Published in José Maria Rosales & Fred Dallmayr (eds), BeyondNationalism? Sovereignty and Citizenship, Lanham (MD), Rowman and

Littlefield, 2001, p. 75-97

War, politics, and sovereignviolence

John Crowley

What is a “political solution” to a conflict? Likemany theoretical questions, this has circumstantialorigins. It so happens that the first draft of thischapter was written in the Spring of 1999 during theNATO bombing campaign against Yugoslavia. A feature ofpublic debate about the rights and wrongs of the warwas general agreement that it was not “political”.Both opponents and supporters of military actioncalled for a “political solution” to the problems ofKosovo. The only difference between them was thatsupporters of the NATO perspective regarded suchaction as a pre-condition for a political solution. Thefailure of the Paris and Rambouillet negotiations inearly 1999 to produce a generally acceptable agreementwas precisely what was judged to justify the war. Onthe contrary, critics of the NATO campaign claimedthat a political solution was what should have beenpursued instead of military action. To that extent,nearly everyone would agree that the Yugoslavgovernment’s acceptance of a de facto UN protectorate inKosovo in exchange for a NATO cease-fire representsthe reestablishment of the previously suspendedpolitical process.

1

2Yet, in spite of this consensus, something seemed

wrong. What, after all, could be more “political” thana war conducted under the aegis of constituted au-thorities, aimed at effecting a change in thedistribution of power within Kosovo in particular andYugoslavia in general, and furthermore saturated byanalysis of its effects on the legitimacy andstability of the Yugoslav government? What could bemore “political” than a public debate on thesignificance of universal human rights, on thecircumstances in which force might be used to promotethem, and on the relation between the moral basis andthe institutional legitimacy of the use of force?Surely, in this case at least, war was – for better orfor worse – a mode, rather than the “opposite”, ofpolitics.

Thus, a circumstantial running commentary oncurrent affairs leads into the heart of politicaltheory. For profound conceptual reasons, but also forpractical prudential reasons that the war of 1999clearly illustrates, theories of politics aregenerally – perhaps necessarily – theories of war.Clausewitz states that war is the prosecution ofpolitical objectives by certain specific techniques.Foucault provocatively flips that famous dictum, andsuggests that politics be analyzed as the continuationof war by other means.1 Weber offers a sociologicalbridge between the two perspectives by defining thestate in terms of the specific means at its disposal:the use of force.2 But to base the state’s existenceas a state on its capacity to monopolize thelegitimate use of force is to make the possibility of war– albeit, internally, in the peculiar form of itssuccessful avoidance – central to politics. Warbetween states therefore appears as a naturalexpression of the state’s characteristic features atthe point where the claim to monopoly no longeroperates. Carl Schmitt arguably pushed to its logicalconclusion the Weberian idea of force as the “joint” –

War, politics, and sovereign violence 3what is called in both ordinary and theoretical French“articulation”, but often lazily translated into Englishas “articulation” – between internal and externalsovereignty. Conversely, liberal theory – of whichRawls and Habermas are exemplary – strongly rejectsthis “warlike” view of politics. But it does so in away that tends implicitly to validate it, by definingpolitics normatively as a realm of reason, absolutelydistinct from war, and therefore dependent on theempirical absence of war. In this sense, a liberaltheory of politics that did not offer a theory of warwould be analytically vacuous.

My contention here will be that standard liberaltheory in fact fails to offer an adequateconceptualization of the relation between “politics”and “war”, and is therefore, if not vacuous, at leastless normatively compelling than it aspires to be. Inorder to substantiate this claim, I propose to focusfirst on the practical politics of war, withparticular emphasis on the theoretically crucial grayareas characteristic of the internal conflictsabsurdly called “civil”. These empirical observationsusefully clarify the more general theoretical issue ofthe relation between “reason” and “violence” – oftenmisleadingly equated with that between politics andwar. Since the criticisms in the first two sectionsfocus largely on contemporary common sense, asexpressed particularly in media debate, moretheoretical argument might seem little affected bythem. In fact, as I propose to illustrate by internalcritique of Rawls and Habermas, this is not true.

Furthermore, a coherent theory of the distinctionbetween politics and war is not simply something thatRawls, Habermas, and others, contingently fail to

1. Michel Foucault, Il faut défendre la société. Cours au Collège deFrance. 1976, (Paris: Gallimard / Seuil, 1997).

2. Weber’s term is die Gewalt, which has in German morecomplex connotations than “force” in English. It can, inparticular, mean both “administration” and “violence”.

