine someone whose conscience was being compelled not wanting to dissociate from those persecuting him. The only way in which to compel conscience would be by denying it the opportunity of leaving the association of people trying to force it.

A free society, then, is fundamentally one marked by freedom of association. This does not mean that other liberties—of speech, for example—are unimportant; but they are less important because they either derive from or are subordinate to this more fundamental freedom.

Now it might, at this point, be observed, that a great deal of trouble has been taken here to establish what is, in many respects, an uncontroversial point—uncontroversial, at least, in the modern day. Who, after all, is against freedom of association or liberty of conscience? Yet the conclusions reached here may be a little more troubling. If liberty of conscience is so important, it might be argued, why think that freedom of association offers it sufficient protection? Why not recognize that, if it is important it ought to be protected by recognizing other basic rights that ensure that people can act according to conscience. Thus, for example, a respect for conscience suggests that it is necessary to ensure that people have the right to act according to conscience within their groups or communities. Or at the very least, it suggests that it is important that the community not have the right to make the costs and risks of exiting the group prohibitive. Why allow the group gratuitously to increase the cost of exit, simply in order to discourage dissenters from acting on their conscience? This seems to uphold the power of the group rather than protect the basic interest of people in ensuring that the cost of exit is not prohibitive.86

(p.116) The reason no greater protection should be offered at this level, however, is that it is not only the consciences of dissenters that are at stake. The conscientious beliefs of the majority or the dominant also have weight. They may conscientiously believe that what the minority thinks or wants to do is wrong. They may also believe that they have a duty to preserve the integrity of the community. But what is to be done when there are two competing claims of conscience? The answer can only be that they should both be entitled to go their separate ways. The terms upon which separation takes place cannot be settled as a matter of fundamental right because what is fundamentally right is what is in dispute.

To put this point most starkly, honouring freedom of conscience must mean granting people the right to reject freedom of conscience as a value. Such persons cannot be forced to live according to the belief that freedom of conscience is paramount. Such people may, in fact, hold other things to be more important: piety, for example. Respecting conscience requires allowing such persons to go off to live with those who share such an understanding. It cannot require more, or less.

If freedom of association is of fundamental importance for a free society—indeed, if freedom of association is the crucial determinant of a society's freedom—then the existence of a diversity of associations is a likely indicator of the freedom of a society. We would expect to find such diversity—manifested in the variety of groups, communities, and legal jurisdictions—because we know people differ, and tend to hold different moral views. If we find a society in which such variety was not present we would have good reason to suspect that freedom of association was not tolerated and that it was not a free society. A society such as the former Soviet Union provides a dramatic example of an unfree society because in it freedom of association was not upheld, and dissociation from the moral norms established by the state was difficult and dangerous.87 To the extent that this society was composed of different associations they were not free associations because their existence did not indicate any freedom to withdraw from the legal and moral control of the central state. Those associations—from the ballet companies to the independent republics—were divisions of the state. They were not free associations in a free society but organizational units in an unfree one.

The obvious contrast is with a society such as the United States described by Tocqueville in the nineteenth century. Here we had a society marked by the independence and vigour of its different associations. The diversity of moral and religious commitments within society was mirrored in the range of moral (p.117) communities and legal jurisdictions found within it. As Tocqueville saw it, there was no strong central state governing the lives of the diverse associations of the society. Or at least, not yet; for Tocqueville thought he detected, in America's democratic institutions, important—perhaps irresistible—tendencies towards the centralization of administrative power. (And, of course, this is all assuming that we can excuse the inexcusable: the fact of slavery.)

Whether or not Tocqueville's fears about democratic despotism in America were well founded is a matter we can leave to one side. What is important is the thought that freedom of association is inconsistent or incompatible with a legal order which subsumes the associations within it, ordering them in accordance with centrally established moral standards. This is so even if those standards are influenced—democratically or otherwise—by the demands of the different associations within the society, by its monks and its clergymen.88

For this reason, the view defended here is at odds with contemporary liberal theories that see diversity as something which has to be constituted or shaped by society's central legal and political institutions. It is also at odds with arguments, such as those advanced by John Rawls, for seeing freedom of association and liberty of conscience as strengthened when grounded in a theory of social justice.89 And it is at odds with arguments such as those put forward by Ronald Dworkin suggesting that there is something wrong with principles which turn society into a ‘checkerboard’of different jurisdictions.90 The good society—the free society—is not one governed by a single standard of social justice.

