Cosmopolitanism and Sovereignty Author(S): Thomas W

Home , Cosmopolitanism, Dogma

Cosmopolitanism and Sovereignty Author(s): Thomas W. Pogge Reviewed work(s): Source: Ethics, Vol. 103, No. 1 (Oct., 1992), pp. 48-75 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2381495 . Accessed: 05/03/2013 13:01

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Ethics.

http://www.jstor.org

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Cosmopolitanismand Sovereignty* ThomasW. Pogge

The human future suddenly seems open. This is an inspiration;we can step back and thinkmore freely.Instead of containmentor detente, politicalscientists are discussinggrand pictures:the end of history,or the inevitableproliferation and mutualpacifism of capitalistdemocracies. And politiciansare speaking of a new world order. My inspirationis a littlemore concrete. Afterdeveloping a rough, cosmopolitan specificationof our task to promote moral progress, I offeran idea for gradual global institutionalreform. Dispersing political authority over nested territorialunits would decrease the intensityof the struggle for power and wealth withinand among states,thereby reducing the incidence of war, poverty,and oppression. In such a multilayered scheme, borders could be redrawn more easily to accord with the aspirationsof peoples and communities. INSTITUTIONAL COSMOPOLITANISM BASED ON HUMAN RIGHTS Three elements are shared by all cosmopolitan positions. First,indi- vidualism: the ultimate units of concern are human beings, or persons- rather than, say, familylines, tribes,ethnic, cultural, or religious communities,nations, or states. The lattermay be units of concernonly indirectly, in virtueof theirindividual members or citizens. Second, universality:the statusof ultimateunit of concern attaches to everyliving human being equally2- not merelyto some subset,such as men, aristocrats,Aryans, whites, or Muslims. Third, generality:this

* This essay has benefited from various incisive comments and suggestions by Andreas F0llesdal, Bonnie Kent, Ling Tong, and myfellow participants at the "Ethikon East/WestDialogue Conferenceon the Restructuringof Politicaland Economic Systems," held in Berlin in January 1991, with fundingprovided by the Pew Charitable Trusts. 1. The differencesbetween the notions of a person and a human being are not essential to the present discussion. 2. There is some debate about the extent to which we should give weightto the interestsof futurepersons and also to those of past ones (whose deaths are stillrecent). I leave thisissue aside because itis at rightangles to the debate betweencosmopolitanism and its alternatives. Ethics103 (October 1992): 48-75 C 1992 by The Universityof Chicago. All rightsreserved. 0014-1704/93/0301-0948$0 1.00

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 49 special status has global force. Persons are ultimateunits of concern for everyone-not only for their compatriots,fellow religionists,or such like. Let me separate three cosmopolitan approaches by introducing two distinctions.The firstis that between legal and moral cosmopolitanism. Legal cosmopolitanismis committedto a concrete political ideal of a global order under which all persons have equivalent legal rightsand duties, that is, are fellowcitizens of a universal republic.3 Moral cosmopolitanismholds that all persons stand in certain moral relations to one another: we are required to respect one another's statusas ultimateunits of moral concern-a requirementthat imposes limitsupon our conductand, in particular,upon our effortsto construct institutionalschemes. This view is more abstract,and in this sense weaker than,legal cosmopolitanism:though compatible with the latter, it is also compatible with other patterns of human interaction,for example, witha systemof autonomous statesand even witha plurality of self-containedcommunities. Here I presenta variantof moral cosmopolitanism,though below I also discuss whetherthis position man- dates effortsto move fromour global statusquo in the directionof a more cosmopolitanworld order (in the sense of legal cosmopolitanism). The central idea of moral cosmopolitanismis that every human being has a global statureas an ultimateunit of moral concern. Such moral concern can be fleshed out in countless ways. One may focus on subjective goods and ills (human happiness, desire fulfillment, preferencesatisfaction, or pain avoidance) or on more objectiveones (such as human need fulfillment,capabilities, opportunities,or resources). Also, one might relativizethese measures, for example, by defining the key ill as being worse off than anyone need be, or as fallingbelow the mean-which is equivalentto replacingstraightforward aggregation (sum ranking or averaging) by a version of maximin or equalitarianism,respectively. In order to get to my topic quickly,I do not discuss these mattersbut simplyopt for a variant of moral cosmopolitanism that is formulated in terms of human rights (with straightforwardaggregation).4 In doing so, I capture what most other variantslikewise consider essential. And my furtherreflections can,

3. One recent argument for a world state is advanced in Kai Nielsen, "World Government,Security, and Global Justice,"in Problemsof InternationalJustice, ed. Steven Luper-Foy (Boulder, Colo.: Westview,1988). 4. I have in mind here a rather minimal conception of human rights,one that rules out trulysevere abuses, deprivations,and inequalitieswhile stillbeing compatible witha wide range of political,moral, and religious cultures.The recent development of, and progress within,both governmentaland nongovernmentalinternational or- ganizations supports the hope, I believe, that such a conception might,in our world, become the object of a worldwideoverlapping consensus. Compare Thomas W. Pogge, RealizingRawls (Ithaca, N.Y.: Cornell UniversityPress, 1989), chap. 5.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 50 Ethics October1992 in any case, easily be generalized to other variantsof moral cosmopolitanism. My second distinctionlies within the domain of the moral. It concerns the nature of the moral constraintsto be imposed. An institutionalconception postulates certain fundamental principles of justice. These apply to institutionalschemes and are thus second-orderprin- ciples: standards for assessing the ground rules and practices that regulate human interactions.An interactionalconception, by contrast, postulatescertain fundamentalprinciples of ethics.These principles, like institutionalground rules,are firstorder in thatthey apply directly to the conduct of persons and groups.5 Interactionalcosmopolitanism assigns direct responsibility for the fulfillmentof human rightsto other (individualand collective)agents, whereas institutionalcosmopolitanism assigns such responsibilityto institutionalschemes. On the latterview, the responsibilityof persons is then indirect-a shared responsibilityfor the justice of any practices one supports: one ought not to participatein an unjust institutional scheme (one that violates human rights)without making reasonable effortsto aid its victimsand to promote institutionalreform. Institutionaland interactionalconceptions are again compatible and thus may be combined.6 Here I focus, however,on a variant of institutionalcosmopolitanism while leaving open the question of its supplementationby a variantof interactionalcosmopolitanism. I hope to show that making the institutionalview primaryleads to a much strongerand more plausibleoverall morality. Let us beginby examining how our two approaches would yield differentaccounts of human rightsand human rightsviolations. On the interactionalview, human rightsimpose constraintson conduct,while on the institutionalview theyimpose constraintsupon

5. Interactionalcosmopolitanism has been defendedin numerousworks. A paradigm example is Henry Shue, Basic Rights(Princeton, N.J.: PrincetonUniversity Press, 1980). Luban, another advocate of this position, puts the point as follows: "A human right, then, will be a right whose beneficiariesare all humans and whose obligors are all humans in a position to effectthe right"(David Luban, "JustWar and Human Rights," in InternationalEthics, ed. Charles Beitz et al. [Princeton,N.J.: Princeton University Press, 1985], p. 209). Robert Nozick's Anarchy,State, and Utopia (New York: Basic, 1974)- howeversurprising the rightshe singlesout as fundamental- is also an instance of interactionalcosmopolitanism. For institutionalcosmopolitanism, see Charles Beitz, PoliticalTheory and InternationalRelations (Princeton, N.J.: Princeton UniversityPress, 1979), pt. 3, and "Cosmopolitan Ideals and National Sentiment,"Journal of Philosophy 80 (1983): 591-600; and Pogge, RealizingRawls, chap. 6. 6. This is done, e.g., by John Rawls, who asserts (i) a natural duty to uphold and promotejust institutionsand also (ii) various othernatural duties that do not presuppose shared institutions,such as duties to avoid injuryand cruelty,duties to render mutual aid, and a duty to bring about just institutionswhere none presentlyexist. See John Rawls, A Theoryof Justice (Cambridge, Mass.: Harvard UniversityPress, 1971), pp. 114-15, 334.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 51 sharedpractices. The latterapproach has twostraightforward limitations. First,its applicabilityis contingent,in thathuman rightsare activated only through the emergence of social institutions.Where such institutionsare lacking,human rightsare merelylatent and human rights violationscannot exist at all. Thus, if we accept a purely institutional conception of human rights,then we need some additional moral conceptionif we wishto deny thatall is permittedin a verydisorganized state of nature. Second, the cosmopolitanismof the institutionalapproach is contingent as well, in that the global moral force of human rights is activated only through the emergence of a global scheme of social institutions,which triggers obligations to promoteany feasiblereforms of this scheme that would enhance the fulfillmentof human rights. So long as thereis a pluralityof self-containedcultures, the responsibility for such violations does not extend beyond their boundaries.7 It is only because all human beings are now participantsin a single,global institutionalscheme-involving such institutionsas the territorialstate and a systemof internationallaw and diplomacy as well as a world marketfor capital, goods, and services-that all human rightsviolations have come to be, at least potentially,everyone's concern.8 These two limitationsdo not violate generality.I have a duty toward every other person not to cooperate in imposing an unjust institutionalscheme upon her, even while this duty triggershuman- rights-basedobligations only to fellow participantsin the same institutional scheme. This is analogous to how the duty to keep one's promises is general even while it triggersobligations only vis-A-vis persons to whom one has actually made a promise. We see here how the institutionalapproach makes available an appealing intermediateposition between two interactionalextremes: it goes beyond simplelibertarianism, according to whichwe may ignore harms that we do not directlybring about, withoutfalling into a util- itarianismof rights'a la Shue, which commands us to take account of all relevant harms whatsoever,regardless of our causal relation to these harms.9