4provide. It is, even in principle, not available. Andthis problem goes deeper than the structure of liberaltheory, and affects any attempt to bring within asingle framework the political issues of foundation andjustification. Any political authority subject tochallenge needs to establish itself, or it byassumption simply fails to exist. Its ability tojustify itself in the eyes of its challengers is, ofcourse, one possible mode of establishment. But whatif such justification fails? The answer is crude andsimple: if it is to exist at all, it must necessarilyestablish itself by other means – means that ignorethe challengers for the purposes of justification(although other logics of justification may be atwork) – means that, in at least a minimal sense, areviolent. We reach here the point where politics andwar merge in the body of what is classically known asthe “sovereign”.3

War as a mode of politics

What the 1999 conflict in the Balkans illustrates isthat war is very widely regarded as something strictlyseparated from politics and in some sense opposed toit. The idea is of course plausible, and it is easyenough to understand how it has come to be generallyaccepted. Nonetheless, when it is pushed a littleharder, a number of interesting and importantunresolved difficulties emerge.

The best way of getting to grips with the problemis simply to look empirically at what is going on inconflicts such as those of the former Yugoslavia,Algeria, or Northern Ireland – to take just threequite different examples. I make no attempt here todescribe them, still less to analyze them, in detail.It is possible, however, to offer a characterizationbased on certain generic features. A “political”

3. On this, see further Giorgio Agamben, Homo sacer. 1, Ilpotere sovrano e la nuda vita (Turin: Einaudi, 1995).

War, politics, and sovereign violence 5solution is usually defined as one in which the use ofviolence is suspended; cease-fires and trucestherefore play a crucial role in the political processby opening spaces for negotiation. Furthermore adurable political solution requires not merely thesuspension of violence but some kind of guaranteeagainst its re-emergence. There is usually strongemphasis on disarming paramilitary groups and somehowtaking their weapons out of circulation. A com-plementary objective is “reconciliation” – a ambiguousnotion that covers a wide range of processes involvingvarious kinds of forgiveness and forgetfulness.4

Whatever precisely it may entail, reconciliation isintended to remove the basis for the re-emergence ofviolence, just as disarmament reduces the capacity toexert it.

It is striking that such processes stand in anambiguous relation to the idea of sovereignty. At onelevel, the suspension of violence in such casesrestores the state in the conventional Weberian sense,since the characteristic feature of the violencesuspended was to be amenable neither to suppressionnor to legitimation. Yet the way in which it does soseems to derogate from the idea of statehood. Thesuspension of violence is not an instance of oneparty’s claim to the monopoly of the legitimate use offorce, nor the consequence of the success of such aclaim – in many cases it is indeed quite the opposite,viz. the recognition that armed confrontation hasproduced stalemate, and is likely to continue to doso. Pacification thus involves the renunciation,within a specific sphere judged to be political, of allclaims to the legitimate use of violence, includingthose of the juridically sovereign authority – hence

4. On the political dimensions and problems ofreconciliation, see the papers collected in Cultures & Conflits,n° 37 (2000, forthcoming); in particular John Crowley, “Lapacification du politique: quelques réflexions sur lestransitions immorales”.

6the central issue of amnesty. The juridical sovereigndoes not, however, thereby cease to exist. Indeed, itsstatus may even be reinforced in the non-politicalsphere by its new-found modesty in the politicalsphere: “peace processes” usually involve, among otherthings, heightened repression of the kind of pettycrime that tends to flourish in the interstices ofpolitical violence. But the limits and legitimacy ofthe use of force by, for example, the Britishgovernment in Northern Ireland or the Israeligovernment in the occupied territories are in practice(and are, at least tacitly, recognized in principle tobe) within the scope of negotiation. Such negotiatedprocesses are thus predicated upon the suspension ofsovereignty, in both the strictly legal and socio-historical senses. They aim of course at the ultimatedisappearance of political violence, and to thatextent at the re-establishment of sovereignty, sincewhat characterizes distinctively political violence isprecisely its claim to legitimacy. However the meansthey use to obtain the temporary suspension ofviolence necessarily recognize the existence offundamental political conflict. Conversely, whatcharacterizes a strictly sovereign perspective onpolitical violence is the refusal to recognize it aspolitical or its perpetrators as people one can “dobusiness with”, and therefore the denial that itssuppression differs from that of ordinary criminalactivity. What outsiders regard as civil war, sove-reign governments call “policing” or “lawenforcement”, or at most – through clenched teeth –“pacification”.

Less theoretically, we may say simply that thecondition for the kinds of peace processes illustratedin various ways in Northern Ireland, South Africa,Lebanon, Mozambique, and many other places, is apartial suspension of questions about legitimacy.Conversely, when, as in Algeria for instance, such acondition cannot be met, not just reconciliation but

War, politics, and sovereign violence 7even minimally civil peace prove elusive. Whether thissuspension involves the explicit recognition of theequal legitimacy of all positions – which means, inpractice, legitimizing the past use of violence –, orthe explicit or tacit denial of legitimacy to any useof violence, makes little difference. The key point issimply that such a framework admits no privilegedposition: at least for the limited purposes of theongoing process, non-state actors are regarded ashaving, de facto, the status of a kind of quasi-state –with, indeed, a higher degree of criminal immunitythan international law now grants to the office-holders of legally sovereign governments.