The Next Step

The defence of freedom of association as a fundamental principle presents the theoretical core of the view defended in this work. Yet the argument is a long way from complete, for not only do a number of questions remain unanswered but the conception of politics and society implicit in this view remains to be fully described. This is the task of the next three chapters, which attempt to spell out what a society upholding freedom of association would embrace, to (p.118) account for the nature of political authority under such a conception of the good society, and to specify the role and limits of the state. The immediate task, then, is to explain what freedom of association means in a world of diversity. What it means is a regime of toleration; for it is toleration which is the political doctrine that lies at the heart of the idea of the good society defended here. Its nature and limits must now be explored.


(1) Bayle (1708: 146).

(2) This, of course, is what Bayle thought the defenders of righteous persecution ended up offering: an argument which said that we can rightly or justifiably force others to live as we think right because we are right. But this argument does not take us very far philosophically if our opponents do not accept that we are indeed right. The only reason they could have to accept the terms we offer then would be that we are stronger. This reason seems no better than Thrasymachus' argument that justice is the interest of the stronger. But might does not make right. Nor, for that matter, need right license might.

(3) For a study of the ways in which border changes affect culture and ethnic development see Sahlins (1989).

(4) Horowitz (1985: 589).

(5) Smith (1981: 85).

(6) Horowitz (1985: 73); see also ch. 2 passim

(7) Horowitz (1985: 66).

(8) Horowitz (1985: 66–7).

(9) Horowitz (1985: 68). See also Alfred P. Rubin (1974).

(10) Horowitz (1985: 195).

(11) Horowitz (1985: 69).

(12) Here it is important not to overstate the case. Culture is not something around which it is possible to draw clear and exclusive boundaries. Indeed, even very different cultural traditions can share a great deal. I have addressed this issue in greater detail in Kukathas (1994: 1–21). My point here is simply that a group's political unity may not reflect cultural homogeneity.

(13) On this see Uchendu (1982: 266–75, at 263).

(14) The Pueblo Indians are an interesting case in point, being a group whose linguistic and cultural diversity has seen it united only twice in 300 years. After 1680, the Pueblo did not unite again until 1920, when they set out to defeat a Congressional bill threatening to allow non‐Pueblo squatters on Pueblo land. But otherwise, their linguistic diversity (several of their languages being mutually unintelligible), and their differences of social, ceremonial, and political organization, made them, in effect, different peoples. See Dozier (1970).

(15) On this see Kua Kia Soong (1990), which makes clear also the depth and extent of this institution, as well as the political difficulties faced by the Chinese over the century in their attempts to preserve it.

(16) Gurr (1993: 15).

(17) See Rawls (1993).

(18) Rawls (1999a: 462).

(19) Rawls (1999a: 462).

(20) Rawls (1999a: 462).

(21) A more detailed defence of this individualist standpoint may be found in Kukathas and Pettit (1990b: 12–16). For a contrary view see Taylor (1990: 45–63); but see the reply by Goodin (1990: 64–79).

(22) Note that I have not made the stronger assertion that it is only the lives of individuals within the group which can relevantly be taken into account.

(23) On this see Horowitz (1985: 71).

(24) Horowitz (1985: 101).

(25) See Horowitz (1985: 225) and more generally chapter 5.

(26) See Wright (1985: 292–9).

(27) A complaint made by a senior Aboriginal Public Servant, Charles Perkins, responding to Aboriginal complaints about his ineffectiveness. See Bennett (1989: 103).

(28) One Malay leader has in fact defended policies of preferential treatment in these terms, arguing that although the benefits fall disproportionately to the Malay elite, the masses, knowing of Malay group success enjoy a vicarious satisfaction more highly prized than personal material gain. ‘With the existence of the few rich Malays at least the poor can say their fate is not entirely to serve rich non‐Malays. From the point of view of racial ego, and this ego is still strong, the unseemly existence of Malay tycoons is essential.’ Mahathir bin Mohammad (1981: 44). See also Sowell (1990: 48–51).