7. On the interactionalapproach, by contrast,any positive human rightswould impose duties on persons anywhere to give possible aid and protectionin specified cases of need. 8. These two limitationsare compatible with the belief that we have a duty to create a comprehensiveinstitutional scheme. Thus, Kant believed thatany persons and groups who cannot avoid influencingone another ought to enter into a juridical state. See Hans Reiss, ed., Kant's PoliticalWritings (Cambridge: Cambridge UniversityPress, 1970), p. 73. 9. The second extreme I am here alluding to is consequentialismin ethics, i.e., any consequentialistview that applies directlyto agents-be it of the ideal or real, of theact, rule, or motivevariety. There are also noninteractionalvariants of consequentialism, such as Bentham's utilitarianismwhich applies to institutions.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 52 Ethics October1992 Consider a human rightnot to be enslaved. On an interactional view, this rightwould constrain persons, who must not enslave one another. On an institutionalview, the rightwould constrainlegal and economic institutions:slavery must not be permittedor enforced.This leads to an importantdifference regarding the moral role of those who are neither slaves nor slaveholders. On the interactionalview, such thirdparties have no responsibilityvis-A-vis existing slaves, unless the human rightin question involved,besides the negative duty not to enslave, also a positive duty to protector rescue others from enslavement. Such positive duties have been notoriouslycontroversial. On the institutionalview, by contrast,some thirdparties may be im- plicated far more directlyin the human rightsviolation. If they are not making reasonable effortstoward institutionalreform, the more privilegedparticipants in an institutionalscheme in which slaveryis permittedor even enforced-even those who own no slaves themselves-are here seen as cooperating in the enslavement,in violation of a negative duty.The institutionalview thus broadens the circle of those who share responsibilityfor certain deprivations and abuses beyond what a simple libertarianismwould justify,and it does so withouthaving to affirmpositive duties. To be sure, workingfor institutionalreform is doing something (positive). But, in the contextof practices,this-as even libertarians recognize-does not entail that the duty in question is thereforea positive one: the negative duty not to abuse just practices may also generate positiveobligations, as when one must act to keep a promise or contractone has made. Once one is a participantin social practices, it may no longer be true that one's negative duties require merely forbearance. The move froman interactionalto an institutionalapproach thus blocks one way in which the rich and mightyin today's developed countrieslike to see themselvesas morallydisconnected from the fate of the less fortunatedenizens of the Third World. It overcomes the claim thatone need only refrainfrom violating human rightsdirectly, that one cannot reasonably be required to become a soldier in the global struggleagainst human rightsviolators and a comforterof their victimsworldwide. This claim is not refutedbut shown to be irrelevant. We are asked to be concerned about human rightsviolations not simply insofar as they exist at all, but only insofaras they are produced by social institutionsin whichwe are significantparticipants. Our negative duty not to cooperate in the impositionof unjust practices,together with our continuing participationin an unjust institutionalscheme, triggersobligations to promote feasible reformsof this scheme that would enhance the fulfillmentof human rights. One may thinkthat a shared responsibilityfor thejustice of the socialinstitutions in whichwe participatecannot plausibly extend beyond

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 53 our national institutionalscheme, in which we participateas citizens, and which we can most immediatelyaffect. But such a limitationis untenable because it treatsas natural or God-giventhe existingglobal institutionalframework, which is in fact imposed by human beings who are collectivelyquite capable of changing it. Therefore at least we-privileged citizens of powerful and approximatelydemocratic countries-share a collectiveresponsibility for the justice of the existing global order and hence also for any contributionit may make to the incidence of human rightsviolations.'0 The practical importanceof this conclusion evidentlyhinges on the extentto whichour globalinstitutional scheme is causallyresponsible for currentdeprivations. Consider this challenge: "Human rightsvi- olationsand theirdistribution have local explanations.In some countries tortureis rampant, while it is virtuallynonexistent in others. Some regions are embroiled in frequentwars, while othersare not. In some countries democratic institutionsthrive, while others bring fortha successionof autocrats.And again, some poor countrieshave developed rapidly,while others are gettingpoorer year by year. Therefore our global institutionalscheme has very littleto do with the deplorable state of human rightsfulfillment on earth." This challenge appeals to true premises but draws an invalid inference.Our global institutionalscheme can obviouslynot figurein the explanationof local human rightsviolations, but only in the macroexplanation of their global incidence. This parallels how Japanese culture may figurein the explanation of the Japanese suicide rate or how the laxityof U.S. handgun legislationmay figurein the explanation of the North American homicide rate, without therebyexplaining particular suicides/homicidesor even intercitydifferentials in rates. In these parallel cases the need for a macroexplanation is obvious from the fact that there are other societies whose suicide/homicide ratesare significantlylower. In the case of global institutions,the need for a macroexplanation of the overall incidence of human rightsviolations is less obvious because-apart fromsome ratherinconclusive historicalcomparisons- the contrastto observable alternativeglobal institutionalschemes is lacking. Still, it is highlylikely that there are feasible(i.e., practicableand accessible) alternativeglobal regimesthat would tend to engender lower rates of deprivation.This is clear, for example, in regard to economic institutions,where the centrifugal tendencies of certain free-marketschemes are well understood from our experience withvarious national and regional schemes. This supports a generalization to the global plane, to the conjecture that the currentconstitution of the world marketmust figureprominently in

