The Promise of Constructivism in Theory Author(s): Ted Hopf Reviewed work(s): Source: International Security, Vol. 23, No. 1 (Summer, 1998), pp. 171-200 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/2539267 . Accessed: 03/01/2012 16:10

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 MIT Press is collaborating with JSTOR to digitize, preserve and extend access to International Security.

http://www.jstor.org The Promiseof TedHopf Constructivismin InternationalRelations Theory

A challenger to the continuing dominance of neorealism and neoliberal institutionalismin the study of internationalrelations in the United States,constructivism is regarded with a greatdeal of skepticismby mainstreamscholars.1 While the reasons for this receptionare many,three central ones are the mainstream'smiscasting of constructivismas necessarilypostmodern and antipositivist;constructivism's own ambivalence about whether it can buy into mainstream social science methods without sacrificingits theoreticaldistinctiveness; and, related to this ambivalence, constructivism'sfailure to advance an alternativeresearch pro- gram. In this article,I clarifyconstructivism's claims, outline the differences between "conventional" and "critical"constructivism, and suggest a research agenda thatboth provides alternativeunderstandings of mainstreaminterna-

Ted Hopf is VisitingProfessor of Peace Research,The MershonCenter, Ohio State University.He is the authorof Peripheral Visions: DeterrenceTheory and American Foreign Policy in the Third World, 1965-1990 (Ann Arbor:University of Michigan Press, 1994) and is at workon ConstructingForeign Policy at Home: Moscow 1955-1999,in whicha theoryof identity and internationalrelations is developed and tested.He can be reachedby e-mailat <>.

I am most gratefulto Matt Evangelista and Peter Katzensteinwho both read and commentedon many less-than-inspiringdrafts of this work,and, more important,supported my overall research agenda. I am also thankfulto Peter Kowert and Nicholas Onuf for invitingme to Miami in the winter of 1997 to a conferenceat Florida InternationalUniversity at which I was compelled to come to grips with the differencebetween critical and conventional constructivisms.I also benefitedfrom especially incisive and criticalcomments from Henrikki Heikka, Badredine Arfi, RobertKeohane, JamesRichter, Maria Fanis, Ned Lebow, Pradeep Chhibber,Richard Herrmann, David Dessler, and one anonymous reviewer. I would also like to salute the members of my graduate seminarin internationalrelations theory at the Universityof Michigan,in particular,Irfan Nooruddin, Frank Penirian,Todd Allee, and JonathanCanedo helped me figureout the relation- ship between the mainstreamand its critics.

1. The canonical neorealistwork remains Kenneth N. Waltz, Theoryof International Politics (Read- ing, Mass.: Addison-Wesley,1979). The debate between neorealismand neoliberalinstitutionalism is presented and summarized in David A. Baldwin, ed., Neorealismand Neoliberalism(New York: Columbia UniversityPress, 1993). Constructivistchallenges can be found in Nicholas Greenwood Onuf, Worldof Our Making:Rules and Rule in Social Theoryand InternationalRelations (Columbia: Universityof South Carolina Press, 1989); PeterJ. Katzenstein, ed., The Cultureof National Security: Normsand Identityin WorldPolitics (New York:Columbia UniversityPress, 1996); and Yosef Lapid and FriedrichV. Kratochwil,eds., The Returnof Ctultureand Identityin IR Theory(Boulder, Colo.: Lynne Rienner,1996).

IhnternationalSecurity, Vol. 23, No. 1 (Summer 1998), pp. 171-200 ? 1998 by the Presidentand Fellows of Harvard College and the Massachusetts Instituteof Technology.

171 InternationalSecurity 23:1 | 172

tional relationspuzzles and offersa few examples of what constructivismcan uniquely bring to an understandingof world politics. Constructivismoffers alternative understandings of a number of the central themesin internationalrelations theory, including: the meaning of anarchyand balance of power, the relationship between state identityand interest,an elaborationof power, and the prospectsfor change in world politics.Construc- tivism itselfshould be understood in its conventionaland criticalvariants, the latterbeing more closely tied to criticalsocial theory.The conventional con- structivistdesire to presentan alternativeto mainstreaminternational relations theory requires a research program. Such a program includes constructivist reconceptualizationsof balance-of-threattheory, the securitydilemma, neolib- eral cooperation theory,and the democraticpeace. The constructivistresearch program has its own puzzles that concentrateon issues of identityin world politics and the theorizationof domestic politics and culture in international relationstheory.

ConventionalConstructivism and Issuesin Mainstream InternationalRelations Theory

Since constructivismis best defined in relation to the issues it claims to apprehend, I presentits position on several of the most significantthemes in internationalrelations theory today.

ACTORS AND STRUCTURES ARE MUTUALLY CONSTITUTED How much do structuresconstrain and enable the actions of actors,and how much can actors deviate fromthe constraintsof structure?In world politics,a structureis a set of relativelyunchangeable constraintson the behavior of states.2Although these constraintscan take the form of systems of material dis/incentives,such as a balance of power or a market,as importantfrom a constructivistperspective is how an action does or does not reproduce both the actor and the structure.3For example, to the extentthat U.S. appeasement in Vietnam was unimaginable because of U.S. identity as a great power,

2. Most importantfor this article,this is the neorealistconceptualization of internationalstructure. All referencesto neorealism,unless otherwisenoted, are fromWaltz, Theory of International Politics. 3. FriedrichKratochwil suggests that this differencein the understandingof structureis because structuralismentered internationalrelations theory not through sociolinguistics,but through microeconomics.Friedrich V. Kratochwil,"Is the Ship of Culture at Sea or Returning?"in Lapid and Kratochwil,The Returnof Cultureand Identity,p. 211. ThePromise of Constructivism | 173

militaryintervention constituted the United States as a greatpower. Appease- ment was an unimaginable act. By engaging in the "enabled" action of inter- vention,the United States reproduced its own identityof great power, as well as the structurethat gave meaning to its action. So, U.S. interventionin Vietnam perpetuated the internationalintersubjective understanding of great powers as those states that use militarypower against others. Meaningful behavior, or action,4is possible only within an intersubjective social context. Actors develop their relations with, and understandings of, others throughthe media of norms and practices. In the absence of norms, exercises of power, or actions, would be devoid of meaning. Constitutive norms define an identityby specifyingthe actions that will cause Others to recognize that identityand respond to it appropriately.5Since structureis meaninglesswithout some intersubjectiveset of norms and practices,anarchy, mainstreaminternational relations theory's most crucial structuralcomponent, is meaningless.Neither anarchy, that is, the absence of any authorityabove the state,nor the distributionof capabilities,can "socialize" statesto the desiderata of the internationalsystem's structureabsent some set of meaningfulnorms and practices.6 A storymany use in first-yearinternational relations courses to demonstrate the structuralextreme, that is, a situation where no agency is imaginable, illustratesthe point. The scenario is a firein a theaterwhere all run for the exits.7But absent knowledge of social practices or constitutivenorms, struc- ture,even in this seemingly overdeterminedcircumstance, is still indetermi- nate. Even in a theaterwith just one door, while all run forthat exit, who goes first?Are they the strongestor the disabled, the women or the children,the aged or the infirm,or is it just a mad dash? Determiningthe outcome will require knowing more about the situationthan about the distributionof ma- terial power or the structureof authority.One will need to know about the culture,norms, institutions, procedures, rules, and social practicesthat consti- tute the actors and the structurealike.

4. The criticaldistinction between action and behavior is made by Charles Taylor,"Interpretation and the Sciences of Man," in Paul Rabinow and WilliamM. Sullivan,eds., InterpretiveSocial Science: A SecondLook (Berkeley: University of CaliforniaPress, 1987), pp. 33-81. 5. Ronald L. Jepperson,,and PeterJ. Katzenstein, "Norms, Identity,and Culture in National Security,"in Katzenstein,The Cultureof National Security, p. 54. 6. David Dessler, "What's At Stake in the Agent-StructureDebate?" InternationalOrganization, Vol. 43, No. 3 (Summer 1989), pp. 459-460. 7. Arnold Wolfers,Discord and Collaboration(Baltimore, Md.: Johns Hopkins UniversityPress, 1962). InternationalSecurity 23:1 | 174

ANARCHY AS AN IMAGINED COMMUNITY Given that anarchy is structural,it must be mutually constitutedby actors employing constitutiverules and social practices,implying that anarchyis as indeterminateas Arnold Wolfers'sfire. Alexander Wendt has offereda con- structivistcritique of this fundamentalstructural pillar of mainstreaminterna- tional relations theory.8But still more fundamentally,this move opens the possibilityof thinkingof anarchy as having multiple meanings for different actors based on theirown communitiesof intersubjectiveunderstandings and practices.And ifmultiple understandings of anarchyare possible, thenone can begin to theorize about differentdomains and issue areas of international politics that are understood by actors as more, or less, anarchic. Self-help,the neorealistinference that all states should prefersecurity inde- pendence whenever possible, is a structurallydetermined behavior of an actor only to the extentthat a single particularunderstanding of anarchyprevails.9 If the implicationsof anarchyare not constantacross all relationshipsand issue areas of internationalpolitics, then a continuumof anarchiesis possible. Where there are catastrophicconsequences for not being able to rely on one's own capacity to enforcean agreement,such as arms controlin a world of offensive militaryadvantage, neorealistconceptualizations of anarchyare most apt. But where actors do not worry much about the potential costs of ceding control over outcomes to other states or institutions,such as in the enforcementof trade agreements,this is a realm of world politics where neorealist ideas of anarchyare just imaginary.