Clearly, the suspension of sovereignty in thissense opens a political space for negotiation, aimedideally both at practical compromises on issues ofspecial urgency and at the development of newprinciples of legitimacy more broadly acceptable thanthose against which, or in the name of which, pastpolitical violence was used. What is far less clear ishow to draw the line between this process and theprior condition of armed confrontation. Yet, unlesssuch a line can be drawn, the claim that there issomething uniquely, or even distinctively, “political”about the negotiation process is difficult to sustain.

In practical terms, the line exists as a matter offact. It is precisely one of the key things thatparties to a peace process agree on. But its functionis primarily symbolic and illocutory. It announcespublicly the possibility of a solution, and thelegitimacy of the process, and thereby possiblycontributes to making such a solution possible. Italso, for the same reason, has a tactical purpose thatis by no means insignificant: to lock the parties into the process by prior commitments that raise thecosts of failure. What this generally means inpractice is making erstwhile “terrorists” – nowaccepted, however reluctantly, as people to “dobusiness” with – (e.g. the IRA) accountable for the

8process to their outside sponsors (e.g. the USgovernment) and thereby acquiring some degree of lev-erage over them. These are risky tactics, of course,since they create issues of “face” that secretcontacts do not, and in addition the stability ofnegotiators’ commitments cannot be guaranteed wheneven limited democratic criteria of representativenessand accountability are in operation. The principle,nonetheless, has been adopted, often effectively, in awide variety of cases.

However, if this is to be regarded as line dividingpolitics from non-politics, it is crucial to note thatit is permeable in both directions. On the one hand,secret contacts between governments and insurgentswith whom, ostensibly, it would be unacceptable to “dobusiness” are an almost invariable feature of civilstrife in the broad sense. The British government hasconducted secret discussions with representatives ofthe Provisional IRA on a number of occasions since1972, in parallel with often brutal counter-insurgencyoperations. The National Party government in SouthAfrica had been talking to the ANC since the mid-1980s, even as it deployed its particularly viciousundercover forces against the very same opponents.Similarly, there was no obvious change in Israelistrategy or tactics to reflect the secret discussionsthat led to the Washington agreement. On the otherhand, violence is an integral part of the politicalprocesses of discussion and negotiation – not in anydeep conceptual sense, but simply because both thetactical use of force or the threat to use force, andthe inability of participants in talks fully tocontrol splinter groups, are a central issue.

Let us take just one example: the Northern Irish“peace process” as it has developed since theagreement signed in April 1998 by most of the relevantpolitical actors, including Sinn Féin, with the strongbacking of the UK, Irish and US governments, andratified by parallel referendums in Northern Ireland

War, politics, and sovereign violence 9and the Republic of Ireland the following month. Thequotation marks around “peace process” underline boththat this is what everyone calls it, and that there issharp disagreement about what the words mean. Evensuch an apparently straightforward notion as “IRAceasefire” has been the subject of persistent semanticwrangling since, empirically, it clearly cannot beequated with the literal suspension of violence. In asense, of course, such ambiguities are the whole pointof processes of this kind: only if all parties canmake some plausible claim to have “won”, and if no onelooks too closely at the inevitable fabrications andmisunderstandings, can any stable solution beachieved. But what this means is simply that violenceis at the heart of the political process, just aspolitics was always at the heart of the war. Thus, thecarefully maintained ambivalence about the relationsbetween Sinn Féin and the IRA and the resultingbrinkmanship over weapons “decommissioning” – whicheventually led to the suspension of the institutionsestablished by the 1998 agreement – are not failuresof the process, but characteristic features of it. Wemay, on normative grounds, prefer impartialdeliberation to bargaining based on a balance of pow-er; and we may, further, reasonably argue thatdeliberation and violence have no theoretical orempirical connection. However, where bargaining aimedat a modus vivendi is the limit of the practical best,things get more complex. No doubt there are excellentnormative reasons to prefer the ballot box to the Ar-malite – to adapt the standard Northern Irish phrase–, but it is implausible to regard them as belongingto radically separate worlds.

Reason, violence, and the “political”

The theoretical implications of these empiricalobservations are far from straightforward. Even if the

10attempt to equate politics exclusively with reason,and violence exclusively with war, were abandoned, itmight still be possible to draw a more subtle linebetween a higher form of politics, grounded solely inreason, and a debased form inseparable from violence.As we shall see, this is what Rawls and Habermas do bytheir dismissive critique of the notion of a modusvivendi. In order to deal with this argument, it is notenough to point to the empirical prevalence of interest-driven bargaining in supposedly deliberative contexts,which no one really denies. What has to be challengedis the normative significance of the ideal deliberativeperspective. The issue, in other words, is therelation between reason and violence as it impacts onthe notion of the political.