10. Talk of such a contributionmakes implicitreference to alternativefeasible global regimes.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 54 Ethics October1992 the explanation of the factthat our world is one of vast and increasing internationalinequalities in income and wealth (withconsequent huge differentialsin nationalrates of infantmortality, life expectancy, disease, and malnutrition).Such a macroexplanation does not preempt mi- croexplanations of why one poor countryis developing rapidly and why another is not. It would explain why so few are while so many are not. Consider thisfurther challenge to the practicalmoral importance of our shared responsibilityfor the justice of our global institutional scheme: "An institutionalscheme can be held responsible for only those deprivationsit establishes,that is (at least implicitly),calls for. Thus, we cannot count against the currentglobal regime the factthat it tends to engender a high incidence of war, torture,and starvation because nothing in the existing (writtenor unwritten)international ground rules calls for such deprivations-they actually forbid both tortureand the waging of aggressive war. The prevalence of such deprivations thereforeindicates no flaw in our global order and, a fortiori,no global duties on our part (though we do of course have some local duties to see to it thatour governmentdoes not bringabout torture,starvation, or an unjust war)." This position is implausible. First,it would be irrationalto assess socialinstitutions without regard to theeffects they predictably engender. For an institutionalchange (e.g., in economic ground rules) might benefiteveryone (e.g., by increasingcompliance, or throughincentive effects).Second, social institutionsare human artifacts(produced and abolished, perpetuated and revised by human beings), and it would be unprecedented not to take account of the predictable effectsof human artifacts.(We choose between two engineeringdesigns by con- sidering not merely their suitabilityfor their particularpurpose but also theirincidental effects, e.g., on pollution and the like, insofaras these are predictable.) Third, we consistentlytake incidental effects into account in debates about the design of domesticinstitutions (in- centive effectsof penal and tax codes, etc.)." These argumentsreaffirm my broadlyconsequentialist assessment of social institutions,which leads us to aim for the feasible global institutionalscheme that produces the best patternof human rights fulfillment,irrespective of the extentto whichthis pattern is established or engendered.We thusconsider the existingglobal institutional scheme unjust insofar as the pattern of human rightsfulfillment it tends to produce is inferiorto the pattern that its best feasible alternatives would tend to produce. This broadly consequentialistvariant of institutionalcosmopolitanism accords withhow the concern for human 11. The supposed moral significanceof the distinctionbetween the established and the engendered effectsof social institutionsis extensivelydiscussed in Pogge, RealizingRawls, secs. 2-4.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 55 rightsis understood withinthe UniversalDeclaration of Human Rights. Section 28 reads: "Everyone is entitled to a social and International order in which the rightsand freedoms set forthin this Declaration can be fullyrealised" (my emphasis).12 This resultsuggests a furtherdifference between the interactional and institutionalapproaches, concerning the way each countsviolations of certain human rights.It cannot reasonably be required of an institutionalscheme, for example, thatit reduce the incidenceof physical assaults to zero. This would be impossible,and approximatingsuch an ideal as closely as possible would require a police state. The institutional approach thus counts a person's human right to physical integrityas fullysatisfied if her physicalintegrity is reasonablysecure.13 This entails that-even in the presence of a shared institutional scheme-some of what count as human rightsviolations on the interactionalview (e.g., certain assaults) do not count as human rights violationson the institutionalview (because the personswhose physical integritywas violated were reasonably well protected). Conversely, some of what count as human rightsviolations on the institutional view (e.g., inadequate protectionagainst assaults) may not registeron the interactionalview (as when insufficientlyprotected persons are not actuallyassaulted). Let me close thismore abstractpart of mydiscussion with a sketch of how my institutionalview relates to social and economic human rightsand the notion of distributivejustice. A man sympatheticto the moral claims of the poor, Michael Walzer, has written:"The idea of distributivejustice presupposes a bounded world,a community,within whichdistributions take place, a group of people committedto dividing, exchanging, and sharing, firstof all among themselves."'14 This is preciselythe pictureof distributivejustice thatRobert Nozick (among others)has so vigorouslyattacked. To the notionof dividinghe objects that "there is no centraldistribution, no person or group entitledto control all the resources,jointly deciding how they are to be doled out."'15 And as for the rest, he would allow persons to do all the

12. Similarlyalso Rawls's firstprinciple of justice: "Every person has the same indefeasibleclaim to a fullyadequate scheme of equal basic liberties,which scheme is compatible with the same scheme of libertiesfor all" (latest version,unpublished). In both cases the postulated entitlementor claim is clearlysecond order. 13. This notion is defined in probabilisticterms, perhaps by taking account of various personal characteristics.Thus it is quite possible thatthe human rightto physical integrityis today fulfilledin the United States for middle-aged whitesor suburbanites but not for black youths or inner-cityresidents. 14. Michael Walzer, "The Distributionof Membership,"in Boundaries,ed. Peter Brown and Henry Shue (Totowa, N.J.: Rowman & Littlefield,1981), p. 1. Compare the largelyidentical chap. 2 of Michael Walzer,Spheres ofJustice (New York: Basic, 1983), p. 31. 15. Nozick, p. 149.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 56 Ethics October1992 exchanging and sharing they like, but stronglyreject any enforced sharing implementedby some redistributionbureaucracy. The institutionalapproach involves a conception of distributive justice that differssharply from the one Walzer supports and Nozick attacks.Here the issue of distributivejustice is not how to distribute a given pool of resourcesor how to improveupon a given distribution but,rather, how to choose or design the economic ground rules,which regulate property,cooperation, and exchange and therebycondition productionand distribution.(On the particularview I have defended, e.g., we should aim for a set of economic ground rules under which each participantwould be able to meet her basic social and economic needs.) These economic ground rules-the object of distributivejustice on the institutionalapproach-are prior to both productionand dis- tributionand thereforeinvolve neither the idea of an already existing pool of stuffto be doled out nor the idea of already owned resources to be redistributed. The institutionalconception of distributivejustice also does not presuppose the existence of a communityof persons committedfirst of all to share with one another. Rather, it has a far more minimal rationale: we face a choice of economic ground rules that is partly open-not determined by causal necessity,nor preempted by some God-givenor naturalor neutralscheme that we mustchoose irrespective of its effects.This choice has a tremendous impact on human lives, an impactfrom which persons cannot be insulatedand cannot insulate themselves. Our present global economic regime produces a stable pattern of widespread malnutritionand starvationamong the poor (withsome 20 millionpersons dying every year from hunger and trivial diseases), and there are likelyto be feasible alternativeregimes that would not produce similarlysevere deprivations. In such a case of avoidable deprivations,we are confrontednot by persons who are merely poor and starving but also by victims of an institutional scheme-impoverished and starved.There is an injusticein thiseco- nomicscheme, which it would be wrongfor its more affluent participants to perpetuate. And that is so quite independentlyof whetherwe and the starvingare united by a communal bond or committedto sharing resources with one another,just as murderinga person is wrong ir- respectiveof such considerations.This is what the assertionof social and economic human rights comes to within my institutionalcosmopolitanism. This institutionalcosmopolitanism does not, as such, entail crisp practical conclusions. One reason for this is that I have not-apart fromallusions to Rawls and the UniversalDeclaration-given a fulllist of preciselydefined human rightstogether with relative weightsor priorityrules. Anotherreason is thatthis institutional cosmopolitanism bears upon the burning issues of the day only in an indirectway,

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 57 mediated by empirical regularitiesand correlations.This is so chiefly because of itsbroadly consequentialist character, that is, itscommitment to take the engendered consequences of an institutionalscheme as seriously,morally, as its established consequences. Whether an institutionalscheme establishesavoidable deprivationsor inequalities(such as slavery or male suffrage) can be read off from the (writtenor unwritten)ground rules characterizingthis scheme. With regard to engendered deprivationsand inequalities,however, we face far more complexempirical questions about how the existinginstitutional scheme, compared to feasiblemodifications thereof, tends to affectthe incidence of human rights violations, such as rates of infant mortality,child abuse, crime, war, malnutrition,poverty, personal dependence, and exclusion fromeducation or health care. The interventionof such empirical matters,and the openness of the notion of human rights,do not mean that no conclusions can be drawn about the burning issues, only that what we can conclude is less precise and less definitethan one mighthave hoped. THE IDEA OF STATE SOVEREIGNTY Before discussing how we should thinkabout sovereigntyin light of myinstitutional cosmopolitanism, let me definethis term, in a somewhat unusual way, as a two-place relation: A is sovereignover B if and only if 1. A is a governmentalbody or officer("agency"), and 2. B are persons, and 3. A has unsupervised and irrevocableauthority over B a) to lay down rules constrainingtheir conduct, or b) to judge their compliance with rules, or c) to enforce rules against them through preemption, pre- vention,or punishments,or d) to act in theirbehalf vis-a'-visother agencies (ones thatdo or do not have authorityover them)or persons(ones whom A is sovereign over, or not). A has absolutesovereignty over B if and only if 1. A is sovereign over B, and 2. no other agency has any authorityover A or over B which is not supervised and revocable by A. Any A having (absolute) sovereigntyover some B can then be said to be an (absolute) sovereign (the one-place predicate).'6 Central to contemporarypolitical thoughtand realityis the idea of the autonomous territorialstate as the preeminentmode of political

16. It is quite possible,and not withouthistorical justification, to definesovereignty the way I have defined absolute sovereignty.In that case the expression "distribution of sovereignty"would be an oxymoron.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 58 Ethics October1992 organization. In the vertical dimension, sovereigntyis very heavily concentratedat a single level; it is states and only states that merit separate colors on a political map of our world. For nearly every human being, and for almost everypiece of territory,there is exactly one government with preeminent authorityover, and primaryre- sponsibilityfor, this person or territory.And each person is thought to owe primarypolitical allegiance and loyaltyto thisgovernment with preeminentauthority over him or her. National governmentsdominate and control the decision making of smaller political units as well as supranationaldecisions, which tend to be made throughintergovern- mental bargaining.'7 From the standpointof a cosmopolitan morality-which centers around the fundamental needs and interestsof individual human beings, and of all human beings-this concentrationof sovereignty at one level is no longer defensible.What I am proposing instead is not the idea of a worldstate, which is reallya variantof thepreeminent- state idea. Rather, the proposal is that governmentalauthority-or sovereignty-be widelydispersed in the verticaldimension. What we need is both centralizationand decentralization,a kind of second- order decentralizationaway fromthe now dominantlevel of the state. Thus, persons should be citizensof, and govern themselvesthrough, a number of political units of various sizes, withoutany one political unit being dominant and thus occupying the traditionalrole of state. And theirpolitical allegiance and loyalties'8should be widelydispersed over these units: neighborhood,town, county, province, state, region, and world at large. People should be politicallyat home in all of them, without converging upon any one of them as the lodestar of their political identity.'9 Before defending and developing this proposal by referenceto myinstitutional cosmopolitanism, let me address twotypes of objection to any verticaldivision of sovereignty.