IDENTITIES AND INTERESTS IN WORLD POLITICS Identitiesare necessary,in internationalpolitics and domestic societyalike, in order to ensure at least some minimal level of predictabilityand order.'0 Durable expectationsbetween states require intersubjectiveidentities that are sufficientlystable to ensure predictablepatterns of behavior.A world without

8. Alexander Wendt, "Anarchy Is What States Make of It: The Social Constructionof Power Politics,"International Organization, Vol. 46, No. 2 (Spring 1992), 391-425. 9. Elizabeth Kier, for example, shows how the same "objective" externalstructural arrangement of power cannot account forFrench military strategy between the two world wars. Elizabeth Kier, "Culture and FrenchMilitary Doctrine before World War II," in Katzenstein,The Cultureof National Security,pp. 186-215. 10. The focus on identitydoes not reflecta lack of appreciationfor other elements in the construc- tivistapproach, such as norms,culture, and institutions.Insofar as identitiesare the most proxi- mate causes of choices,preferences, and action,I concentrateon them,but withthe fullrecognition that identitiescannot be understood withouta simultaneous account of normative,cultural, and institutionalcontext. ThePromise of Constructivism | 175

identitiesis a world of chaos, a world of pervasive and irremediableuncer- tainty,a world much more dangerous than anarchy.Identities perform three necessaryfunctions in a society:they tell you and otherswho you are and they tell you who othersare.11 In tellingyou who you are, identitiesstrongly imply a particularset of interestsor preferenceswith respect to choices of action in particulardomains, and with respect to particularactors. The identityof a state implies its preferencesand consequent actions.12 A state understands othersaccording to the identityit attributesto them,while simultaneouslyreproducing its own identitythrough daily social practice.The crucial observationhere is thatthe producer of the identityis not in controlof what it ultimatelymeans to others; the intersubjectivestructure is the final arbiterof meaning. For example, during the Cold War,Yugoslavia and other East European countriesoften understood the Soviet Union as Russia, despite the factthat the Soviet Union was tryinghard not to have thatidentity. Soviet control over its own identitywas structurallyconstrained not only by East European understanding,but also by daily Soviet practice,which of course included conversingwith East Europeans in Russian. Whereas constructivismtreats identity as an empirical question to be theo- rized within a historicalcontext, neorealism assumes that all units in global politics have only one meaningfulidentity, that of self-interestedstates. Con- structivismstresses that this proposition exempts fromtheorization the very

11. Henri Tajfel,Human Groupsand Social Categories:Studies in Social Psychology(Cambridge, U.K.: Cambridge UniversityPress, 1981), p. 255. Although there are many accounts of the origin of identity,I offera cognitiveexplanation because it has minimala prioriexpectations, assuming only that identitiesare needed to reduce complexityto some manageable level. 12. Dana Eyre and Mark Suchman, for example, findthat, controlling for rational strategic need, domestic coalition politics, and superpower manipulation, countries in the third world prefer certain weapons systems over others because of their understanding of what it means to be "modern" in the twentiethcentury. Dana P. Eyre and Mark C. Suchman, "Status,Norms, and the Proliferationof Conventional Weapons: An InstitutionalTheory Approach," in Katzenstein,The Cultureof National Security,pp. 73-113. Other examples of empirical research that have linked particularidentities to particularsets of preferencesare "civilized" identitiesdriving attitudes toward weapons of mass destruction;notions of what constitutes"humanitarian" shaping deci- sions to intervenein otherstates; the identityof a "normal" stateimplying particular Soviet foreign policies; and "antimilitarist"identities in Japan and German shaping their post-World War II foreignpolicies. These argumentscan be found in Richard Price and Nina Tannenwald, "Norms and Deterrence:The Nuclear and Chemical Weapons Taboos," pp. 114-152; Martha Finnemore, "ConstructingNorms of Humanitarian Intervention,"pp. 153-185; Robert Herman, "Identity, Norms,and National Security:The Soviet ForeignPolicy Revolutionand the End of the Cold War," pp. 271-316; and Thomas U. Berger,"Norms, Identity,and National Securityin Germany and Japan,"pp. 317-356. All of the above are in Katzenstein,The Culture of National Security. On identity and mutual intelligibility,see Roxanne Lynn Doty, "The Bounds of 'Race' in InternationalRela- tions,"Millennium: Journal of International Studies, Vol. 22, No. 3 (Winter1993), p. 454. InternationalSecurity 23:1 | 176

fundamentalsof internationalpolitical life, the nature and definitionof the actors. The neorealistassumption of self-interestpresumes to know, a priori, just what is the self being identified.In otherwords, the state in international politics,across time and space, is assumed to have a single eternalmeaning. Constructivisminstead assumes that the selves, or identities,of states are a variable; theylikely depend on historical,cultural, political, and social context. Constructivismand neorealism share the assumption that interestsimply choices, but neorealism furtherassumes that states have the same a priori interests.Such a homogenizing assumption is possible only if one denies that interestsare the productsof the social practicesthat mutually constitute actors and structures.13Given thatinterests are the productof identity,that is, having the identity"great power" implies a particularset of interestsdifferent from those implied by the identity"European Union member,"and that identities are multiple,constructivist logic precludes acceptance of pregiven interests.14 By making interestsa centralvariable, constructivism explores not only how particular interestscome to be, but also why many interestsdo not. The tautological,and thereforealso true,most common,and unsatisfyingexplana- tion is that interestsare absent where there is no reason for them, where promised gains are too meager. Constructivism,instead, theorizes about the meaning of absent interests.Just as identities and interestsare produced throughsocial practices,missing interests are understood by constructivistsas produced absences, omissions that are the understandable product of social practicesand structure.The social practicesthat constitutean identitycannot imply intereststhat are not consistentwith the practices and structurethat constitutethat identity. At the extreme,an actor would not be able to imagine an absent interest,even if presentedwith it.15

13. calls the failure to contextualize interestsone of the major weaknesses of mainstreaminternational relations theory.Robert 0. Keohane, "InternationalInstitutions: Two Approaches," InternationalStudies Quarterly, Vol. 32, No. 4 (December 1988), pp. 390-391. 14. JeffreyLegro, forexample, has shown how the preferencesof greatpowers beforeand during WorldWar II withrespect to the use and nonuse of strategicbombing, and chemicaland submarine warfare,are unfathomablewithout first understanding the identitiesof the militaryorganizations responsible for shaping those preferences.Jeffrey W. Legro, "Culture and Preferencesin the InternationalCooperation Two-Step,"American Political Science Review, Vol. 90, No. 1 (March 1996), pp. 118-137. 15. See, for example, Tannenwald, "Norms and Deterrence," and Kier, "Culture and French MilitaryDoctrine before World War II," p. 203. For a brilliantaccount of how social structure enables and impedes the constructionof identityand interest,see JaneK. Cowan, "Going Out for Coffee?Contesting the Grounds of Gendered Pleasures in Everyday Sociability,"in Peter Loizos and Evthymios Papataxiarchis, eds., ContestedIdentities: Gender and Kinship in Modern Greece (Princeton,N.J.: Princeton University Press, 1991), pp. 196-197. The Promiseof Constructivism | 177

The consequences of this treatmentof interestsand identitieswork in the same directionas constructivism'saccount of structure,agency, and anarchy: states are expected to have (1) a far wider array of potentialchoices of action before them than is assumed by neorealism, and (2) these choices will be constrainedby social structuresthat are mutuallycreated by states and struc- tures via social practices. In other words, states have more agency under constructivism,but that agency is not in any sense unconstrained. To the contrary,choices are rigorouslyconstrained by the webs of understandingof the practices,identities, and interestsof otheractors that prevail in particular historicalcontexts.

THE POWER OF PRACTICE Power is a centraltheoretical element forboth mainstreamand constructivist approaches to internationalrelations theory,but their conceptualizations of power are vastly different.Neorealism and neoliberal institutionalismassume that materialpower, whethermilitary or economic or both, is the single most importantsource of influenceand authorityin global politics.16 Constructivism argues that both materialand discursive power are necessary for any under- standing of world affairs.I emphasize both because oftenconstructivists are dismissed as unRealistic for believing in the power of knowledge, ideas, culture,ideology, and language, that is, discourse.17 The notion that ideas are a formof power, that power is more than brute force,and that material and discursive power are related is not new. Michel Foucault's articulationof the power/knowledge nexus, Antonio Gramsci's theoryof ideological hegemony, and Max Weber's differentiationof coercion fromauthority are all precursors to constructivism'sposition on power in political life.18Empirical work exists

16. A rare effortin the mainstreamliterature to break away fromthis focus on materialpower is JudithGoldstein and Robert 0. Keohane, eds., Ideas and ForeignPolicy (Ithaca, N.Y.: Cornell UniversityPress, 1993). 17. As R.B.J.Walker has clarified,"To suggest thatculture and ideology are crucial forthe analysis of world politicsis not necessarilyto take an idealist position.... On the contrary,it is important to recognizethat ideas, consciousness,culture, and ideology are bound up with more immediately visible kinds of political,military, and economic power." In R.B.J.Walker, "East Wind, West Wind: Civilizations,Hegemonies, and World Orders," in Walker,ed., Culture,Ideology, and WorldOrder (Boulder, Colo.: Westview Press, 1984), p. 3. See also Onuf, Worldof Our Making,p. 64. 's conceptualizationof "soft" power could be usefullyread througha constructivistinterpre- tation. See Joseph S. Nye, Jr.,Bound to Lead: The ChangingNature of AmericanPower (New York: Basic Books, 1991), esp. pp. 173-201. 18. Colin Gordon,ed., Power/Knowledge:Selected Interviews aned Other Writinigs, 1972-1997, by Michel Foucault(Brighton, Sussex, U.K.: HarvesterPress, 1980); AntonioGramsci, Selectionis from the Prison Notebooks,trans. and ed., Quinton Hoare and GeoffreyNowell Smith (New York: International InternationalSecurity 23:1 | 178

in both internationalrelations theoryand securitystudies that demonstrates the need to appreciateboth the materialand the discursiveaspects of power.19 Given that the operation of the material side of power is familiarfrom the mainstreamliterature, here I concentrateon the discursive side, the power of practicein constructivism. The power of social practiceslies in theircapacity to reproducethe intersub- jective meanings that constitutesocial structuresand actors alike. The U.S. militaryintervention in Vietnamwas consistentwith a number of U.S. identi- ties: great power, imperialist,enemy, ally, and so on. Others observing the United States not only inferredU.S. identityfrom its actions in Vietnam,but also reproduced the intersubjectiveweb of meaning about what precisely constitutedthat identity.To the extent,for example, that a group of countries attributedan imperialistidentity to the United States,the meaning of being an imperialiststate was reproduced by the U.S. militaryintervention. In thisway, social practicesnot only reproduce actors throughidentity, but also reproduce an intersubjectivesocial structurethrough social practice. A most important power of practiceis its capacity to produce predictabilityand so, order.Social practicesgreatly reduce uncertaintyamong actors withina socially structured community,thereby increasing confidence that what actions one takes will be followed by certainconsequences and responses fromothers.20 An actor is not even able to act as its identityuntil the relevantcommunity of meaning,to paraphrase Karl Deutsch,21acknowledgesthe legitimacyof that