(29) Horowitz (1985: 226).

(30) I owe this point to Richard Mulgan, who suggests that the urbanized Maori in New Zealand might provide an example of elite interests.

(31) Sandel (1982).

(32) Barber (1984).

(33) But see Taylor (1989: 59–82).

(34) Young (1990: 229). Young is, however, equally critical of the communitarian viewpoint, adding: ‘Proponents of community, on the other hand, deny difference by positing fusion rather than separation as the social ideal. They conceive the social subject as a relation of unity of mutuality composed by identification and symmetry among individuals within a totality. Communitarianism represents an urge to see persons in unity with one another in a shared whole.’

(35) Young (1990: 44).

(36) Young (1990: 45).

(37) Young (1990: 45).

(38) Young (1990: 44).

(39) Young (1990: 45).

(40) Hindess (1987: 110).

(41) Hindess (1987: 111).

(42) Although, of course, not all southern states joined the Confederacy, several agreeing to fight on the Union side under an assurance that slavery would be permitted within their own boundaries.

(43) Young (1990: 228).

(44) There is a critical difference between this example and the previous one: the trade union member in societies or professions where union membership is compulsory can (almost always) exit by leaving his job.

(45) There are also rights which an individual living within a cultural community might have as a member of the wider society. Some of these might be exercised by an individual while living in the cultural community within that society; other rights it might not be open for the individual to take up without leaving his cultural community. For example, an individual might not be free to exercise the right to marry whomsoever she wishes if such a right is recognized by the wider society but not by her religious community—unless she chooses to leave her religion.

(46) It is, of course, possible in principle that a society not be made up of many societies or associations and still be free. What is assumed in the argument here is that the human condition is one of diversity; so the diversity of outlooks is held to give rise, ineluctably, to a multiplicity of associations.

(47) Innocent X in his papal Bull, ‘Zelo domus Dei’, of 26 November 1648; quoted in Davies (1996: 568).

(48) This distinction does not depend upon any clear semantic distinction resulting simply from the use of one preposition or the other. Meaning obviously depends on context.

(49) It might seem that Mytopia can include or subsume Panoptica. This is indeed so, even though Panoptica is a society which forbids its members to exit, since in Mytopia no group or society or association is recognized as having any right to forbid exit.

(50) Kymlicka (1989: 242).

(51) See Lomasky (1987).

(52) See Lomasky (1987: 248–50).

(53) Lomasky (1987: 42).

(54) Lomasky (1987: 27–8).

(55) Maddock (1972: 129).

(56) Here I am not sure how strongly Lomasky wants to insist that projects are what provide personal standards by which individuals make choices. Elsewhere in his book he seems to suggest that any display of commitment may be taken as evidence of the existence of a project. See, for example, p. 45 where he says that all ‘patterns of motivated activity that form the structure of a scrutable life . . . merit recognition as projects’. If so, there may be less reason for disagreement between us.

(57) Van Dyke (1977: 360).

(58) Van Dyke (1977: 360; see also p. 350).

(59) Van Dyke (1982: 26).

(60) Van Dyke (1982: 28–9).

(61) Van Dyke (1982: 27).

(62) This point is discussed by Karl Popper in his account of the development of the ‘open society’ and the breakdown of tribalism. See Popper (1977: 176–7). See also Hayek (1989).

(63) See Mulgan (1989b: 12).

(64) Horowitz (1985: 591).

(65) ‘Where groups are territorially concentrated, devolution may have utility, not because it provides “self‐determination”, but because, once power is devolved, it becomes somewhat more difficult to determine who the self is.’ Horowitz (1985: 617).

(66) The Salish people sought protection for their traditional culture through the legal process, winning special rights enabling them to continue traditional fishing practices. Indeed, the courts chose to recognize the cultural importance of traditional fishing and provided greater fishing opportunities on the basis of treaty rights. But the implementation of the decision exacerbated hitherto unimportant divisions within Salish society between riparian and marine fisherfolk. This made for the disintegration of that cultural community to the extent that it has seen the development of new inequalities, several lawsuits, and even open hostility between tribes. On this see Anderson (1987: 125–42). Anderson's sobering conclusion is that, of the factors explaining the ‘decay of traditional cultural community’ among the Salish, ‘of greatest importance . . . was the preexisting cultural divisions within the tribes which encouraged treaty fishers to take advantage of their expanded fishing rights in distinctly different ways. While equal before the law in terms of treaty rights, the Salish were and are culturally diverse: traditional and capitalistic marine fishers have different values and ways of viewing the world. It is perhaps not surprising that expanded opportunities accorded a culturally diverse group only reinforced the diversity’ (pp. l39–40).