17. One promisingexception to this is the European Parliament. 18. This includes the sentimentsof patriotism,if such there must be. Beitz points out two respects in which patrioticallegiance to political units may be desirable: it supports a sense of shared loyalty("Cosmopolitan Ideals," p. 599); and it allows one to see oneself as a significantcontributor to a common culturalproject: "Justas we can see ourselvesas strivingto realize in our own livesvarious forms of individualperfection, so we can see our countries as strivingfor various forms of social and communal perfection"("Cosmopolitan Ideals," p. 600). Neitherof these considerationsentail that, say, Britain must be the sole object of your patrioticallegiance ratherthan some com- bination of Glasgow, Scotland, Britain, Europe, humankind, and perhaps even such geographicallydispersed units as theAnglican church, the WorldTrade Union Movement, PEN, or AmnestyInternational. 19. Many individualsmight, of course, identifymore withone of theircitizenships than withthe others.But in a multilayeredscheme such prominentidentifications would be less frequentand, most important,would not converge: even if some residentsof Glasgow would see themselvesas primarilyBritish, others would identifymore with Europe, with Scotland, withGlasgow, or with humankind at large.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 59 Objections of type 1 dispute that sovereigntycan be divided at all. The traditionalform of this objection rests on the belief that a juridical state (as distinctfrom a lawless state of nature) presupposes an absolute sovereign.This dogma of absolute sovereigntyarises (e.g., in Hobbes and Kant) roughlyas follows.Ajuridical state,by definition, involvesa recognized decision mechanism that uniquely resolves any dispute. This mechanismrequires some agencybecause a mere written or unwrittencode (constitution,holy scripture)cannot settledisputes about its own interpretation.But so long as this agency is limitedor divided-whether horizontally(i.e., by territoryor by governmental function)or vertically(as in my proposal)-a juridical state has not been achieved because there is no recognized way in which conflicts over the precise location of the limitor divisioncan be authoritatively resolved. A genuine state of peace requires then an agency of last resort-ultimate, supreme, and unconstrained.Such an agency may stillbe limitedby (codified or unmodified)obligations. But these can obligate merelyin foro interno because to authorize subjects,or some second agency, to determinewhether the firstagency is overstepping its bounds would enable conflictsabout this question for which there would be no legal path of resolution.20 This argument,which-strictly construed-would require an absolute world sovereign,has been overtakenby the historicalfacts of the last two hundred years or so, which show conclusivelythat what cannot work in theory works quite well in practice. Law-governed coexistence is possible withouta supreme and unconstrainedagency. There is, it is true, the possibilityof ultimateconflicts: of disputes in regardto whicheven thelegally correct method of resolutionis contested. To see this,one need only imagine how a constitutionaldemocracy's threebranches of governmentmight engage in an all-outpower struggle, each going to the verybrink of what, on its understanding,it is con- stitutionallyauthorized to do. From a theoreticalpoint of view, this possibilityshows that we are not insured against, and thus live in permanent danger of, constitutionalcrises. But this no longer un- dermines our confidence in a genuine division of powers: we have learned that such crises need not be frequentor irresolvable.From a practical point of view, we know that constitutionaldemocracies can endure and can ensure a robustjuridical state.

20. This dogma-prefigured in Aquinas, Dante, Marsilius,and Bodin-is most fullystated in chaps. 14, 26, and 29 of Thomas Hobbes, Leviathan(Harmondsworth: Penguin, 1981), who also introduces the idea of obligationsin foro intern. For Kant's statementsof it, see Reiss, ed., pp. 75, 81, 144-45. The dogma maintained its hold well into the twentiethcentury, when it declined togetherwith the Austinianconception ofjurisprudence. See GeoffreyMarshall, Parliamentary Sovereignty and theCommonwealth (Oxford: Oxford UniversityPress, 1957), pt. 1; S. I. Benn and R. S. Peters, Social Principlesand theDemocratic State (London: Allen & Unwin, 1959), chaps. 3, 12; and Herbert L. A. Hart, The Conceptof Law (Oxford: Oxford UniversityPress, 1961).

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 60 Ethics October1992 This same point applies in the verticaldimension as well: just as it is nonsense to suppose that (in a juridical state) sovereigntymust restwith one of the branches of government,it is similarlynonsensical to thinkthat in a multilayeredscheme sovereignty must be concentrated on one level exclusively.As the historyof federalistregimes clearly shows,a verticaldivision of sovereigntycan workquite well in practice, even while it leaves some conflictsover the constitutionalallocation of powers withouta legal path of authoritativeresolution. Objections of type2 oppose, more specifically,a verticaldispersal of sovereignty:there are certain verticallyindivisible governmental functionsthat form the core of sovereignty.Any politicalunit exercising these core functionsmust be dominant-free to determinethe extent to which smallerunits within it may engage in theirown local political decision making, even while its own political process is immune to regulation and review by more inclusive units. Vertical distributions of sovereignty,if they are to exist at all, must thereforebe lopsided (as in currentfederal regimes). To be assessable, such a claim stands in need of two clarifications, whichare rarelysupplied. First, when one thinksabout it morecarefully, it turns out to be surprisinglydifficult to come up with examples of indivisiblegovernmental functions. Eminent domain, economic policy, foreign policy,judicial review; the control of raw materials,security forces, education, health care, and income support; the regulation and taxation of resource extractionand pollution,of work and con- sumptioncan all be handled at variouslevels and indeed are so handled in existingfederal regimes and confederations.So what are the gov- ernmentalfunctions that supposedly are verticallyindivisible? Second, is theirindivisibility supposed to be derived froma conceptual insight, fromempirical exigencies, or frommoral desiderata?And whichones? Since I cannot here discuss all possible type 2 objections,let me concentrateon one paradigm case: Walzer's claim that the authority to fixmembership, to admitand exclude,is at least partof an indivisible core of sovereignty:"At some level of politicalorganization something like the sovereign state must take shape and claim the authorityto make its own admissions policy,to controland sometimesto restrain,, the flowof immigrants.",2'Walzer's "must" does not reflecta conceptual or empirical necessity,for in those senses the authorityin question quite obviously can be divided-for example, by allowing political units on all levels to veto immigration.It is on moral grounds that Walzer rejects such an authorityfor provinces,towns, and neighborhoods: it would "create a thousand pettyfortresses."22 But if smaller unitsare to be precluded fromcontrolling the influxof new members,

21. Walzer, "Distribution,"p. 10. 22. Ibid., p. 9.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 61 then immigrationmust be controlled at the state level: "Only if the state makes a selectionamong would-be membersand guarantees the loyalty,security, and welfare of the individuals it selects, can local communitiestake shape as 'indifferent'associations, determined only by personalpreference and marketcapacity."23 The assertedconnection is again a moralone: itis certainlyfactually possible for local communities to exist as indifferentassociations even while no control is exercised over migrationat all; as Walzer says,"The fortressestoo could be torn down, of course."24 Walzer's point is, then, that the insistence on openness (to avoid a thousand pettyfortresses) is asking too much of neighborhoods,unless the state has control over immigration:"The distinctivenessof cultures and groups depends upon closure.... If thisdistinctiveness is a value, . . . thenclosure mustbe permittedsome- where."25 But is the conventionalmodel, withthis rationale, really morally necessary?To be sure,Walzer is rightto claimthat the value of protecting cohesiveneighborhood cultures is betterserved by nationalimmigration controlthan by no controlat all.26But it would be much betterserved still if the state were constrained to admit only immigrantswho are planning to move into a neighborhood that is willingto accept them. Moreover,since a neighborhoodculture can be as effectivelydestroyed by the influxof fellownationals as by thatof immigrants,neighborhoods would do even better,if theyhad some authorityto selectfrom among prospective domestic newcomers or to limit their number. Finally, neighborhoods may often want to bring in new members from abroad-persons to whom theyhave specialethnic, religious, or cultural ties-and they would thereforebenefit from a role in the national immigrationcontrol process that would allow them to facilitatethe admission of such persons. Thus there are at least three reasons for believingthat Walzer's rationale-cohesive neighborhoodcultures ought to be protectedwithout becoming pettyfortresses-is actuallybetter served by a divisionof the authorityto admit and exclude than by the conventionalconcentration of this authorityat the level of the state. SOME MAIN REASONS FOR A VERTICAL DISPERSAL OF SOVEREIGNTY Having dealt withsome preliminaryobstacles, let me now sketchfour main reasonsfavoring, over the statusquo, a worldin whichsovereignty is widelydistributed vertically. 1. Peace/security.-Under the current regime, interstaterivalries are settledultimately through military competition, including the threat