Publishers,1992); and Max Weber,From Max Weber,ed., Hans Gerth and C. WrightMills (New York: OxfordUniversity Press, 1946). 19. Price and Tannenwald show that even power as material as nuclear missiles and chemical artilleryhad to be understood and interpretedbefore it had any meaning. In Price and Tannen- wald, "Norms and Deterrence."Robert Cox has provided an account of the rise,reproduction, and demise of nineteenth-centuryBritish supremacy, and the rise and reproductionof U.S. dominance in the twentiethcentury through a close reading of the interactionbetween materialand discursive power. RobertW. Cox, "Social Forces, States, and World Orders: Beyond InternationalRelations Theory,"Millennium: Journal of International Studies, Vol. 10, No. 1 (Spring 1981), pp. 126-155. 20. Onuf sees these reproduciblepatterns of action as the product of "reflexiveself-regulation," wherebyagents referto theirown and other'spast and anticipatedactions in deciding how to act. Onuf,World of Our Making, p. 62. 21. Karl W. Deutsch, Nationalismand Social Communication:An Inquiry into the Foundationsof Nationality(New York:MIT Press, 1953), pp. 60-80. Deutsch was a constructivistlong ahead of his time to the extentthat he argued that individuals could not engage in meaningfulaction absent some community-wideintersubjectivity. Another work constructivistin essence is RobertJervis's The Logic ofImages in InternationalRelations (Princeton, N.J.: PrincetonUniversity Press, 1970). Applying Erving Goffmann'sself-presentation theory to internationalpolitics, Jervis pointed out that state actions, such as gunboat diplomacy, were meaningless unless situated in a larger intersubjectivecommunity of diplomatic practice. The Promiseof Constructivism | 179

action,by that actor,in thatsocial context.The power of practiceis the power to produce intersubjectivemeaning withina social structure.It is a shortstep fromthis authorizingpower of practice to an understandingof practice as a way of bounding, or disciplininginterpretation, making some interpretations of realityless likely to occur or prevail within a particularcommunity.22 The meanings of actions of members of the community,as well as the actions of Others,become fixed throughpractice; boundaries of understandingbecome well known. In this way, the ultimatepower of practice is to reproduce and police an intersubjectivereality.23 Social practices,to the extentthat they autho- rize, discipline,and police, have the power to reproduce entirecommunities, including the internationalcommunity, as well as the many communitiesof identityfound therein.24 State actions in the foreignpolicy realm are constrainedand empowered by prevailing social practicesat home and abroad. Richard Ashley,for example, writes of a foreignpolicy choice as being a kind of social practicethat at once constitutesand empowers the state,defines its sociallyrecognized competence, and secures the boundaries that differentiatethe domestic and international economic and political spheres of practice and, with them, the appropriate domains in which specificactors may secure recognitionand act competently. Finally,Ashley concludes, foreignpolicy practice depends on the existenceof intersubjective"precedents and shared symbolicmaterials-in orderto impose interpretationsupon events,silence alternativeinterpretations, structure prac- tices,and orchestratethe collectivemaking of history."25 Although I have necessarily concentratedon articulatinghow discursive power works in this section, the power to control intersubjectiveunder- standing is not the only formof power relevantto a constructivistapproach to world politics. Having resources that allow oneself to deploy discursive power-the economic and militarywherewithal to sustain institutionsneces-

22. See Doty,"The Bounds of Race," p. 454; and Carol Cohn, "Sex and Death in the RationalWorld of Defense Intellectuals,"Signs: Journal of Womenin Cultureand Society,Vol. 12, No. 32 (Summer 1987), pp. 687-718. 23. See Richard K. Ashley, "Untying the Sovereign State: A Double Reading of the Anarchy Problematique,"Millennium: Journal of International Studies, Vol. 17, No. 2 (Summer 1988), p. 243, fora discussion of this process. 24. Richard K. Ashley,"The of Geopolitical Space: Toward a CriticalSocial Theory of InternationalPolitics, " Alternatives,Vol. 12, No. 4 (October-December1987), p. 409. 25. RichardK. Ashley,"Foreign Policy as PoliticalPerformance," International Studies Notes (1988), p. 53. InternationalSecurity 23:1 | 180

sary forthe formalizedreproduction of social practices-is almost always part of the storyas well.

CHANGE IN WORLD POLITICS Constructivismis agnostic about change in world politics.26It restoresmuch varietyand differenceto world affairsand points out the practicesby which intersubjectiveorder is maintained,but it does not offerany more hope for change in world politicsthan neorealism.Constructivism's insight that anarchy is what states make of it, for example, implies that there are many different understandingsof anarchyin the world, and so state actions should be more varied than only self-help.But thisis an observationof already-existingreality, or, more precisely,a set of hypothesesabout the same. These differentunder- standings of anarchy are still rooted in social structures,maintained by the power of practice,and quite impervious to change. What constructivismdoes offeris an account of how and where change may occur. One aspect of constructivistpower is the power to reproduce,discipline, and police. When such power is realized, change in world politics is very hard indeed. These intersubjectivestructures, however, although difficultto chal- lenge, are not impregnable.Alternative actors with alternativeidentities, prac- tices, and sufficientmaterial resources are theoreticallycapable of effecting change. RobertCox's account of Britishand Americansupremacy, for example, perhaps best illustratesthe extraordinarystaying power of a well-articulated ideological hegemony,but also its possible demise. And Walker rightlyob- serves that constructivism,to the extentthat it surfaces diversity,difference, and particularity,opens up at least potentialalternatives to the currentprevail- ing structures.27Constructivism conceives of the politics of identityas a con- tinual contestfor controlover the power necessary to produce meaning in a social group. So long as thereis difference,there is a potentialfor change. Thus, contraryto some critics28who assert that constructivismbelieves that change in world politics is easy, that "bad" neorealiststructures need only be thoughtaway, in factconstructivism appreciates the power of structure,if for no other reason then it assumes that actors reproduce daily their own con- straintsthrough ordinary practice. Constructivism'sconceptualization of the

26. Criticalconstructivism denies this vigorously. 27. R.B.J.Walker, "Realism, Change, and InternationalPolitical Theory," International Studies Quar- terly,Vol. 31, No. 1 (March 1987), pp. 76-77. 28. See, for example, John J. Mearsheimer,"The False Promise of InternationalInstitutions," InternationalSecurity, Vol. 19, No. 1 (Winter1994/1995), pp. 5-49, esp. 37-47. The Promiseof Constructivism | 181

relationshipbetween agency and structuregrounds its view thatsocial change is both possible and difficult.Neorealism's position thatall statesare meaning- fullyidentical denies a fairamount of possible change to its theoreticalstruc- ture. In sum, neorealismand constructivismshare fundamentalconcerns with the role of structurein world politics,the effectsof anarchyon state behavior,the definitionof state interests,the natureof power, and the prospectsfor change. They disagree fundamentally,however, on each concern. Contra neorealism, constructivismassumes that actors and structuresmutually constituteeach other;anarchy must be interpretedto have meaning; stateinterests are part of the process of identityconstruction; power is both materialand discursive;and change in world politics is both possible and difficult.