I do not wish to argue here that one should rejectthe conceptual distinction between reason andviolence, still less that violence might be somehowsuperior to reason if looked at from the appropriateangle. Such views are in fact surprisingly common,especially perhaps in the French intellectualtradition. In Derrida, for instance, violence becomesa kind of quasi-metaphysical category based on aconfluence of Bataille, Lacan and Levinas.5 The essayForce de loi6 is most explicit in this respect,particularly in its reading of Benjamin, but thefascination with the moment of foundation or origin,associated in terms of the fixing of meaning with thecritique of structuralism, is a leitmotiv of Derrida’swork in general. One possible consequence is that“reason” and “violence” may become simplyinterchangeable, as different ways of describing the

5. “Violence and metaphysics” is the title of the essayon Levinas in the early collection L’écriture et la différence(Paris: Seuil, 1967). The theme of violence and origin,however, runs through the whole book, and is ultimatelywhat gives it its unity.

6. Jacques Derrida, Force de loi. Le “Fondement mystique del’autorité” (Paris: Galilée, 1994).

War, politics, and sovereign violence 11asymptotic point of origin: the point irreducible tointerpretation or explanation. By this time, however,violence in any ordinary, concrete sense, as somethingpeople may do to other people, has entirely droppedout of the equation.

Another, rather different, denial of thesuperiority of reason to violence is at work in thecelebration of emancipatory violence. The locus classicusof this reversal is Sartre’s famous, or infamous,preface to Fanon’s Les damnés de la terre.7 Here, at least,the concrete reality of violence is not elided: asSartre puts it, when the native shoots the settler,both the oppressor and the oppressed disappear, leav-ing a dead man and a free man. The political point isthat such violence does not increase the amount ofviolence in a social system, but merely displaces andre-arranges it. Domination is, in the terms ofBourdieu’s sociology, symbolic violence; real violence,so long as it is politically informed, is not “worse”,but “better”. But sociology, of course, offers nobasis for such normative judgment, especially when ithas resolved violence as a relation between peopleinto violence as a relation between social positions.At least in Bourdieu, what serves to justify suchnormative statements, which are increasingly common inhis work, is a largely unargued criterion of“transparency” – one that is ultimately quite close tometaphysical concern with the “origin”.

More generally, what is at stake is the tendency ofviolence to become an object of fascination rather thanof analysis. Such fascination can work in two ap-parently opposite ways. On the one hand, violence maybe regarded as that about which nothing, ultimately,can be said.8 On the other, it can become a purelyerotic-aesthetic obsession. The status of Sade inFrench intellectual life, and the way in whichKubrick’s Clockwork Orange has been received in France –

7. Frantz Fanon, Les damnés de la terre (Paris: Maspéro,1961).

12much to the chagrin of Anthony Burgess, the author ofthe book on which the film was based – are indicativeof this climate. It is fairly clear, however, thatfascination, whether positive or negative, is here anobstacle to understanding.

The question, therefore, is whether it is possibleto find ways of talking about violence that are bothanalytically useful and morally defensible – that keepin touch, in other words, with both the fundamentalmoral fact that violence involves suffering, and theepistemological commonplace that the victim’s positionis not sociologically privileged. At its most generallevel, I shall simply avoid this question. My aim hereis more modest: to suggest a framework capable ofbringing some aspects of violence at least within thedomain of political theory in order to support thenegative theorem that normative support for reasonagainst violence – which I fully share – cannot bebased on assigning them to radically separate worlds.For this purpose, some kind of definition or criterionis useful. What I shall regard here as irreduciblyviolent is the application of force, with the potential to inflict pain,to a victim capable of giving consent, without that victim’s consent. Ido not mean that everything one might appropriatelyregard as violent fits this criterion; rather thatanything that fits it would meet almost any standardfor violence. Furthermore, the definition raises someserious difficulties, particularly with regard to theperpetrator’s intentionality and to the victim’s“capacity to give consent”. However, even if werestrict ourselves to the most straightforward cases,where both the use of force and the denial of consentare entirely deliberate, it is still capable ofproving fruitful.

8. References on this point are extremely numerous. Agood recent statement is Jean Leca, “Violence et ordre,”in Jean Hannoyer (ed.), Guerres civiles. Économies de la violence,dimensions de la civilité (Paris: Karthala – CERMOC, 1999): 315-26.

War, politics, and sovereign violence 13The most interesting consequence of this approach

is the light it throws on the conception of reason ascommunicative, of which Habermas is currently the mostinfluential proponent. Is violence “irrational” in thesense that its use is something that could never beregarded as legitimate in a context of free discus-sion? The answer is a complicated one, since itdepends not only on the ideal conditions relevant tothe justification of principles, but also on the pre-sumed characteristics of the situations to which theyare designed to apply. If these are imperfect,violence may in practice have a role to play. Manypeople regard as legitimate in principle both a puni-tive criminal law and the use of violence, in certaincircumstances and subject to certain restrictions, tocombat profound injustice. They may be wrong, but theargument that some people’s consent may count for less– because they are criminals or exploiters – iscertainly not obviously irrational. While it may con-flict with a Kantian view of equal moral worth, it iseven universalizable in the straightforward sense thatit is both intelligible and empirically observable todefend one’s use of violence by recognizing one’svictim’s equal entitlement to it. However morallyunpleasant, this is not obviously incoherent or self-contradictory. To make sense of violence in terms of“monological” reason, on the other hand, is far moredifficult.