23. Ibid. 24. Ibid. 25. Ibid., pp. 9-10. 26. Ibid., p. 9.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 62 Ethics October1992 and use of militaryforce. Moreover, within their own territories,national governmentsare free to do virtuallyanything they like. Such governments thereforehave very powerful incentivesand very broad opportunitiesto develop their militarymight. This is bound to lead to the furtherproliferation of nuclear, biological,chemical, and conven- tional weapons of mass destruction.And in a world in which dozens of competingnational governments control such weapons,the outbreak of devastatingwars is only a matterof time. It is not feasibleto reduce and eliminatenational control over weapons of mass destructionthrough a program that depends upon the voluntarycooperation of each and everynational government.What is needed, therefore,is the centrally enforcedreduction and eliminationof such weapons-in violationof the prevalentidea of statesovereignty. Such a program,if implemented soon, is much less dangerous than continuingthe statusquo. It could gain the support of most peoples and governments,if it increases the securityof all on fairterms that are effectivelyadjudicated and enforced. 2. Reducingoppression. -Under the currentglobal regime,national governmentsare effectivelyfree to control"their" populations in what- ever way they see fit. Many make extensive use of this freedom by torturingand murderingtheir domestic opponents, censoringinfor- mation,suppressing and subvertingdemocratic procedures, prohibiting emigration,and so forth.This problem could be reduced through a verticaldispersal of sovereigntyover various layers of political units that would check and balance one another as well as publicize one another's abuses. 3. Globaleconomic justice. The magnitude and extentof current economic deprivations-over 20 million persons die everyyear from poverty-relatedcauses-calls for some modificationin the prevailing scheme of economic cooperation. One plausible reformwould involve a global levyon the use of natural resources to support the economic developmentin the poorest areas.27Such a levywould tend to equalize per capita endowmentsand also encourage conservation.Reforms for the sake of economic justice would again involve some centralization-though withoutrequiring anything like a global welfarebu- reaucracy. Global economicjustice is an end in its own right,which requires, and thereforesupports, a reallocation of politicalauthority. But it is also importantas a means toward the firsttwo purposes. War and oppressionresult from the contestfor power withinand among political units, which tends to be the more intense the higher the stakes. In fightsto govern states, or to redraw their borders, far too much is now at stake by way of control of people and resources. We can best 27. For furtherdiscussion of such a reform-backed perhaps by the idea thatthe world's resources should be owned or controlled by all its inhabitantsas equals-see Beitz, PoliticalTheory, pp. 136-43; and Pogge, RealizingRawls, pp. 250-52, 263-65.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 63 lower the stakes by dispersing political authorityover several levels and institutionallysecuring economic justice at the global level. This importantpoint suggestswhy my firstthree considerations -though each supportssome centralization-do not on balance support a world state. While a world state could lead to significantprogress in termsof peace and economic justice, it also poses significantrisks of oppression. Here the kind of multilayeredscheme I propose has the great advantages of affordingplenty of checks and balances and of assuring that,even when some politicalunits turn tyrannicaland op- pressive,there will always be other, already fullyorganized political units (above, below, or on the same level) which can render aid and protectionto the oppressed, publicize the abuses, and, if necessary, fightthe oppressors. There are two furtherimportant reasons against a world state. Cultural and social diversityare likely to be much better protected when the interestsof culturalcommunities at all levelsare represented (externally)and supported (internally)by coordinate political units. And the scheme I propose could be gradually reached from where we are now (throughwhat I have called second-orderdecentralization), while a world state-involving, as it does, the annihilationof existing states-would seem reachable only throughrevolution or in the wake of some global catastrophe. 4. Ecology.-Modern processes of production and consumption are liable to generate significantnegative externalities that, to a large and increasingextent, transcend national borders. In a world of com- peting autonomous states,the internalizationof such externalitiesis generallyquite imperfectbecause of familiarisolation, assurance, and coordination problems. Treaties among a large number of very dif- ferentlysituated actors require difficult and time-consumingbargaining and negotiations,which often lead to only veryslight progress, if any. And even when treatiesare achieved, doubts about the fullcompliance of other parties tend to erode each party'sown commitmentto make good-faithefforts toward compliance. Now one might think that this fourthreason goes beyond my institutionalcosmopolitanism because there is no recognized human rightto a clean environment.Why should people not be free to live in a degraded natural environmentif they so choose? In response, perhaps they should be, but for now they won't have had a choice. The degradation of our naturalenvironment ineluctably affects us all. And yet,most people are effectivelyexcluded fromany say about this issue which,in the currentstate-centric model, is regulatedby national governmentsunilaterally or through intergovernmentalbargaining heavilyinfluenced by huge differentialsin economicand militarymight. This response suggestsreplacing ecologywith a deeper and more general fourthreason, whichmight be labeled democracy:persons have

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 64 Ethics October1992 a rightto an institutionalorder under which those significantlyand legitimately28affected by a political decision have a roughly equal opportunityto influence the making of this decision-directly or through elected delegates or representatives.29Such a human right to politicalparticipation also supportsgreater local autonomyin matters of purely local concern than exists in most current states or would existin a world state,however democratic.In fact,it supportsjust the kind of multilayeredinstitutional scheme I have proposed. Before developing thisidea further,let me consideran objection. One might say, against a human rightto political participation,that what mattersabout politicaldecisions is that theybe correct,not that they be made democraticallyby those concerned. But this objection applies, firstof all, only to politicalchoices thatare morallyclosed and thus can be decided correctlyor incorrectly.I believe that we should rejecta view on whichalmost all politicalchoices are viewed as morally closed (withthe correctdecision determined,perhaps, throughutility differentials),but I have no space here to defend thisbelief. Second, even when politicalchoices are morallyclosed, the primaryand ultimate responsibilityfor their being made correctlyshould lie withthe persons concerned.Of course,some otherdecision procedure-such as a group of experts-may be more reliable for this or that kind of decision, and such procedures (judges, parliaments,cabinets, etc.) should then be put in place. This should be done, however,by the people delegating, or abstaining from,such decisions. It is ultimatelyup to them, and not to self-appointedexperts, to recognize the greater reliabilityof, and to institutionalize,alternative decision-making procedures.

28. The qualification"legitimately" is necessaryto rule out claims such as this: "I should be allowed a vote on the permissibilityof homosexuality,in all parts of the world, because the knowledge that homosexual acts are performedanywhere causes me great distress."I cannot enter a discussion of this proviso here, except to say that the argumentsrelevant to its specificationare by and large analogous to the standard argumentsrelevant to the specificationof Mill's no-harm principle. 29. I understand opportunityas being impaired only by (social) disadvantages not by (natural) handicaps. This is plausible only on a narrowconstrual of "handicap." Although being black and being female are natural features,they reduce a person's chances to affectpolitical decisions only in certain social settings (in a racist/sexist culture). Such reductions should thereforecount as disadvantages. By contrast,those whose lesser abilityto participatein public debate is due to theirlow intelligenceare not disadvantaged but handicapped. They do not count as having a less-than-equal opportunity.The postulatedhuman rightis not a group right.Of course,the inhabitants of a townmay appeal to thisright to show thatit was wrongfor the nationalgovernment, say,to impose some politicaldecision that affects only them. In such a case, thetownspeople forma group of those having a grievance.But theydo not have a grievanceas a group. Rather,each of them has such a grievance of not having been given her due political weight-just the grievance she would have had, had the decision been made by other townspeople with her excluded.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 65 Given the postulated human rightto political participation,the proper verticaldistribution of sovereigntyis determinedby three sets of considerations. The firstfavor decentralization,the second centralization,while the thirdmay correctthe resultingbalance in either direction. First,decision making should be decentralizedas far as possible. This is desirable in part, of course, in order to minimizethe decision- makingburdens upon individuals.But thereare moreimportant reasons as well. Insofar as decisions are morally closed, outsiders are more likely to lack the knowledge and sensitivitiesto make responsible judgments-and the only practicable and morallyacceptable way of delimitingthose who are capable of such judgments is by rough geographical criteria.Insofar as decisions are morallyopen, the end must be to maximize each person's opportunityto influencethe social conditions that shape her life-which should not be diluted for the sake of enhancing persons' opportunitiesto influencedecisions of merely local significanceelsewhere. At least persons should be left free to decide for themselves to what extent to engage in such exchanges. The firstconsideration does not then rule out voluntarycreation of central decision-makingmechanisms (even though their structure- dependent upon unanimous consent-would tend to reflectthe participants'bargaining power). Such centralizationmay be rational,for example,in cases of conflictbetween local and globalrationality (tragedy- of-the-commonscases: fishing,grazing, pollution) and also in regard to desired projectsthat require manycontributors because theyinvolve coordinationproblems or economies of scale, forexample, or because theyare simplytoo expensive (constructionand maintenanceof trans- portationand communicationsystems, research and technology,space programs,and so forth). The second consideration favors centralizationinsofar as this is necessary to avoid excluding persons from the making of decisions that significantly(and legitimately)affect them. Such decisions are of two-possibly three-kinds. Inhabitingthe same naturalenvironment and being significantlyaffected by what others do to it, we have a rightto participatein regulatinghow it may be used. And since the lives each of us can lead are very significantlyshaped by prevailing institutions-suchas marriage,reproduction and birthcontrol, property, money,markets, and formsof politicalorganization-we have a right to participatein theirchoice and design. These two kinds of decision arise directlyfrom Kant's point that human beings cannot avoid influencingone another:through direct contact and throughtheir impact upon the natural world in which they coexist. A rightto participate in decisions of the thirdkind is more controversial.There are contexts, one might say, in which we act as a species and thus should decide together how to act. Examples might be our conduct toward other