Constructivisms:Conventional and Critical

To the degree that constructivismcreates theoreticaland epistemologicaldis- tance between itselfand its originsin criticaltheory, it becomes "conventional" constructivism.Although constructivismshares many of the foundationalele- ments of criticaltheory, it also resolves some issues by adopting defensible rules of thumb,or conventions,rather than following critical theory all the way up the postmodern critical path.29I situate constructivismin this way to highlightboth its commonalitieswith traditionalinternational relations theory and its differenceswith the criticaltheory with which it is sometimesmislead- ingly conflated.30Below I sketch out the relationshipbetween conventional constructivismand criticalsocial theoryby identifyingboth those aspects of critical theory that constructivismhas retained and those it has chosen to conventionalize. The result, conventional constructivism,is a collection of principles distilled from critical social theorybut without the latter's more consistenttheoretical or epistemologicalfollow-through. Both criticaland con- ventionalconstructivism are on the same side of the barricadesin YosefLapid's characterizationof the battle zone: the fixed, natural, unitary,stable, and

29. Jepperson,Wendt, and Katzensteindifferentiate the kind of "sociological" analysis performed in theirvolume fromthe "radical constructivistposition" of RichardAshley, David Campbell, R.B.J. Walker,and Cynthia Weber. See Jepperson,Wendt, and Katzenstein,"Norms, Identity,and Cul- ture,"p. 46, notes 41 and 42. 30. As, for example, in Mearsheimer,"The False Promise of InternationalInstitutions," wherein constructivism,, postmodernism, and poststructuralismare all reduced to "critical theory,"p. 37, note 128. InternationalSecurity 23:1 | 182

essence-like,on the one (mainstreaminternational relations theory) hand, and the emergent,constructed, contested, interactive, and process-like,on the other (constructivist)one.31 Conventional and criticalconstructivism do share theoreticalfundamentals. Both aim to "denaturalize" the social world, that is, to empiricallydiscover and reveal how the institutionsand practicesand identitiesthat people take as natural,given, or matterof fact,are, in fact,the product of human agency, of social construction.32Both believe thatintersubjective reality and meanings are criticaldata for understandingthe social world.33Both insist that all data must be "contextualized,"that is, theymust be related to, and situatedwithin, the social environmentin which they were gathered,in order to understand their meaning.34Both accept the nexus between power and knowledge, the power of practice in its disciplinary,meaning-producing, mode.35 Both also accept the restorationof agency to human individuals. Finally,both stressthe reflexivityof the self and society,that is, the mutual constitutionof actor and structure.36 Perhaps where constructivismis most conventionalis in the area of meth- odology and epistemology.The authors of the theoreticalintroduction to The Cultureof National Security, for example, vigorously,and perhaps defensively, deny that their authors use "any special interpretivistmethodology."37 The authors are carefulto stressthat they do not depart from"normal science" in this volume, and none of the contributorseither deviates fromthat ground or questions whether it is appropriate.38This position is anathema to critical theorywhich, as part of its constitutiveepistemology, has a lengthybill of particularsagainst positivism.

31. Yosef Lapid, "Culture's Ship: Returnsand Departures in InternationalRelations Theory," in Lapid and Kratochwil,The Returnof Cultureand Identity,pp. 3-20. 32. Mark Hoffman,"Critical Theory and the Inter-ParadigmDebate," Millennium:Journal of Inter- nationalStudies, Vol. 16, No. 2 (Summer 1987), pp. 233-236. 33. Ashley,"The Geopolitics of Geopolitical Space," p. 403. 34. In this respect,both criticaland conventionalconstructivism can be understood as sharingan interpretivistepistemology, more generally.See Taylor,"Interpretation and the Sciences of Man." 35. James Der Derian, On Diplomacy.A Genealogyof WesternEstrangement (Oxford, U.K.: Basil Blackwell, 1987), p. 4. 36. R.B.J.Walker, "World Politics and WesternReason: Universalism,Pluralism, Hegemony," in Walker,Culture, Ideology, and WorldOrder, p. 195; and Ashley, "The Geopolitics of Geopolitical Space," pp. 409-410. 37. Jepperson,Wendt, and Katzenstein,"Norms, Identity,and Culture," p. 67. 38. The only,even partial,exceptions are Price and Tannenwald, "Norms and Deterrence,"and Michael N. Barnett,"Institutions, Roles, and Disorder: The Case of the Arab States System," InternationalStudies Quarterly, Vol. 37, No. 3 (September1993), pp. 271-296. ThePromise of Constructivism | 183

Conventional constructivism,while expectingto uncover differences,iden- tities,and multiple understandings,still assumes that it can specifya set of conditionsunder which one can expect to see one identityor another.This is what Mark Hoffmanhas called "minimal foundationalism,accepting that a contingentuniversalism is possible and may be necessary."In contrast,critical theoryrejects either the possibilityor the desirabilityof a minimal or contin- gentfoundationalism.39 Ashley chides all noncriticalapproaches for"anticipat- ing analysis coming to a close." In allowing for such prematureclosure, the analyst participatesin the normalization or naturalization of what is being observed, and risks hiding the patternsof domination that mightbe revealed if closure could only be deferred.40To reach an intellectuallysatisfying point of closure, constructivismadopts positivistconventions about sample charac- teristics,methods of difference,process tracing,and spuriousness checks. In making this choice, criticaltheorists argue, constructivismcan offeran under- standingof social realitybut cannot criticizethe boundaries of its own under- standing,and this is preciselywhat criticaltheory is all about.41 So, for example, Thomas Bergermakes claims about Japanese and German national identitiesthat imply a certainoutcome foran indefiniteperiod of time to come.42Such a claim requires the presumed nonexistenceof relevantunob- servables,as well as the assumption thatthe practices,institutions, norms, and power relationsthat underlay the productionof those identitiesare somehow fixedor constant.Critical theorists would see thisas an illusion of control;none of these factorscan be so easily immobilized foreither analysis or prediction. This differencemanifests itself as well in how critical and conventional constructivismunderstand identity.Conventional constructivistswish to dis- cover identitiesand their associated reproductivesocial practices,and then offeran account of how those identities imply certain actions. But critical theoristshave a differentaim. They also wish to surface identities,not to articulatetheir effects,but to elaborate on how people come to believe in a

39. Mark Hoffman,"Restructuring, Reconstruction, Reinscription, Rearticulation: Four Voices in Critical InternationalTheory," Millennium: Journal of InternationalStudies, Vol. 20, No. 1 (Spring 1991), p. 170. David Campbell argues that no identity(or any other theoreticalelement for that matter)may be allowed to be fixedor final.It mustbe criticallydeconstructed as soon as it acquires a meaning.David Campbell, "ViolentPerformances: Identity, Sovereignty, Responsibility," in Lapid and Kratochwil,The Returnof Cultureand Identity,pp. 164-166. See also Stephen J.Rosow, "The Forms of Internationalization:Representation of WesternCulture on a Global Scale," Alternatives, Vol. 15, No. 3 (July-September1990), p. 289, fordifferences on this issue. 40. Ashley,"The Geopolitics of Geopolitical Space," p. 408. 41. Hoffman,"Restructuring, Reconstruction, Reinscription, Rearticulation," p. 232. 42. Berger,"Norms, Identity,and National Securityin Germanyand Japan." InternationalSecurity 23:1 | 184

single version of a naturalized truth.In other words, criticaltheory aims at exploding the mythsassociated with identityformation, whereas conventional constructivistswish to treatthose identitiesas possible causes of action.Critical theorythus claims an interestin change, and a capacity to fosterchange, that no conventionalconstructivist could make. In addition,and in a relatedvein, criticaltheorists self-consciously recognize their own participationin the reproduction,constitution, and fixingof the social entitiesthey observe.43 They realize thatthe actorand observercan never be separated. Conventional constructivistsignore thisinjunction, while largely adopting interpretivistunderstandings of the connectivityof subjects with othersubjects in a web of intersubjectivemeaning. The observernever becomes a subject of the same self-reflectivecritical inquiry. Conventional and criticalconstructivists also split over the origins of iden- tity.44Whereas conventionalconstructivists accommodate a cognitiveaccount foridentity, or offerno account at all, criticalconstructivists are more likelyto see some formof alienation drivingthe need foridentity. As remarkedabove, conventional constructivismaccepts the existence of identitiesand wants to understandtheir reproduction and effects,but criticalconstructivists use criti- cal social theory to specify some understanding of the origin of identity. Tzvetan Todorov and Ashis Nandy, forexample, assume thatEuropean iden- titieswere incomplete(indeed, every selfis incompletewithout an other)until theyencountered peoples in the Americas and India, respectively.45The neces- sity of differencewith an other to produce one's own identityis found in Hegel's bondsman's tale, where the more powerful slaveowner can neither know his own identitynor exercisehis superiorpower untilhis slave, his other, helps him constructthat identity through practice. Perhaps conventionalcon- structivismcould accept this assumption: the need for others to construct oneself,but criticalconstructivism moves beyond this position with the aid of Nietzsche, Freud, and Lacan.46The formerallows differenceto reign,whereas

43. Cynthia Weber points this out as a very importantdistinction between her approach to the state and more modernistapproaches. Weber similarlyseparates conventionalconstructivists from criticaltheorists. Max Weber,Simulating Sovereignty: Intervention, the State,and SymbolicExchange (Cambridge,U.K.: Cambridge UniversityPress, 1995), p. 3. 44. For a review of thisissue see FriedrichKratochwil, "Is the Ship of Cultureat Sea or Returning?" pp. 206-210. 45. The discussion of the work of Todorov and Nandy is in Naeem Inayatullah and David L. Blaney,"Knowing Encounters:Beyond Parochialismin InternationalRelations Theory,"in Lapid and Kratochwil,The Returnof Cultureand Identity,pp. 65-84. 46. For an account of identitybased on these three theorists,see Anne Norton, Reflectionson PoliticalIdentity (Baltimore, Md.: JohnsHopkins UniversityPress, 1988). The Promiseof Constructivism | 185

the latterimplies eitherthe assimilation of the other,if deemed equal, or his oppression,if inferior.47 Critical theory'sapproach toward identityis rooted in assumptions about power.48Critical theorists see power being exercisedin every social exchange, and thereis always a dominantactor in thatexchange. Unmasking these power relations is a large part of criticaltheory's substantive agenda; conventional constructivism,on the otherhand, remains "analyticallyneutral" on the issue of power relations.Although conventionalconstructivists share the idea that power is everywhere,because they believe that social practices reproduce underlyingpower relations,they are not necessarilyinterested in interrogating those relations. Critical theory's assumption that all social relations are in- stances of hierarchy,subordination, or domination ironicallyappears similar to the expectationsof realists and neorealists about world politics.49The dif- ferent conceptualizations of power imply differenttheoretical agendas. Whereas conventional constructivismis aimed at the production of new knowledge and insightsbased on novel understandings,"critical theory ana- lyzes social constraintsand cultural understandingsfrom a supreme human interestin enlightenmentand emancipation."50 Although conventionaland criticalconstructivism share a number of posi- tions-mutual constitutionof actors and structures,anarchy as a social con- struct,power as both materialand discursive,and stateidentities and interests as variables-conventional constructivismdoes not accept critical theory's ideas about its own role in producing change and maintainsa fundamentally differentunderstanding of power.51