This point is reinforced by consideration of theempirical forms that violence takes, at least in itspolitical manifestations – regarding as “political”for these purposes contexts where the infliction ofpain is associated with a claim about its legitimacythat exceeds the personal relation between perpetratorand victim. A wealth of evidence suggests that suchviolence is in a very real sense a mode of commu-nication. Recognition of the victim’s capacity to giveconsent is essential to the process, and indifferenceto consent is not usually an expression of callousness

14or lack of moral imagination, but a deliberate form ofsymbolic violence superimposed on the infliction ofphysical pain. Indeed, it is precisely in its mosthorrible forms (torture, mutilation, sacrilege, …),which tend to be symbolically charged, ritualized andusually stereotyped, that violence has the mostclearly defined discursive dimension.9 Furthermore,communication in such cases exceeds by design thebilateral relation between perpetrator and victim.Extreme violence is usually staged for the “benefit”of a group of spectators.10

Nor is this discursive dimension of violencepeculiar to such extreme cases. A terrorist bomb isalways, among other things but quite consciously, ademonstration of the capacity to cause murderousexplosions and therefore a threat, express or implied,to do it again. Whether or not it takes place in anongoing context of negotiations, it is thus usually,in a very tangible sense, a discursive move.11 This isprecisely why ceasefires (as in Northern Ireland) donot really take violence out of politics. Furthermore,this relation is at least partially reversible. Justas violence is, among other things, a form of dis-

9. I refer here in particular to the studies collectedin Françoise Héritier (ed.), De la violence (Paris: Éd. OdileJacob – volume 1, 1996; volume 2, 1999); and Hannoyer,Guerres civiles. See also John Keane, Reflections on violence(London: Verso, 1996).

10. It should be stressed that some important cases seemto detract from this characterization, most strikingly theNazi genocide. Such exceptions obviously cast doubt on theability of the indications given here to offer any basisfor a general “theory” of violence, whether in philosophicalor sociological terms. A useful philosophical discussionof this point is Myriam Revault d’Allonnes, Ce que l’homme faità l’homme. Essai sur le mal politique (Paris: Seuil, 1995).

11. A useful statement of the instrumental analysis ofpolitical violence, with particular reference toterrorism, is Yves Michaud, La violence, 3rd edition (Paris:PUF, 1996).

War, politics, and sovereign violence 15course, so discourse habitually involves a dimensionof violence. As post-structuralist theory hasstressed, the language of politics often affirms ordenies identities, affiliations, memories, dignity orpride, and has a capacity to inflict emotional painquite comparable to certain forms of mental torture.In more practical terms, and in some ways moreimportantly, the language of deliberation, especiallywhen public, has real effects, both deliberate andunintended, often indistinguishable from thestraightforward use of violence. Mental torture andsymbolic humiliation are fairly natural preludes tophysical extermination.12 Genocide in Rwanda started,in a very tangible sense, on the radio; and legalcodes in many countries define, for very good reasons,criminal offences of “incitement to (racial,religious, …) hatred”.

This does not mean, of course, that all forms ofhuman communication are violent or that differencesbetween modes of violence in terms of kind or of de-gree are morally indifferent. A consequence is,however, that normative arguments about violencecannot simply rely on reason as defining an inherentlynon-violent mode of interaction. Even in an idealdeliberative situation, from which violence is exhypothesi excluded as a valid mode of interaction, itis at stake with regard to the non-ideal conditions ofordinary social interaction. The justification of violence isnot some kind of absurd scandal, but one of the mostroutine of political issues. Furthermore, the relationof violence to reason has to do with tensions withinthe idea of reason itself. Violence is perhaps unrea-sonable rather than irrational; and, as we shall see withreference to Rawls, reasonableness and rationality arevery different faculties.

12. A helpful summary of these multifaceted interactionsbetween discourse and violence is Vivienne Jabri, Discourseson violence. Conflict analysis reconsidered (Manchester: ManchesterUniversity Press, 1996).