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 66 Ethics October1992 biological species (extinction,genetic engineering,cruelty), ventures into outer space, and the preservationof our human heritage(ancient skeletonsand artifacts,great works of art and architecture,places of exceptional natural beauty). In all these cases it would seem wrong for one person or group to take irremediablesteps unilaterally. The significanceof the second considerationdepends heavilyupon empirical matters,though it does so in a ratherstraightforward and accessible way. It is obvious upon minimal reflectionthat the devel- opmentsof the past fewcenturies have greatlyincreased the significance of thisconsideration in favorof centralization.This is so partlybecause of rising population density,but much more importantlybecause of our vastlymore powerfultechnologies and the tremendouslyincreased level of global interdependence.Concerning technologies, the factthat what a population does withinits own national borders-stockpiling weapons of mass destruction,depleting nonrenewable resources, cutting down vegetation essential for the reproduction of oxygen, emitting pollutants that are destroying the ozone layer and cause global warming-now often imposes verysignificant harms and risksupon outsidersbrings into play the politicalhuman rightsof these outsiders, therebymorally undermining the conventionalinsistence on absolute stateautonomy. Global interdependenceis bestillustrated by the emergence of trulyglobal capital and commoditymarkets (as dramatically illustratedby the stock marketcrash of October 1987): a change in Japanese interestrates, or a speculativefrenzy of short-sellingon the Chicago Futures Exchange, can literallymake the differencebetween life and death for large numbers of people half a world away-in Africa,for example, where manycountries depend upon foreignbor- rowing and cash crop exports. Such interdependence is not bad as such (it can hardly be scaled back in any case), but it does require democraticcentralization of decisionmaking: as more and more persons are significantlyaffected by certaininstitutions, more and more persons have a rightto a politicalrole in shaping them. The possibilityof free bargaining over the design of such institutionsdoes not satisfythe equal-opportunityprinciple, as is illustratedin the case of commodity marketsby the factthat African populations simplylack the bargaining power thatwould allow them significantlyto affecthow such markets are organized. (This argument withstandsthe communitarianclaim that we must reject supranational democratic processes for the sake of the value of national autonomy.Such rejectiondoes indeed enhance the national autonomy of the advantaged First World populations. But theirgain is purchased at the expense of poorer populations who, despite fictionalor de jure state sovereignty,have virtuallyno control over the most basic parameters that shape their lives-a problem heightenedby the factthat even theirown, rather impotent governments face strongincentives to cater to foreigninterests rather than to those of theirconstituents.)

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 67 The firsttwo considerationsby themselvesyield the result that the authorityto make decisions of some particularkind should rest with the democratic political process of a unit that (i) is as small as possible but still (ii) includes as equals all persons significantlyand legitimatelyaffected by decisionsof thiskind. In practice,some trading- offis required between these two considerationsbecause therecannot always be an established political process that includes as equals all and only those significantlyaffected. A matteraffecting the populations of two provinces,for example, mightbe referredto the national parliament or might be left to bargaining between the two provincial governments.The formersolution caters to (ii) at the expense of (i): involvingmany persons who are not legitimatelyaffected. The latter solutioncaters to (i) at the expense of (ii): givingthe personslegitimately affectednot an equal opportunityto influencethe matterbut one that depends on the relativebargaining power of the two provincialgovernments. The firsttwo considerations would sufficeon the ideal-theory assumption that any decisions made satisfyall moral constraintswith regard to both procedure (the equal-opportunityrequirement) and output (thisand otherhuman rights).This assumption,however, could hardlybe strictlytrue in practice. And so a thirdconsideration must come into play: what would emerge as the proper verticaldistribution of sovereigntyfrom a balancing of the firsttwo considerationsalone should be modified-in eitherdirection-if such modificationsignif- icantlyincreases the democratic nature of decisionmaking or itsreliability (as measured in terms of human rightsfulfillment). Let me briefly discuss how this thirdconsideration might make a difference. On the one hand, one must ask whetherit would be a gain for human rightsfulfillment on balance to transferdecision-making authority"upward" to largerunits-or (perhaps more plausibly)to make the politicalprocess of smallerunits subject to regulationand/or review by the politicalprocess of more inclusiveunits. Such authoritywould allow the larger unit,on human rightsgrounds,30 to require revisions in the structureof the political process of the smaller one and/orto nullifyits political decisions and perhaps also to enforcesuch revisions and nullifications. Even when such interventionsreally do protecthuman rights,this regulationand reviewauthority has some costsin termsof the political human rightsof the membersof the smallerunit. But then,of course, the largerunit's regulationand reviewprocess may itselfbe unreliable and thus may produce human rightsviolations either by overturning