47. Inayatullah and Blaney,"Knowing Encounters,"pp. 65-66. For a very useful analysis of how differentaccounts of identityhave made theirway throughfeminist theorizing, see Allison Weir, SacrificialLogics: Feminist Theory and theCritique of Identity (New York: Routledge, 1996). 48. My views on the differencesseparating criticaland conventionalconstructivist positions on power were shaped in conversationwith JimRichter. 49. See ArturoEscobar, "Discourse and Power in Development:Michel Foucault and the Relevance of His Work to the Third World," Alternatives,Vol. 10, No. 4 (October-December 1984), esp. pp. 377-378. 50. This is takenfrom Andrew Linklater,"The Question of the Next Stage in InternationalRelations Theory: A Critical-TheoreticalPoint of View," Millennium:Journal of InternationalStudies, Vol. 21, No. 1 (Spring 1992), p. 91, and is based on his interpretationof JurgenHabermas. For a view on preciselythe point of the emancipatorypower of criticaltheory, see Chris Brown,"'Turtles All the Way Down': Anti-Foundationalism,Critical Theory, and InternationalRelations," Millennium: Jour- nal ofInternational Studies, Vol. 23, No. 2 (Summer 1994), p. 219. 51. For an alternativeaccount of internationalrelations theory from a criticaltheory perspective in which conventional constructivism'spositions can be found as well, see Richard K. Ashley, "Three Modes of Economism," InternationalStudies Quarterly,Vol. 27, No. 4 (December 1983), pp. 477-491. On the constructionof anarchy,in particular,see Ashley, "Untying the Sovereign InternationalSecurity 23:1 | 186

A ConstructivistResearch Agenda

This section aims at moving constructivismfrom the margins52by articulating a loosely Lakatosian research program for a constructiviststudy of interna- tional relations.53I presentthis research agenda in threesections. The firststep is to show that constructivismoffers competing understandings of some key puzzles frommainstream international relations theory. The second move is to suggest what new and innovative puzzles constructivismpromises to raise. The last step is forconstructivism to point out its own weaknesses.

MAINSTREAM PUZZLES, CONSTRUCTIVIST SOLUTIONS Constructivismcan provide alternativeaccounts of the balance of threat,secu- ritydilemmas, neoliberal institutionalistaccounts of cooperation under anar- chy,and the liberal theoryof the democraticpeace. BALANCE OF THREAT. Neorealism tells us that states ally against power. Steven Walt rightlyobserved that this is empiricallywrong. He suggested, instead, that states ally against threats.The attemptedfix was to claim that states will balance, not against power, but against particularkinds of power. The latter is the power possessed by a relatively capable, geographically proximate state with offensivemilitary capabilities and perceived hostile in- tentions.54Whereas geographicalproximity and offensivemilitary capacity can be established a priori,perceived intentionsthreaten tautology. Several con- structivistscholars have pointed to balance of threatas one of the mainstream

State," p. 253. In addition, conventionalconstructivism is more willing to accept the ontological status of the state when theorizing,whereas criticaltheory demands that the state remain a zone of contestation,and should be understood as such; its autonomous existence should not be accepted. For the formerconventional view, see Alexander Wendt, "ConstructingInternational Politics," InternationalSecurity, Vol. 20, No. 1 (Summer 1995), p. 72. For the criticalview of the state,see Ashley,"Untying the Sovereign State," pp. 248-251. 52. For the challenge to constructiviststo develop a research program or be marginalized,see Keohane, "InternationalInstitutions," p. 392. For criticismin a similar vein, see Thomas J.Bier- steker,"Critical Reflectionson Post-Positivismin InternationalRelations," InternationalStudies Quarterly,Vol. 33, No. 3 (September1989), p. 266. 53. It is a loose adaptation because, while I adopt Lakatosian criteriafor what constitutesa progressiveand degenerativeshift in a researchprogram, I do not adopt his standards of falsifica- tionismor theirassociated "protectivebelts" of auxiliaryhypotheses. See Imre Lakatos, "Falsifica- tion and the Methodology of ScientificResearch Programmes," in Imre Lakatos and Alan Musgrave, eds., Criticismand the Growthof Knowledge(Cambridge, U.K.: Cambridge University Press, 1970), pp. 91-196. 54. Stephen M. Waltz, The Originsof Alliances (Ithaca, N.Y: Cornell UniversityPress, 1987,p. 5. By acknowledging that "one cannot determine a priori . . . which sources of threatwill be most importantin any given case; one can say only that all of them are likely to play a role," Waltz does not offera nontautologicalmeans forspecifying threat. Quotation on p. 26. The Promiseof Constructivism | 187

accounts most susceptibleto a constructivistalternative.55 What is missinghere is a theory of threatperception, and this is precisely what a constructivist account of identityoffers. Distributionof power cannot explain the alliance patternsthat emerged after World War II; otherwise,the United States would have been balanced against, not the Soviet Union. Instead,the issue must be how France,Britain, Germany, and the United States came to understand Soviet militarycapabilities and geographical proximityas threatening.The neorealistaccount would be that the Soviet Union demonstratedby its behavior that it was an objective threat to WesternEurope. A constructivistaccount would be that the state identities of WesternEurope, the United States, and the Soviet Union, each rooted in domesticsociocultural milieus, produced understandingsof one anotherbased on differencesin identityand practice. The potential advantage of this ap- proach is that it is more likely to surfacedifferences in how the Soviet threat was constructedin differentsites than is the neorealist approach, which ac- cords objectivemeaning to Soviet conduct. Let us imagine, for example, that the United States balanced against the Soviet Union because of the latter'scommunist identity, and what that meant to the United States. If true,it means thatother possible Soviet identities,such as an Asian, Stalinist,Russian, or authoritarianthreat, were not operative. So what? First,how the United Statesunderstood the Soviet threat,as communist, not only explains the anticommunistdirection of U.S. actions in the Cold War, but it also tellsus thatthe United Statesunderstood itself as the anticommunist protectorof a particularset of values both at home and abroad. Second, how the United States constructedthe Soviet communistthreat needs to be under- stood in relation to how WesternEuropeans understood that threat.If, for example, France understood the Soviet threatas a Russian threat,as an in- stance of superior Russian power in Europe, then France would not readily join in U.S. anticommunistventures against the Soviet Union. In particular, whereas the United States saw the thirdworld duringthe Cold War as an arena forbattling communism, as in Vietnam,Europeans very rarelyunderstood it in those terms,instead regardingthird world states as economic actors or as formercolonies.

55. See Thomas Risse-Kappen, "Collective Identityin a Democratic Community: The Case of NATO," in Katzenstein,The Cultureof NationalSecurity, pp. 361-368; Barnett,"Identity and Alli- ances," pp. 401-404; Peter J. Katzenstein, "Introduction:Alternative Perspectives on National Security,"in Katzenstein,The Cultureof National Security, pp. 27-28; Jepperson,Wendt, and Katzen- stein,"Norms, Identity,and Culture,p. 63; and Wendt,"Constructing International Politics," p. 78. InternationalSecurity 23:1 | 188

SECURITY DILEMMAS. Securitydilemmas are the products of presumed un- certainty.56They are assumed to be commonplace in world politics because states presumably cannot know, with sufficientcertainty or confidence,the intentionsof others. But as importantas the securitydilemma is to under- standing conflictualrelations among states,we do not see much evidence of securitydilemmas among many pairs or groups of states:members of the same alliance, members of the same economic institution,perhaps two peaceful states or two neutral states,and so on. In the study of world politics,uncer- taintymight be best treatedas a variable, not a constant.Constructivism can provide an understanding of what happens most of the time in relations between states, namely, nothing threateningat all. By providing meaning, identitiesreduce uncertainty.57 States understand differentstates differently.Soviet and French nuclear capabilities had differentmeanings for Britishdecision makers. But of course certaintyis not always a source of security.Knowing that another state is an aggressorresolves the securitydilemma, but only by replacingit with certain insecurity,an increased confidencethat the other state is in fact threatening. As Richard Ashley,bowing generouslyto Karl Deutsch, pointed out, politics itselfis impossible in the absence of "a backgroundof mutual understandings and habitual practices that orients and limits the mutual comprehensionof practices,the significationof social action."58Constructivism's empirical mis- sion is to surface the "background" that makes uncertaintya variable to understand,rather than a constantto assume. NEOLIBERAL COOPERATION. Neoliberalism offers compelling arguments about how states can achieve cooperation among themselves.Simple iterative interactionamong states,even when they preferto exploit one another,may stilllead to cooperativeoutcomes. The conditionsminimally necessary for such outcomes include transparencyof action,capacity to monitorany noncoopera- tive behavior and punish the same in a predictablefashion, a sufficientlylow discount (high appreciation)rate forfuture gains fromthe relationship,and an expectationthat the relationshipwill not end in the foreseeablefuture.59

56. RobertJervis, "Cooperation under the SecurityDilemma," WorldPolitics, Vol. 30, No. 2 (March 1978), pp. 167-214. 57. I thankMaria Fanis forbringing home to me the importanceof thinkingabout world politics in this way. 58. Ashley,"Three Modes," p. 478; see also Ashley,"The Geopoliticsof GeopoliticalSpace," p. 414. 59. Kenneth A. Oye, "Explaining Cooperation under Anarchy: Hypotheses and Strategies,"in KennethA. Oye, ed., Cooperationunder Anarchy (Princeton, N.J.: Princeton University Press, 1986), pp. 1-24. The Promiseof Constructivism | 189