16These observations have distasteful implications

and raise a series of awkward philosophical problems,which is perhaps why even those who do succumb to thefascination of violence generally prefer to avoidthem. In addition, the classic response to thechallenge is deeply unsatisfactory. Thus Carl Schmittgrounds the “concept of the political” in what hecalls “anthropological pessimism”.13 This, in hisview, is one answer to two supposedly identical ques-tions: whether “man” is good or evil;14 and “whetherman is a dangerous being or not, a risky or a harmlesscreature”.15 “Pessimism” states, of course, that manis evil and dangerous. Furthermore, Schmitt regardsthe only other possible answer as “optimism” – theclaim that man is inherently good –, and this as ab-surd. “[A]ll genuine political theories presuppose manto be evil, i.e., by no means an unproblematic but adangerous and dynamic being.”16 It is no doubt truethat no political theory would subscribe to Schmitt’s“optimism”, but that does not entail “pessimism”because the questions themselves are loaded. As theuse of the word “evil” (in a non-moral, non-ethicalsense) implies, what Schmitt calls “anthropology” isreally theology. In the absence of his theologicalpresupposition, what changes is not the answer but the

War, politics, and sovereign violence 17question. The truly political question is whetherhuman intercourse is a risky business. This has su-perficial similarities to Schmitt’s formulation, butin dropping not just the category of “evil” but alsothat of “man” – as implying a supposedly uniform“human nature” –, it is in fact quite different.However, dislike of Schmitt’s answers has tended topreclude criticism of his questions,17 with the resultthat something like his “optimism” tends to be tacitlyadopted by his intellectual opponents. This makes itimpossible even to formulate the problem of violenceas an object for political theory.

It is impossible here to take this line of thoughtany further. What I wish to do is simply to draw fromthe understanding of contemporary political conflicts,and of the conceptual relation between reason andviolence, that I have sketched so far two postulatesthat, if valid, would point to a series of unacknow-ledged gaps in political theory. First, politics andwar are not sharply distinct forms or spheres of

13. Carl Schmitt, Der Begriff des Politischen (second edition,1932). English translation: The concept of the political (Chicago:University of Chicago Press, 1996).

14. “The distinction is to be taken here in a rathersummary fashion and not in any specifically moral orethical sense.” (Concept of the political: 58) – whatever thatmight mean.

15. The concept of the political, esp. §7. An unusuallyinteresting recent variant on the “anthropological”perspective is Wolfgang Sofsky, Traktat über die Gewalt (Frank-furt: S. Fischer Verlag, 1996). Sofsky attempts explicitlyto include within the analysis of violence theunderstanding of the fascination it often seems to exer-cise.

16. Concept of the political: 61.17. This tendency is broad, but by no means universal.

Schmitt’s thought continues, for instance, to exertsignificant influence among left-wing critics ofliberalism. See, for example, Chantal Mouffe (ed.), Thechallenge of Carl Schmitt (London: Verso, 1999).

18activity; reason and violence are not strictly sepa-rate modes of interaction; and still less can the twobinary oppositions be neatly superimposed andcorrelated. Second, whatever ethical or moraljudgement one may wish to make about violence, itrelies on extrinsic considerations. The use ofviolence may, in practice, be self-limiting in manycontexts; but the claim that violence is in principlea legitimate resource does not seem to be self-contradictory or self-undermining. Perhaps the mostconcise way of bringing together the two postulates isto state, in line with my opening remarks, that theidea of a “just war” (whether in bello or ad bellum) isinherently and profoundly political, and necessarilybased on an appeal to reason.

All this may seem blindingly obvious. In fact,whether or not people would accept the postulates asstated above, they are habitually denied or ignored.An anecdote may clarify the nature of this denial. In1998, I was asked to contribute a short piece on the“peace process” in Northern Ireland to the journal Cri-tique internationale.18 The main purpose in the editor’smind was to assist understanding of the reasons for,and the meaning of, the unexpectedly comprehensive“Good Friday” agreement finalized on April 10th 1998and endorsed by referenda in both the North and theRepublic the following month. Defining myself bycontrast with Gerard Delanty’s analysis of the peaceprocess as reflecting the emergence of a post-ethnicand post-nationalist form of political consciousnessopening a space for Habermasian deliberation,19 I hadsought to stress the shifting balance of power, socialcleavages and political aspirations that hadfacilitated a messy compromise by which most partiesagreed to fudge the key underlying issues and which,far from excluding violence, was intimately involvedwith it. Wishing to stress that such compromises were

War, politics, and sovereign violence 19as much the stuff of politics as higher-mindeddeliberation, I suggested the title “L’espace du politique[the space of the political] en Irlande du Nord”. Theeditor and managing editor disliked the title, whichthey took to miss the key feature of the peaceprocess, namely the re-emergence of politics as aresult of the suspension of violence. They thereforesuggested titles on the lines of “Le retour du politique”[the return of the political]. I had to fight quitehard to get my point across, precisely because it cutsagainst the grain of what now seems to be acceptedthinking about the nature of politics. Eventually, weagreed on “La pacification du politique”, a better and moreexplicit title than the original: the idea of“peaceful politics” clearly points to the conceptualcoherence and empirical possibility of “warlikepolitics”.