30. Though not in defense of other procedural or substantiveconstraints to which the smaller unit may have chosen to commititself. Compare here the situationin the United States,where federal courts may reviewwhether laws and decisions at the state level accord withsuperordinate federal requirements, but not whetherthey accord with superordinaterequirements of that state itself.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 68 Ethics October1992 unobjectionable structuresor decisions (at even greater cost to the politicalhuman rightsof members of the smaller unit) or by forcing the smaller unit to adopt structuresand decisions thatdirectly violate human rights. On the other hand, there is also the inverse question: whether the thirdconsideration might support a move in the directionof decentralization.Thus one must ask to what extentthe politicalprocess of a larger unit is undemocraticor unreliable, and whetherit might be a gain forhuman rightsfulfillment on balance to transferdecision- makingauthority "downward" to smallerunits-or to investthe political process of such subunits with review authority.Such an authority might,for example, allow provincialgovernments, on human rights grounds, to block the application of national laws in their province. This authorityis justifiedif and only if its benefits(laws passed in an undemocraticmanner or violatinghuman rightsare not applied) out- weigh its costs (unobjectionable laws are blocked in violation of the political rightsof members of the larger unit). How such mattersshould be weighedis a highlycomplex question, which I cannot here address with any precision. Let me make two points nevertheless.First, a good deal of weight should be given to the actual viewsof those who sufferabridgments of theirhuman rights and forwhose benefita regulationand/or review authority might thus be called for. If most blacks in some statewould rathersuffer discrim- ination than see their state governmentconstrained by the federal government,then the presumptionagainst such an authorityshould be much weightierthan if the opposition came only fromthe whites. This is not to deny that victimsof injustice may be brainwashed or may sufferfrom false consciousness of various sorts. It may still be possible to make the case fora regulationand/or review authority. But it should be significantlymore difficultto do so. Second, commonalitiesof language, religion,ethnicity, or history are strictlyirrelevant. Such commonalitiesdo not give people a claim to be part of one another's political lives, nor does the lack of such commonalitiesargue against restraints.The presence or absence of such commonalitiesmay stillbe empiricallysignificant, however. Thus suppose that the members of some smaller unit share religious or ethnic characteristicsthat in the larger unit are in the minority(e.g., a Muslim provincewithin a predominantlyHindu state).Our historical experience withsuch cases may well support the viewthat a regulation and reviewauthority by the larger unit would probablybe frequently abused or that a review authorityby the smaller unit would tend to enhance human rightsfulfillment overall. The relevance of such in- formationbrings out that the required weighingsdo not depend on value judgments alone. They also depend on reasonable expectations about how alternativearrangements would actually work in one or another concrete context.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 69 The thirdconsideration must also play a centralrole in a special case: the question of where decisions about the proper allocation of decisionmaking should be made. For example,should a disputebetween a provincialparliament and a national legislatureover which of them is properlyin chargeof a particulardecision be referredto the provincial or the nationalsupreme court? Here again one mustpresent arguments to the effectthat the preferredlocus of decision making is likelyto be more reliable than its alternative. Nothing definitecan be said about the ideal number of levels or the exact distributionof legislative,executive, and judicial functions over them. These mattersmight vary in space and time, depending on the prevailingempirical factsto be accommodated by my second and thirdconsiderations (externalities, interdependence; unreliabilities) and on persons' preferencesas shaped by the historical,cultural, lin- guistic,or religious ties among them. The human right to political participationalso leaves room fora wide variety,hence regionaldiversity, of decision-making procedures-direct or representative,with or withoutpolitical parties, and so on. Democracy may take many forms. THE SHAPING AND RESHAPING OF POLITICAL UNITS One great advantage of the proposed multilayeredscheme is, I have said, that it can be reached gradually from where we are now. This requiresmoderate centralizingand decentralizingmoves involvingthe strengtheningof politicalunits above and below the level of the state. In some cases, such unitswill have to be created,and so we need some ideas about how the geographical shape of new politicalunits is to be determined.Or, seeing thatthere is considerabledissatisfaction about even the geographical shape of existingpolitical units, we should ask more broadly: What principlesought to govern the geographical sep- aration of political units on any level? Guided again by the cosmopolitan ideal of democracy,I suggest these two procedural principlesas a firstapproximation: 1. The inhabitantsof any contiguousterritory of reasonableshape may decide-through some majoritarianor supermajoritarian procedure-to join an existingpolitical unit whose territory is contiguous with theirsand whose population is willing- as assessed through some majoritarianor supermajoritarian procedure-to accept themas members.3'This libertyis con-

31. I won't tryto be precise about "reasonable shape." The idea is to rule out areas with extremelylong borders, or borders that divide towns, integrated networksof economic activity,or the like. Perhaps the inhabitantsin question should have to be minimallynumerous; but I think the threshold could be quite low. If a tinyborder village wants to belong to the neighboringprovince, why should it not be allowed to switch?The contiguitycondition needs some relaxing to allow territoriesconsisting of a small number of internallycontiguous areas whose access to one another is not controlled by other political units. The United States of America would satisfythis

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 70 Ethics October1992 ditional upon the political unit or units that are truncated throughsuch a move eitherremaining viable (with a contiguous territoryof reasonable shape and sufficientpopulation) or being willinglyincorporated, pursuant to the firstclause, into another political unit or other political units. 2. The inhabitantsof anycontiguous territory of reasonableshape, if sufficientlynumerous, may decide-through some majoritarianor supermajoritarianprocedure-to formthemselves intoa politicalunit of a level commensuratewith their number. This liberty is subject to three constraints:there may be subgroups whose members, pursuant to their libertyunder 1, are free to reject membershipin the unit to be formed in favorof membershipin another politicalunit. There may be subgroups whose members, pursuant to their libertyunder 2, are free to reject membershipin the unit to be formedin favorof formingtheir own politicalunit on the same level.32 And the politicalunit or unitstruncated through the requested move must either remain viable (with a contiguous territory of reasonable shape and sufficientpopulation) or be willingly incorporated,pursuant to the firstclause of 1, into another political unit or other political units. It will be said that acceptance of such principles would triggeran avalanche of applications. It is surely true that a large number of existinggroups are unhappy with their currentmembership status; there is a significantbacklog, so to speak, that might pose a serious short-termproblem. Once this backlog will have been worked down, however,there may not be much redrawingactivity as people willthen be contentwith their political memberships, and most borders will be supported by stable majorities. Moreover, as the advocated vertical dispersal of sovereigntyis implemented,conflicts over borders will lose much of theirintensity. In our world,many such conflictsare motivatedby morally inappropriate considerations-especially the followingtwo. There is competition over valuable or strategicallyimportant territories and groups because their possession importantlyaffects the distributionof international bargainingpower (economic and militarypotential) for the indefinite future. And there are attempts by the more affluentto interpose

relaxed condition through secure access among Puerto Rico, Alaska, Hawaii, and the remaining forty-eightcontiguous states. 32. What if minoritysubgroups are geographicallydispersed (like the Serbs in Croatia)? In such cases, there is no attractiveway of accommodatingthose opposed to the formationof the new political unit. My second principlewould let the preference of the majoritywithin the relevant territoryprevail nevertheless.This is defensible,I think,so long as we can bracketany concern forhuman rightsviolations. Where justice is not at stake, it seems reasonable, if legitimatepreferences are opposed and some must be frustrated,to let the majorityprevail.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 71 bordersbetween themselves and the poor in order to circumventwidely recognized duties of distributivejustice among compatriots.33Under the proposed multilayeredscheme-in which the political authority currentlyexercised by national governmentsis both constrainedand dispersed over several layers,and in which economic justice is insti- tutionalizedat the global level and thusinescapable-territorial disputes on any level would be only slightlymore intense than disputes about provincial or county lines are now. It is quite possible that my two principles are not suitable for defining a right to secession in our present world of excessivelysovereign states.34But their plausibility willincrease as the proposed second-orderdecentralization progresses.35 Finally,the incidence of applicationscan be reduced throughtwo reasonable amendments. First,the burden of proof, in appealing to eitherof the two principles,should restwith the advocates of change, who must map out an appropriate territory,organize its population, and so forth.This burden would tend to discourage frivolousclaims. Second, it maybe best to require some supermajoritarianprocess (e.g., proponents must outnumberopponents plus nonvotersin three con- secutivereferenda over a two-yearperiod). Some such provisionwould especiallyhelp preventareas changingback and forthrepeatedly (with outside supportersmoving in, perhaps, in order to tip the scales). Let me brieflyillustrate how the two principleswould workin the case of nested political units. Suppose the Kashmirisagree that they want to belong togetheras one province but are divided on whether this should be a province of India or of Pakistan. The majorityWest Kashmiris favor affiliationwith Pakistan, the East Kashmiris favor affiliationwith India. There are four plausible outcomes: a united Kashmiriprovince of Pakistan(P), a united Kashmiriprovince of India (I), a separate state of Kashmir (S), and a divided Kashmirbelonging partlyto Pakistan and partlyto India (D). Since the East Kashmiris can, by principle 2, unilaterallyinsist on D over P, they enjoy some protectionagainst the West Kashmirimajority. They can use thispro- tectionfor bargaining,which may result in outcome S (if this is the second preferenceon both sides) or even in outcome I (if that is the

33. See Alan Buchanan, Secession(Boulder, Colo.: Westview,1991), pp. 114-25; and Thomas W. Pogge, "Loopholes in Moralities,"Journalof Philosophy 89 (1992): 79- 98, pp. 88-90. 34. That topic is extensivelydiscussed by Buchanan. While he takes the current statessystem for granted and adjusts his theoryof secession accordingly,I am arguing thata more appealing theoryof secessionwould be plausiblein the contextof a somewhat differentglobal order. I therebyoffer one more reason in favorof the latter. 35. For example, as European stateswill increasingly become subject to global and regional constraints-regarding militarymight, pollution, exploitation of resources, treatmentof itscitizens, etc. -the importanceof whether there is one state(Czechoslovakia) or two states (one Czech, one Slovak) would tend to decline: for the Slovaks, for the Czechs, and for any third parties in the vicinity.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 72 Ethics October1992 second preference of the West Kashmiris while the East Kashmiris preferD or S over p).36 The conventionalalternatives to mycosmopolitan view on settling the borders of politicalunits reservea special role eitherfor historical statesand theirmembers (compatriots) or fornations and theirmembers (fellownationals). The formerversion is inherentlyconservative, the latterpotentially revisionist (by including,e.g., the Arab, Kurdish,and Armeniannations and by excludingmultinational states like the Soviet Union or the Sudan). The two key claims of such a position are: (a) Only (encompassing) groups of compatriots/fellownationals have a rightto self-government.(b) Such governmentmay be exercised even over unwillinggeographical subgroups of compatriots/fellownationals (who at most have a libertyof individual emigration).37Those who hold such a conventionalposition are liable to rejectmy cosmopolitan viewas excessivelyindividualist, contractarian, or voluntaristic.Examples of thissentiment are easy to find:"The more importanthuman group- ings need to be based on shared history,and on criteriaof nonvolun- taristic(or at least not whollycontractarian) membership to have the value that they have."38 Insofar as this is an empirical claim-about the preconditionsof authenticsolidarity and mutual trust,perhaps- I need not disagree with it.39If indeed a political unit is far more