Internationalinstitutions, whether in the formof regimes,laws, treaties,or organizations, help provide these necessary conditions for cooperation. By having rules about what constitutesa violation of a relationship,institutions help increase the confidenceof each state thatit will not be exploited and that its own cooperative move will be reciprocated.By establishingformal mecha- nisms of surveillance,institutions enable states to see what other states are doing, again enhancing confidencethat a defectionwill be seen and a coop- erative action will be followed by the same. By creatingrules and procedures for surveillance and sanction, all parties can have greater confidence that violations will be punished. By formalizingthese relationships,institutions help reduce each state's discount rate for futuregains while increasingeach state's expectationthat the relationshipwill continueinto the future.60 Constructivismshares neoliberalism'sconclusion that cooperation is possible under anarchy,but offersa very differentaccount of how that outcome emerges. Robert Keohane presents as the heart of neoliberalism two funda- mental assumptions: thereare potentiallybeneficial agreements among states that have not been reached, and they are hard to achieve.61A constructivist approach mightbegin by investigatinghow states understand theirinterests withina particularissue area. The distributionof identitiesand interestsof the relevant states would then help account for whethercooperation is possible. The assumption of exogenous interestsis an obstacle to developing a theory of cooperation. Sittingdown to negotiatea trade agreementamong friends(as opposed to adversaries or unknowns) affectsa state's willingnessto lead with a coopera- tive move. Perhaps it would no longerunderstand its interestsas the unilateral exploitationof the otherstate. Instead it mightsee itselfas a partnerin pursuit of some value otherthan narrow strategicinterest. In Logicof Collective Action, Mancur Olson bracketeda host of situationswhere cooperationwas relatively easy,despite large numbersof players,the absence of a group large enough to provide a public good, but sufficientlysmall to avert coordinationproblems (a k-group),no hegemonic leadership,and so on. These were situationswhere communitiesof identityexisted such that the players were not in a noncoop- erativegame in the firstplace. Too littleattention has been paid to thisinsight.

60. The regimes literatureis vast. For an early foundational volume that includes theoretical specification,empirical illustration, and some self-critique,see Stephen D. Krasner,ed., International Regimes(Ithaca, N.Y.: Cornell UniversityPress, 1983). Elaboration of the marketfailure logic is in Robert0. Keohane, AfterHegemony (Princeton, N.J.: Princeton University Press, 1984). 61. Keohane, "InternationalInstitutions," p. 386. InternationalSecurity 23:1 | 190

A constructivistaccount of cooperationwould reconstructsuch intersubjective communitiesas a matterof course. When a neoliberalwrites of difficultyin reachingan agreement,she usually has one particular problem in mind: uncertainty.Many of the institutional mechanismsdescribed above are aimed at reducinguncertainty among states: provision of transparency;facilitation of iteration;enabling of decomposition; and of course the developmentof rules,monitoring capabilities, and adjudica- tion procedures.A constructivistwould agree thatthese are all veryimportant, but that a prior issue must be raised: Is it not likelythat the level of certainty is a variable associated with identityand practice,and that,ceteris paribus, the less certaintyone has, the more institutionaldevices are necessaryto produce cooperation,the harderthat cooperation will be to achieve, and the more likely it will be to break down? Neoliberalismhas concluded thatan importantpart of ensuringcompliance with agreementsis the development of reputationsfor reliability.62 One of the most importantcomponents of discursive power is the capacity to reproduce order and predictabilityin understandingsand expectations.In this respect, identitiesare a congealed reputation,that is, the closest one can get in social life to being able to confidentlyexpect the same actions fromanother actor time aftertime. Identitiessubsume reputation;being a particularidentity is sufficientto provide necessary diagnostic informationabout a state's likely actions with respect to other states in particulardomains.63 On the other side of the life cycle, neoliberals argue that institutionsdie when membersno longer "have incentivesto maintainthem."64 But one of the more enduringpuzzles forneoliberals is why these institutionspersist past the

62. On the criticalimportance of a theoryof reputationto account foreconomic transactions,such as contracts,see David M. Kreps, "Corporate Culture and Economic Theory,"in JamesE. Alt and Kenneth A. Shepsle, eds., Perspectiveson PositivePolitical Economy (Cambridge, U.K.: Cambridge UniversityPress, 1990), pp. 90-143. Formal game-theoreticwork on reputationconsistently shows thatit should matter,and it does, but only when assumed to do so. Empiricalwork in international relationshas shown thatreputations do not work as hypothesizedby most internationalrelations theory.See JonathanMercer, Reputation and InternationalPolitics (Ithaca, N.Y.: Cornell University Press, 1996); Ted Hopf, PeripheralVisions: Deterrence Theory and AmericanForeign Policy in theThird World,1965-1990 (Ann Arbor: Universityof Michigan Press, 1994); Richard Ned Lebow, Between Peace and War: The Natureof International Crisis (Baltimore,Md.: JohnsHopkins UniversityPress, 1981); and Jervis,Logic of Images in InternationalRelations. 63. For a recognitionthat "shared focal points," a la Thomas Schelling,have much in common with intersubjectivereality and its capacity to promote cooperative solutions to iterativegames, see GeoffreyGarrett and BarryR. Weingast,"Ideas, Interests,and Institutions:Constructing the European Community's Internal Market," in Goldstein and Keohane, Ideas and ForeignPolicy, pp. 173-206. 64. Keohane, "InternationalInstitutions," p. 387. ThePromise of Constructivism I 191

point that great powers have an apparent interestin sustaining them. Their answers include lags caused by domesticpolitical resistance to adjustment,the stickinessof institutionalarrangements, and the transactioncosts entailed in the renegotiationof agreementsand the establishmentof a new order.65An alternativeconstructivist hypothesis would be that if the identitiesbeing re- produced by the social practicesconstituting that institution have gone beyond the strategicgame-playing self-regardingunits posited by neoliberals, and have developed an understandingof each otheras partnersin some common enterprise,then the institutionwill persist,even if apparent underlyingpower and interestshave shifted.66Duncan Snidal, in his formal representationof what is most likelyto happen as a hegemon falters,includes as an untheorized variable "interestin the regime," with the obvious positive relationshipbe- tween interestin the regime and willingnessto expend resourcesto maintain it after hegemonic decline.67Constructivist research, through exploring the natureof the norms,practices, and identitiesconstituting membership in some institution,can provide some measurable substantivecontent for that variable. Although constructivistsand neoliberals agree that anarchy does not pre- clude cooperation among states, how they understand the emergence and reproductionof such cooperation yields very differentaccounts and research agendas. THE DEMOCRATIC PEACE. The observation that democratic states have not fought each other is an empirical regularityin search of a theory.Neither structuralnor normativeaccounts farevery well.68 The formerrequires assum- ing a consistentlybellicose executivebeing constrainedby a pacificpublic and its duly-electedrepresentative institutions-but only when democraticadver-

65. On lags and stickiness,see Stephen D. Krasner,State Power and the Structureof International Trade," WorldPolitics, Vol. 28, No. 3 (April 1976), pp. 317-343. On transactioncosts, see Keohane, AfterHegemony. 66. Anotherconstructivist hypothesis offers itself here: institutionalizedcooperation will be more likelyto endure to the extentthat the identitiesof the membersof that institutionare understood as common and they are reproduced by a thick array of social practices. This is meant as a continuum,with narrow self-interestbeing arrayed at one end of the spectrum,neoliberal institu- tionalizationof self-interestedcooperation in the middle, communityof identitytoward the other end, and harmonyat the otherpole. 67. Duncan Snidal, "The Limitsof Hegemonic StabilityTheory," International Organization, Vol. 39, No. 4 (Autumn 1985), esp. pp. 610-611. 68. For a comprehensivereview of the most recent literatureon the democraticpeace, and an empiricaltest that shows that satisfactionwith the status quo (a variable subject to constructivist interpretation)is the single most importantfactor affecting the use of force,by democracies and authoritarianstates alike, see David L. Rousseau, ChristopherGelpi, and Dan Reiter,"Assessing the Dyadic Nature of the Democratic Peace, 1918-1988,"American Political Science Review, Vol. 90, No. 3 (September1996), p. 527. InternationalSecurity 23:1 | 192

saries are about. The latterhas more promise,but its naturalizationof certain aspects of liberalism-the market,nonviolent resolution of differences,the franchise,the FirstAmendment-and its crucial assumption that these norms actually matterto decision makers in democraticstates when making choices about war and peace with other democracies, are untenable and untested, respectively. Constructivismis perfectlysuited to the task of testingand fundamentally revising the democraticpeace.69 Its approach aims at apprehending how the social practicesand norms of states constructthe identitiesand interestsof the same. Ergo, if democracies do not fighteach other,then it must be because of the way they understand each other,their intersubjectiveaccounts of each other,and the socio-internationalpractices that accompany those accounts.70 But constructivismcould offera more general account of zones of peace, one not limitedto democracies.Different periods of the historiesof both Africaand Latin America have been marked by long stretchesof little or no warfare between states. These pacific periods are obviously not associated with any "objective" indicatorsof democracy.By investigatinghow Africanand Latin American states constructedthemselves and others,it might be possible to understand these neglected zones of "authoritarianpeace."

ConstructivistPuzzles

Constructivismoffers an account of the politicsof identity.71It proposes a way of understandinghow nationalism,ethnicity, race, gender,religion, and sexu- ality,and otherintersubjectively understood communties,are each involved in an account of global politics. Understandinghow identitiesare constructed, what norms and practicesaccompany theirreproduction, and how they con- structeach otheris a major part of the constructivistresearch program.