(67) Horowitz (1985: 633).

(68) On this see Entelis (1988: 216–40).

(69) It is perhaps in this way that we should look to deal with some of the problems confronting Aboriginal and Indian communities who have claims based on past injustice. The solution is not to be found in recognizing group rights. Rather it is to be found in specifying the entitlements of particular historical communities who were the victims of past injustice. Once the validity of their claims has been established (and I have said nothing here about how past injustice should be established or about the principles by which it should be treated), we can turn to the question of the institutional mechanisms needed to satisfy legitimate claims and rectify the injustice. If the granting of land rights is deemed an appropriate solution this would open up the issue of the form these rights should take given the interests of the individuals in question and their communities.

(70) This seems to be too strong. It implies, for example, that the character, Alex, in Anthony Burgess's novel, A Clockwork Orange, is not rendered unfree by the conditioning he undergoes to turn him from a Beethoven‐loving sociopath into a more harmless, Beethoven‐averse citizen. My response to this that the new Alex is free, but is no longer the old Alex. The doctors have changed him. (Whether they had the right to do so is, to put it mildly, still doubtful.) A person's preferences do not determine whether a person is free; but they do bear directly upon who he is.

(71) Barry (2000: 148).

(72) Here it is not clear that Barry really means what he says, since he observes that, in cases of voluntary sado‐masochism, he thinks public policy should simply treat individual proclivities with a measure of indulgence, ‘prohibiting the infliction of bodily harm only when it is of a kind liable to lead to permanent injury’. Barry (2000: 148). This would give the makers public policy warrant to outlaw all kinds of dangerous activity, from football to boxing. The ‘sane consulting adult’ defence seems to vanish at this point.

(73) Barry (2000: 149).

(74) Barry (2000: 150).

(75) Barry (2000: 151) for Barry's discussion of these distinctions. Unfortunately, Barry does not make any clearer what it is that makes some costs imposed by the group not ‘legitimate’. I have discussed this omission in Kukathas (2002c: 184–203, at 187).

(76) Kukathas (2002c: 152).

(77) Kukathas (2002c: 152).

(78) I argue this more fully in Kukathas (2002c).

(79) Barry (2000: 153).

(80) Barry (2000: 153).

(81) Barry (2000: 154).

(82) On this see Lecler (1960: 11–31).

(83) Liberty of conscience is important because, if there are any basic human interests, one of these is an interest in living in accordance with the demands of conscience. Among the worst fates that can befall a person is that he be unable to avoid acting against conscience—that he be unable to avoid doing what he thinks wrong.

(84) I treat too quickly, in particular, the vexing question of whether an erring conscience is absolved of sin when the individual acts out of good intention. Abelard scandalized the church by arguing that it is.

(85) These two conclusions are defended by St Paul, I Corinthinans 8: 9–13. See the discussion in Lecler (1960: 16f.). These conclusions are also defended very vigorously by Bayle in his Philosophical Commentary.

(86) This objection was put to me by Kymlicka.

(87) On this see Harry Rigby's account of the way in which elements of ‘civil society’ survived in homes and networks of friends which were too small to be penetrated by the state. Rigby (1991).

(88) ‘It is plain, then, that if the persecution of sects could ever be a necessary evil, it would be so through the fault of the sovereigns handing themselves over to the mercy of monks and all the clergymen, either through lack of understanding or corrupt motives.’ Bayle (1708: 143).

(89) In accounting for how freedom of association and liberty of conscience come to be upheld in his theory, Rawls writes: ‘The characteristic feature of these arguments for liberty of conscience is that they are based solely on a conception of justice.’Rawls (1971: 214). Rawls asserts that this supplies a better foundation for toleration and liberty of conscience generally than does scepticism. This issue is, I suspect, one which deserves (again) further exploration.

(90) See Dworkin (1986).