Rawls and Habermas on moral pluralism andpolitical conflict

If such implicit preconceptions about the nature ofpolitics and its relation to reason, violence and warwere simply a reflection of lack of thought or of theinfluence of journalistic clichés, they would notrequire extended discussion. I wish to suggest,however, that this is not the case. The presumptionthat politics “re-emerges” in places like NorthernIreland as the guns fall silent is central to

18. John Crowley, “La pacification du politique enIrlande du Nord,” Critique Internationale, n° 1 (1998): 35-42.

19. Gerard Delanty, “Negotiating the peace in NorthernIreland,” Journal of Peace Research, 32 (1995): 57-64; “Habermasand postnational identity: theoretical perspectives on theconflict in Northern Ireland,” Irish Political Studies, 11 (1996):20-32; “Social exclusion and the new nationalism: Europeantrends and their implications for Ireland,” Innovation: theEuropean Journal of Social Sciences, 10(2) (1997): 127-44.

20contemporary political theory, and, precisely becauseit is not usually examined at all closely, is asignificant source of weakness.

The problem is at its most striking in the work ofJohn Rawls, particularly in his attempts to revise histheory to respond to criticism of A theory of justice. Thefocus for the issues raised here is what Rawls regardsin his later work as the “fact of pluralism”. By thishe means the empirical fact of deep conflicts aroundfundamental values in contemporary societies, and theabsence of any reasonable or rational standardaccording to which such conflicts might be ad-judicated. Pluralism, understood as the irreduciblediversity of “comprehensive conceptions of the good”thus has a moral, and not simply factual, status: itis both the cause of conflict, and a constraint on theinstitutions designed to cope with conflict. As weshall see, what is ultimately at stake is the signi-ficance for democratic theory of the sociology ofconflict, which is implicit in the notion of“stability” to which the second version of Rawls’theory gives great prominence. In a word, contemporaryliberalism and its communitarian critics need a theoryof conflict, but are prevented from acquiring onebecause of their commitment to a peculiar idea ofpluralism. They are thus forced to write conflict outof the scope of theory just as they claim to betheorizing it. For the moment, however, I shallconcentrate on Rawlsian exegesis.

Any discussion of Rawls must take into account thesignificant differences between the two book-lengthversions of his thinking on social justice.20 For mypurposes here, the key point is the changing role ofthe notion of “stability”, which points to a subtle,unacknowledged shift at the most fundamental level,namely the point of the whole exercise, the purpose thetheory is intended to serve. The most basic premise ofany theory of justice is a set of assumptions aboutthe kind of people who make up a society, and the

War, politics, and sovereign violence 21various ways in which they interact. Simplifying verycrudely, TJ reasoned from the structural necessitiesof cooperation, whereas PL gives priority to a newlysalient fact of (reasonable) moral pluralism. The kindof people whose interactions produce pluralism in thissense are essentially the bearers of comprehensivedoctrines. It is plausible enough consider that anyfree society is highly likely to include significantgroups of citizens strongly committed to incompatiblebut reasonable comprehensive doctrines. However, Rawlsprovides inadequate evidence for the primacy attri-buted by him to this problem, and, quite apart fromhis argument, adequate evidence is not in factavailable.

The starting point is again the complex relationbetween TJ and PL. Rawls’ own statement is at onceaccurate and subtly misleading: “the account of sta-bility in part III of Theory [sc. TJ] is not con-sistent with the view as a whole. I believe alldifferences are consequences of removing that incon-sistency. Otherwise these lectures take the structureand content of Theory to remain substantially thesame.” (PL: xv-xvi). In fact the differences caused bythe new account of stability are far-reaching(although, tautologically, TJ is otherwise unaffected)because they affect not simply the substantive contentof stability but its very theoretical status. To putthe contrast very crudely, TJ is a theory of justicesensitive to the social context of its reproduction,

20. A theory of justice (Harvard: Harvard University Press,1971) [hereafter quoted as TJ]; Political liberalism, (New York:Columbia University Press, 1993) [hereafter quoted as PL].Rawls’ subsequent debate with Habermas [Journal of Philosophy,92/3 (1995)] introduces some additional complications thatwill be referred to where necessary but not discussed indetail. References are to the French edition [JürgenHabermas and John Rawls, Débat sur la justice politique (Paris: Éd.du Cerf, 1997)] which contains Habermas’s PolitischerLiberalismus, Rawls’ Reply to Habermas (49-142; quoted here asRH), and Habermas’s Einbeziehung des Anderen.

22whereas PL is a theory of social order and of the roleof justice in promoting it. However this conceptualreversal is never stated or defended.21 Rather itarises from features peculiar to PL that are justifiedprimarily in formal rather than substantive terms,viz. the fact of pluralism and the related distinctionbetween comprehensive and political liberalism. Thedistinctive contribution of PL, compared to TJ, is theimplied claim that stable justice is necessary for thecollective survival of a society primarily threatened bypolitical conflict arising from the clash of reasonable com-prehensive doctrines. The underlying theme, in otherwords, is a simple dichotomy: “mortal conflictmoderated only by circumstance and exhaustion, orequal liberty of conscience and freedom of thought.”(PL: xxxi). This implies two distinct assumptions: (1)that the clash of doctrines necessarily produces se-vere political conflict, and (2) that nothing else