69. For a very well developed researchdesign to test constructivistversus mainstreamaccounts of the democraticpeace, see Colin Kahl, "Constructinga Separate Peace: Constructivism,Collective Liberal Identity,and the Democratic Peace," SecurityStudies (forthcoming). 70. For accounts of the democraticpeace that focus on its contextualintersubjective characters, see Ido Oren, "The Subjectivityof the 'Democratic' Peace: Changing U.S. Perceptionsof Imperial Germany,"International Security, Vol. 20, No. 2 (Fall 1995), pp. 147-184; Thomas Risse-Kappen, Cooperationamong Democracies,p. 30; and Risse-Kappen, "Collective Identityin a Democratic Community,"pp. 366-367. 71. I do not tryto compile a comprehensiveset of questions forconstructivists, but instead merely elaborate general themes for research,themes that do not have a prominentplace in mainstream internationalrelations theory. The Promiseof Constructivism| 193

Although nationalism and ethnicityare receivingmore attentionin main- stream internationalrelations theory, attention to gender,sexuality, race, and religionhave received much less, and certainlynone of them is part of either neorealist or neoliberal accounts of how the world works.72Constructivism promises to deal with these issues, not merely because they are topical or heretoforeundervalued, but because as varietiesof identity,they are centralto how constructivismgenerates understandings of social phenomena. Construc- tivismassumes, a priori,that identitiesare potentiallypart of the constitutive practicesof the state,and so, productive of its actions at home and abroad.73 One of the most importantby-products of this concernwith identitypolitics is the returnof differencesamong states. The same state is, in effect,many differentactors in world politics,and differentstates behave differentlytoward other states,based on the identitiesof each. If true, then we should expect differentpatterns of behavior across groups of states with differentidentities and interests.74Although it is temptingto assert that similaritybreeds coop- eration,it is impossible to make such an a priori claim. Identitieshave much more meaning for each state than a mere label. Identitiesoffer each state an understandingof other states,its nature,motives, interests,probable actions, attitudes,and role in any given political context. Understandinganother state as one identity,rather than another,has conse- quences for the possible actions of both. For example, Michael Barnetthas speculated that the failure of deterrenceagainst Iraq in Kuwait in 1990 is because Saudi Arabia was seen as an "Arab," ratherthan a "sovereign,"state. Iraq's understandingof Saudi Arabia as an Arab state implied that Riyadh would never allow U.S. forcesto deploy on Arab territory.If, instead, Iraq had

72. For a criticalview of neorealism'sbelated effortsto capture nationalism,see Yosef Lapid and FriedrichKratochwil, "Revisiting the 'National': Toward an IdentityAgenda in Neorealism?, in Lapid and Kratochwil,The Returnof Cultureand Identity,pp. 105-126. For a most imaginative criticalconstructivist treatment of nationalism,see , "Ground Identity:Nature, Place, and Space in Nationalism,"in ibid., pp. 129-145; see also Roxanne Lynn Doty,"Sovereignty and the Nation: Constructingthe Boundaries of National Identity,"in Thomas J. Bierstekerand CynthiaWeber, eds., StateSovereignty as Social Construct(Cambridge, U.K.: Cambridge University Press, 1996) pp. 121-147. 73. For example,J. Ann Ticknerobserves thatcontemporary masculinized Westernunderstandings of themselveslead to feminizedportrayals of the South as "emotional and unpredictable.Tickner, "Identityin InternationalRelations Theory: FeministPerspectives," in Lapid and Kratochwil,The Returnof Culture and Identity, pp. 147-162. 74. For example,Risse-Kappen, "Collective Identityin a DemocraticCommunity," finds a common identitywithin the North AtlanticTreaty Organization; see also Iver B. Neumann and JenniferM. Welsh, "The Other in European self-definition,"Review of InternationalStudies, Vol. 17, No. 4 (October 1991), pp. 327-348, for an exploration of "Christian" and "European" states versus "Islamic" "Asiatic" Turkey. InternationalSecurity 23:1 | 194

understood Saudi Arabia as a sovereignstate, in a realistworld, it would have perhaps expected Saudi balancing against Iraqi actions in Kuwait, including U.S. militaryintervention, and would have been deterred.75In other words, neorealistpredictions of balancing behavior,such as thatof Saudi Arabia, rely on a single particularidentity being ascribed to that countryby Iraq. But if alternativeidentities are possible, as constructivismsuggests, the neorealist world is smaller than alleged. Or anotherstate may not be seen as another"state" at all, but instead as an ally, friend,enemy, co-guarantor,threat, a democracy,and so on.76 Finally, constructivism'sexpectation of multiple identitiesfor actors in world politics rests on an openness to local historicalcontext. This receptivityto identities being generated and reproduced empirically,rather than restingon pregiven assumptions, opens up the study of world politics to differentunits alto- gether.77Hypothesizing differences among statesallows formovement beyond the typical binary characterizationsof mainstream internationalrelations: democratic-nondemocratic,great power-non-greatpower, North-South,and so forth.While these common axes of analysis are certainlyrelevant, construc- tivism promises to explain many other meaningfulcommunities of identity throughoutworld politics. A thirdconstructivist promise is to returnculture and domestic politics to internationalrelations theory. To the extentthat constructivism is ontologically agnostic-that is, it does not include or exclude any particularvariables as meaningful-it envisions no disciplinarydivides between internationalrela- tions and comparativesubfields (or any fieldsfor that matter). Constructivism has no inherentfocus on "second image" accounts of world politics. In fact, an appropriate criticismwould be that it has remained far too long at the systemiclevel of analysis.78Nevertheless, constructivism provides a promising

75. Michael N. Barnett,"Institutions, Roles, and Disorder: The Case of the Arab States System," InternationalStudies Quarterly, Vol. 37, No. 3 (September1993), pp. 271-296. 76. See Risse-Kappen,"Collective Identityin a DemocraticCommunity," and Michael N. Barnett, "Sovereignty,Nationalism, and Regional Order in the Arab System," InternationalOrganization, Vol. 49, No. 3 (Summer 1995), pp. 479-510, forexamples. 77. Yale Ferguson and Richard Mansbach, for example, offera rich varietyof "polities," such as city-states,civilizations, polis, empires, kingdoms, caliphates, each of which had and, in some cases, has and will have, meaningfulidentities in world politics. Ferguson and Mansbach, "Past as Prelude," pp. 22-28, and Sujata ChakrabartiPasic, "CulturingInternational Relations Theory," both in Lapid and Kratochwil,The Returnof Cultureand Identity,pp. 85-104. 78. Keohane, in "InternationalInstitutions," p. 392, has made this observationabout "reflectivist" scholarship..For similarlaments, see Dessler, "What's At Stake," p. 471; and Barnett,"Institutions, Roles, and Disorder," p. 276. Alexander Wendt acknowledges he has "systematicallybracketed" domestic factorsin Wendt,"Anarchy Is What States Make of It," p. 423. The Promiseof Constructivism| 195

approach foruncovering those featuresof domestic society,culture, and poli- ticsthat should matterto stateidentity and stateaction in global politics.There are many differentways in which a constructivistaccount can operate at the domestic level. I mentiononly several here. Any state identityin world politics is partly the product of the social practicesthat constitute that identity at home.79In this way, identitypolitics at home constrainand enable state identity,interests, and actions abroad. Ashis Nandy has writtenabout the close connectionbetween VictorianBritish gen- erationaland genderidentities at home and the colonizationof India. Victorian Britaindrew a very strictline between the sexes and also between generations, differentiatingthe latter into young and old, productive and unproductive, respectively.British colonial dominance was understood as masculine in rela- tionshipto Indian's femininesubmission, and Indian culturewas understood as infantileand archaic. In these ways Victorianunderstandings of itselfmade India comprehensibleto Britainin a particularway.80 Whereas conventional accounts of colonialismand imperialismrely on disparitiesin relativematerial power to explain relations of domination and subordination,constructivists would add thatno account of such hierarchicaloutcomes is complete without exploring how imperial identities are constructedboth at home and with respectto the subordinatedOther abroad.81 Even ifmaterial power is necessary to produce imperialism,its reproductioncannot be understood withoutinves- tigating the social practices that accompanied it and the discursive power, especially in the formof related identities,they wielded. Within the state itself might exist areas of cultural practice, sufficiently empowered throughinstitutionalization and authorization,to exerta constitu- tive or causative influence on state policy.82The state's assumed need to constructa national identityat home to legitimizethe state'sextractive author- ity has effectson state identityabroad. A more criticalconstructivist account

79. Two works thatmake the connectionbetween domesticidentity construction at home and state identityare Audie Klotz, Normsin internationalrelations: the struggle against apartheid (Ithaca, N.Y.: Cornell UniversityPress, 1995); and Peter J. Katzenstein, CulturalNorms and National Security (Ithaca, N.Y.: Cornell UniversityPress, 1996). 80. Inayatullahand Blaney,"Knowing Encounters,"pp. 76-80. 81. Compare this,for example, to Richard Cottam's very interestingaccount of imperial British images of Egypt. The critical differenceis that Cottam does not see British constructionsof themselvesor theirsociety's parts as relevantto an understandingof Britishimages of Egyptians. Richard Cottam,Foreign Policy Motivation: A GeneralTheory and Case Study(Pittsburgh: University of PittsburghPress, 1977). 82. One mightsay this about the Frenchmilitary between World Wars I and II. See Kier,"Culture and FrenchMilitary Doctrine beforeWorld War II." InternationalSecurity 23:1 | 196

mightbegin by positing the state's need for an Other in world politics,so as to justifyits own rule at home.83 A last promise of constructivismconcerns not so much research issues as research strategy.Constructivism offers a heterogamous research approach: thatis, it readilycombines with differentfields and disciplines.Constructivism itselfis the product of structurallinguistics, postmodern political theory,criti- cal theory,cultural and media studies, literarycriticism, and no doubt others. Far fromclaiming primacy as a theoryof internationalpolitics, constructivism lends itselfto collaborationwith otherapproaches, both withinpolitical science and outside. Literaturesin decision making,political culture, socialization, and experimentalcognitive and social psychologywould seem to be most promis- ing partners.