36. Obviously, this storyis not meant to reflectthe actual situationon the Indian subcontinent. 37. While the precise definitionof 'nation' and 'nationality'is not essential to my discussion, I do assume that nationalityis not defined entirelyin voluntaristicterms (e.g., "a nation is a group of persons all of whom desire to constituteone politicalunit of which they are the only members" ), in which case the two claims would become trivial.The definitionmay stillcontain significantvoluntaristic elements, as in Renan's proposal: "A nation is a grand solidarityconstituted by the sentimentof sacrificeswhich one has made and those one is disposed to make again. It supposes a past" (quoted in Brian Barry, "Self-GovernmentRevisited," in The Natureof Political Theory, ed. David Miller and Larry Siedentop [Oxford: Clarendon, 1983], p. 136). So long as some non- voluntaristicelement is present,at least one of the two claims can get offthe ground: those who want to belong togetheras one politicalunit may be preventedfrom doing so when they lack an appropriate historyof solidarityand sacrifices. 38. Avishai Margalit and Joseph Raz, "National Self-Determination,"Journal of Philosophy57 (1990): 439-461, p. 456. 39. Though one should ask how this claim squares withthe historyof the United States,in the nineteenthcentury, say. Those who enjoyed the rightsof citizenshipwere highlyheterogeneous in descentand upbringing,and theycame as immigrants,through sheer choice. I do not believe these facts significantlyreduced the level of solidarity and mutual trustthey enjoyed, compared to the levels enjoyed in the major European states of that period. A careful study of this case mightwell show that people can be bound togetherby a common decision to follow the call of a certain constitutionand ideologyas well as the promiseof opportunitiesand adventure.If so, thiswould suggest that what matters for solidarityand mutual trust is the will to make a political life together and that such will is possible without unchosen commonalities.This result would hardlybe surprising,seeing how easilythe closestfriendships we formtranscend such commonalitiesof facialfeatures, native language, culturalbackground, and religious convictions.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 73 valuable for its members when they share a common descent and upbringing (language, culture, religion), then people will recognize this fact and will themselvesseek to form political units along these lines. I don't doubt thatgroups seekingto change theirpolitical status under the twoprinciples would forthe mostpart be groupscharacterized by such unchosen commonalities. But would I not give any other group, too, the rightto change its political status,even if this means exchanging a more valuable for a less valuable membership?Margalit and Raz ridiculethis idea through their examples of "the Tottenham Football Club supporters," "the fiction-readingpublic," and "thegroup of all the people whose surnames begin with a 'g' and end with an 'e.' "40 Yet these examples-apart from being extremelyfarfetched-are ruled out by the contiguity requirement,which a "voluntarist"can and, I believe, should accept in light of the key functionof government:to support shared rules among persons who cannot avoid influencingone another through direct interactionand through their impact upon theircommon environment.A more plausible example would then be that of the inhabitantsof a culturallyand linguisticallyItalian border village who preferan (exhypothesi) less valuable membershipin France over a more valuable membershipin Italy.Here I ask, Do theynot, France willing, have a right to err? Or should they be forced to remain in, or be turned over to, a superordinatepolitical unit against theirwill? This example bringsout the underlyingphilosophical value conflict. My cosmopolitanismis committedto the freedomof individualpersons and thereforeenvisions a pluralistglobal institutionalscheme. Such a scheme is compatible with political units whose membershipis ho- mogeneous withrespect to some partlyunchosen criteria(nationality, ethnicity,native language, history,religion, etc.), and itwould certainly engender such units. But it would do so only because persons choose to share theirpolitical life with others who are like themselvesin such respects-not because persons are entitledto be part of one another's politicallives if and only if they share certain unchosen features. One way of supportingthe conventionalalternative involves re- jecting the individualistpremise that only human beings are ultimate units of moral concern.41 One could then say that, once the moral claimsof states/nationsare takeninto account alongside those of persons, one maywell find that, all thingsconsidered, justice requires institutional arrangements that are inferior,in human rights terms, to feasible alternatives-institutionalarrangements, for example, under which the interestof Italyin itsborder village would prevailover the expressed interestof the villagers.

40. Margalit and Raz, pp. 443, 456. 41. For an example, see Brian Barry,"Do Countries Have Moral Obligations?"in TheTanner Lectures on Human Value,vol. 2, ed. S. M. McMurrin(Salt Lake City:University of Utah Press, 1981), pp. 27-44.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions 74 Ethics October1992 This justificatorystrategy faces two main problems. It is unclear how states/nationscan have interestsor moral claims that are not reducible to interestsand moral claims of theirmembers (which can be accommodated within a conception of human rights).This idea smacksof bad metaphysics42and also is dangerouslysubject to political/ ideological manipulation (as exemplifiedby Charles de Gaulle who was fond of adducing the interestsof la nation against those of his French compatriots).Moreover, it is unclear whythis idea should work here, but not in the case of other kinds of (sub- and supranational) political units, nor in that of religious,cultural, and athleticentities. Whyneed we not also take intoaccount the moralclaims of Catholicism, art, or baseball? These problems suggest the other justificatorystrategy, which accepts the individualistpremise but then formulatesthe political rightsof persons with essential reference to the state/nationwhose membersthey are. This strategyhas been defended,most prominently, by Michael Walzer, albeit in a treatisethat focuses on international ethics(interactions) rather than international justice (institutions).Walzer approvingly quotes Westlake: "The duties and rights of states are nothing more than the duties and rightsof the men who compose them," adding "the rights ... [to] territorialintegrity and political sovereignty. . . belong to states,but they derive ultimatelyfrom the rightsof individuals,and fromthem they take theirforce.... States are neitherorganic wholes nor mysticalunions."43 The keyquestion is, of course, how such a derivationis supposed to work. There are two possibilities.The direct route would be to postulate either a human rightto be governed by one's compatriots/ fellow nationals44or a human rightto participatein the exercise of sovereigntyover one's compatriots/fellownationals. The formerof these rights is implausibly demanding upon others (the Bavarians could insiston being part of Germany,even if all the other Germans wanted nothing to do with them) and would still fail to establish b, unless it were also unwaivable-a duty,really. The latterright is im- plausiblydemanding upon those obligated to continue to abide by the common will merelybecause theyhave once (howeverviolently) been

42. Rawls makes this point: "We want to account for the social values, for the intrinsicgood of institutional,community, and associativeactivities, by a conception of justice that in its theoreticalbasis is individualistic.For reasons of clarityamong others, we do not want to . . . suppose that societyis an organic whole with a life of its own distinctfrom and superior to thatof all its membersin theirrelations with one another" (p. 264). 43. Michael Walzer,Justand Unjust Wars(New York: Basic, 1977), p. 53; cf.Walzer, "The Moral Standing of States," in Beitz et al., eds., p. 219. 44. Walzer suggests this tack: "Citizens of a sovereign state have a right,insofar as theyare to be ravaged and coerced at all, to sufferonly at one another'shand" (Wars, p. 86).

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions Pogge Cosmopolitanism 75 incorporated into a state or merely because they have once shared solidarityand sacrifices. The indirect,instrumental route would involvethe empiricalclaim thathuman rights(on a noneccentricdefinition) are more likelyto be satisfied,or are satisfiedto a greaterextent, if there is, foreach person, one political unit that decisivelyshapes her life and is dominated by her compatriots/fellownationals. This route remainsopen on mycos- mopolitanconception (via the thirdconsideration), though the relevant empirical claim would not seem to be sustainable on the historical record. Supposing that this sort of argumentfails on empiricalgrounds, my institutionalcosmopolitanism would favora global order in which sovereigntyis widelydistributed vertically, while the geographicalshape of political units is determined by the autonomous preferencesof situated individuals in accordance with principles 1 and 2.

This content downloaded on Tue, 5 Mar 2013 13:01:55 PM All use subject to JSTOR Terms and Conditions