CONSTRUCTIVIST PROBLEMS A constructivistresearch program, like all others,has unexplained anomalies, but theirexistence need not necessitatethe donning of protectivebelts of any sort.Conventional constructivism has one large problemthat has several parts. FriedrichKratochwil has observed thatno theoryof culturecan substitutefor a theoryof politics.84Paul Kowert and JeffreyLegro have pointed out that thereis no causal theoryof identityconstruction offered by any of the authors in the Katzensteinvolume.85 Both criticismsare as accurate as theyare differ- ent, and imply differentremedies. Kratochwil's statementreinforces the point that constructivismis an ap- proach, not a theory.And if it is a theory,it is a theoryof process, not sub- stantive outcome. In order to achieve the latter,constructivism must adopt some theoryof politics to make it work. Criticaltheory is far more advanced in thisregard than conventionalconstructivism, but it comes at a price,a price thatone may or may not be willingto pay,depending on empirical,theoretical, and/or aestheticinterests. I have described how differentlycritical and con- ventional constructivismtreat the originsof identityand the nature of power.

83. This is done by David Campbell, WritingSecurity: United States Foreign Policy and thePolitics ofIdentity (Minneapolis: Universityof Minnesota Press, 1992) and JimGeorge, Discoursesof Global Politics:A Critical(Re)Introduction to InternationalRelations (Boulder, Colo.: Lynne Rienner,1994). 84. Kratochwil,"Is the Ship of Culture at Sea or Returning?"p. 206. 85. Paul Kowert and JeffreyLegro, "Norms, Identity,and Their Limits:A TheoreticalReprise," in Katzenstein,The Cultureof National Security, p. 469. For othercritical reviews of constructivismand world politics,see JeffreyT. Checkel, "The ConstructivistTurn in InternationalRelations Theory," WorldPolitics, Vol. 50, No. 2 (January1998), pp. 324-348, and Emanuel Adler,"Seizing the Middle Ground: Constructivismin World Politics,"European Journal of International Relations, Vol. 3, No. 3 (1997), pp. 319-363. The Promiseof Constructivism | 197

It is here thatcritical theory finds its animatingtheory of politics.By assuming that the identitiesof the Self and Other are inextricablybound up in a rela- tionship of power, and that the state is a dominating instrument,critical theoristscan offertheoretically informed accounts of the politicsof identity:at least along the dimensions specified,that of hierarchy,subordination, domina- tion,emancipation, and state-societystruggle. The price paid for such theories of politics,however, is an ironic one that naturalizes certain "realities," privilegingsocial relations of dominance and hierarchy.Of course, criticaltheory asserts its ultimateopenness to variation and change, but the point here is that its theoryof politics,a priori,is more closed than thatof its conventionalversion, which stands accused of theoretical underspecification.The problem of underspecificationexists because conven- tional constructivism,as a theoryof process,does not specifythe existence,let alone the precise nature or value, of its main causal/constitutiveelements: identities,norms, practices, and social structures.Instead, constructivismspe- cifies how these elements are theoreticallysituated vis-a-vis each other,pro- viding an understanding of a process and an outcome, but no a priori predictionper se. The advantages of such an approach are in the nonpareil richness of its elaboration of causal/constitutivemechanisms in any given social contextand its openness (and not just in the last instance,as in critical theory)to the discoveryof othersubstantive theoretical elements at work. The cost here, however,is the absence of a causal theoryof identity. The dilemma is thatthe more conventionalconstructivism moves to furnish such a causal theory,the more it loses the possibility of maintaining the ontological openness thatits interpretivistmethods afford.But the dilemma is a continuum,not a binaryopposition. Conventionalconstructivists can and do specify their theoreticalelements in advance in practice. Just to take one example, not a single author in the Katzensteinvolume assessed gender,class, or race in any of theiranalyses. This observation(not criticism)is intended to underline how conventionalconstructivists already bound theira prioritheo- retical domains according to empirical interestand theoreticalpriors. More- over, conventionalconstructivists can make predictions,if they choose. Their only constraintis just how durable theybelieve the social structuresto be that they have demonstratedare constrainingthe reproductionof identities,inter- ests, norms, and practices,in some social context.For example, when Risse- Kappen argues that North Atlantic Treaty Organization (NATO) members regardeach otheras liberal allies, ratherthan as realiststates balancing against a threat,he is making a prediction:if NATO memberssee each otheras liberal allies, NATO will persistbeyond the point where the threatdisappears. InternationalSecurity 23:1 | 198

One obstacle to the development of a causal model of identityis conven- tional constructivism'ssilence on the issue of intentionality.Critical theorists confidentlydeclare theirindifference to the issue: establishingcausality is an illusory goal. Kowert and Legro point out the failure of any author in the Katzensteinvolume to establish more than a correlativerelationship between an identityand an outcome. In fact,the authors do far more than that: they control for alternativeexplanations and they show the connectionbetween norms and interestsand outcomes. But what is missing is the decision based on the identity.Here again, constructivistheterogamy allows foran attempted fix.The answer may lie in tryingto marryconstructivist process to psychologi- cal process. Kowert and Legro discuss the possibilityin terms of the experi- mental social psychologicalwork of MarilynBrewer and JonathanTurner.86 To the extentit is possible to establisha causal link between a particularidentity, such as Japanese antimilitarism,and an interestin opposing Japanese military expenditures(or between beliefin a norm,such as humanitarianintervention- ism, and an action to fulfillthat norm), it mightbe attainablethrough ongoing work on the connectionbetween identityand behavior in social psychology. The last problem with constructivismis really not so much a problem as it is an advantage. Constructivism'stheory of process and commitmentto inter- pretivistthick description place extraordinarydemands on the researcherto gathermountains of elaborate empirical data. To reconstructthe operation of identitypolitics, even in a limited domain for a short period, requires thou- sands of pages of reading, months of interviewsand archival research,and a host of less conventionalactivities, such as ridingpublic transportation,stand- ing in lines, and going to bars and caf6s to participatein local practices.(The latterneed not be so onerous.) The point here is that the evidence necessary to develop an understandingof, say, a national identity,its relationto domestic identities,the practicesthat constituteboth, implied interestsof each, and the overall social structureis necessarily vast and varied. Constructivismis no shortcut.

TheConstructivist Promise

The assumptions thatunderlay constructivismaccount forits differentunder- standing of world politics. Since actors and structuresare mutually con- structed,state behavior in the face of differentdistributions of power or

86. Ibid., p. 479. ThePromise of Constructivism I 199

anarchyis unknowable absent a reconstructionof the intersubjectivemeaning of these structuresand actors.Since actors have multipleidentities, and these identitiesimply differentinterests, the a priori and exogenous attributionof identicalinterests to states is invalid. Since power is both materialand discur- sive, patternedbehavior over time should be understoodas a resultof material or economic power workingin concertwith ideological structures,social prac- tices,institutionalized norms, and intersubjectivewebs of meaning. The great- est power of all is thatwhich disciplines actorsto naturallyimagine only those actions that reproduce the underlyingarrangements of power-material and discursive.Since constructivistsocial structuresare both enduring and muta- ble, change in world politics is considered both difficultand possible. A conventionalconstructivist recasting of mainstreaminternational relations puzzles is based on the implicationsof its assumptions.Since what constitutes a threatcan never be stated as an a priori,primordial constant, it should be approached as a social constructionof an Other,and theorized at that level. Since identities,norms, and social practices reduce uncertainty,the security dilemma should not be the startingpoint foranalyzing relationsamong states. Since states are already situated in multiple social contexts,any account of (non)cooperation among them should begin by exploring how their under- standingsof each othergenerate their relevant interests. Since communitiesof identityare expected to exist,patterns of behavior that spur scholars to con- sider a liberalpeace should instead provoke us to considerzones of peace more generally. A conventional constructivistaccount of politics operates between main- streaminternational relations and criticaltheory. Conventional constructivism rejectsthe mainstreampresumption that world politicsis so homogenous that universallyvalid generalizationscan be expected to come of theorizingabout it. It denies the criticalconstructivist position that world politics is so hetero- geneous thatwe should presume to look foronly the unique and the differen- tiating.Contrary to both these two approaches, conventional constructivism presumes we should be looking forcommunities of intersubjectivityin world politics,domains withinwhich actors share understandingsof themselvesand each other,yielding predictable and replicable patterns of action within a specificcontext. Mainstream internationalrelations theorytreats world politics as an inte- grated whole, undifferentiatedby either time or territory.Critical theoryre- gards world politicsas an arrayof fragmentsthat can never add up to a whole, and regards effortsto constructsuch a whole as a political move to impose InternationalSecurity 23:1 | 200

some kind of rationalistic,naturalized order on irrepressibledifference. Con- ventional constructivism,on the other hand, regards the world as a compli- cated and vast array of differentdomains, the apprehension of all of which could never yield a fullycoherent picture of internationalpolitics. The failure to account forany one of them,however, will guarantee a theoreticallyunsat- isfyingunderstanding of the world. In effect,the promise of constructivismis to restore a kind of partial order and predictabilityto world politics that derives not fromimposed homogeneity,but froman appreciationof difference.

Corrections:

In Alexei G. Arbatov,"Military Reform in Russia: Dilemmas, Obstacles, and Prospects,"Vol. 22, No. 4 (Spring 1998): p. 86 line 13 should read "The quantity of militarypersonnel . .. must be sacrificedfor higher quality arms"; p. 90 line 17 should read "Numerical Balance"; p. 92 line 3 should read "reinforcement advantages and interdictioncapabilities against Russian reinforcements"; p. 106 line 10 should read "has never been preprogrammedinto"; p. 109 line 11 should read "to findits forcelevels and structureon a prioritybasis"; p. 130 line 1 should read "down to a level of 1.2 millionby 1999"; and p. 130 line 25 should read "are not carried out."