21. Indeed, it is explicitly denied on at least oneoccasion: “I have not supposed that an overlappingconsensus is necessary for certain kinds of social unityand stability.” (PL: 149). The context is the (correct)emphasis on the fact that an overlapping consensus is nota political compromise. But the denial remains, in view ofRawls’ discussion as a whole, essentially unconvincing. Inhis own words, “social unity (...) as deriving from anoverlapping consensus on a political conception of justicesuitable for a constitutional regime. (...) is the mostdesirable conception of unity available to us; it is thelimit of the practical best.” (PL: 201-2). Rawls is infact consistently ambivalent about the status of a “modusvivendi”: is it a kind of “social unity and stability”,albeit a morally inferior one; or is it too thin and toodependent on a contingent balance of power to qualify as“stable” at all? I think it is easy to show that in PLRawls clearly prefers the second answer. The ambivalenceis on the other hand more marked in RH, where Rawlsretreats slightly from some of the less plausiblepositions criticized here.

War, politics, and sovereign violence 23produces conflicts of comparable severity. Neither issatisfactorily established.

With respect to the first assumption, even grantedthe existence of fundamental doctrinal oppositionbetween at least some groups in any given society, itdoes not follow that civil war is just around thecorner. Despite the massive historical significance ofthe post-Reformation European wars of religion, theyare just an illustration from societies of a verydifferent kind,22 and do not constitute proof. Rawlshimself appears ambivalent on this point, and seems tohesitate in PL between a robust view of politics and astrongly anti-political reference point of a societyentirely without conflict, expressed by his constantconcern to insulate justice from politics (”in thewrong sense”), and to distinguish sharply justice asfairness from a “mere” modus vivendi. In the absence oftheory of belief and motivation, the conflictualimplications of comprehensive doctrines are simplyindeterminate.

The case is if anything stronger with respect tothe second assumption. One way of looking at the issueis to consider the idea of “circumstances of justice”,a phrase used in TJ (and carried over into PL) torefer precisely to the basic problem the theory ofjustice is designed to solve: “although society is acooperative venture for mutual advantage, it is

22. Whether they even illustrate Rawls’ point is in factarguable. The theological differences between, say,Anglicans and Catholics in late 17th century England wereminimal, and their political outlooks in nearly allrespects were quite compatible. There was certainly farless difference than between either and the Puritans. Thereal issue was geopolitical. Catholics were regarded astraitors because of (supposed) subservience to the interestsof France or Spain, even more than as heretics. Converselytolerance had as much to do with a diminished sense ofexternal threat (especially after 1745) as with internalpolitics. Mutatis mutandis, the difficulties faced byEuropean Muslims may be analyzed in a similar way.

24typically marked by a conflict as well as by anidentity of interests. There is an identity ofinterests since social cooperation makes possible abetter life for all than any would have if each wereto live solely by his own efforts. There is a conflictof interests since persons are not indifferent as tohow the greater benefits produced by their collabo-ration are distributed, for in order to pursue theirends they each prefer a smaller to a larger share.”(TJ: 4). The conflict of interests referred to is apolitical problem requiring a political solution.Rawls provides a series of convincing reasons forbelieving that the political management of socialcooperation would be more effective, and more widelyacceptable, if it was based on principles incorpora-ting an ideal position of impartiality outsidepolitics. It is important, however, (and entirelyconsistent with Rawls’ theory) to stress that theissues are not thereby removed from politics, nor arethe underlying reasons for conflict somehowdissipated. The precise extent to which, in variouscircumstances, class conflict, say, may be a threat tosocial order is an important question of politicalsociology. Rawls’ failure to address it makes hisemphasis on the primacy of doctrinal conflict seemmerely dogmatic.

A complementary angle is provided by Rawls’insistence on circumscribing the domain of justice byrationality (the moral power to define conceptions ofthe good) and reasonableness (the willingness toaccept fair terms of social cooperation). From theperspective of a theory of justice this limitation isperfectly understandable. If all established positionsand all forms of belief and behaviour deserve equalrespect, any current injustice is simply entrenched.23

The view of social and political conflict that itproduces is however so thin-blooded that it is hardlyrelevant at all. Accepting, as a constraint on justiceas it relates to social order, the status quo with

War, politics, and sovereign violence 25respect to the actual distribution of comprehensivedoctrines within society, including the weird, thewacky and the downright nasty, is a recipe forvacuousness, but so is failure to take account of thecircumstances that create the problem one is trying tosolve. The distinction is a matter of judgement, towhich Rawls gives insufficient attention. “prejudiceand bias, self- and group interest, blindness andwillfulness, play their all too familiar part inpolitical life. But these sources of unreasonabledisagreement stand in marked contrast to thosecompatible with everyone’s being fully reasonable.(...) It is unrealistic – or worse, it arouses mutualsuspicion and hostility – to suppose that all ourdifferences are rooted solely in ignorance andperversity, or else in the rivalries for power,status, or economic gain.” (PL: 58) In other words,even if people were in fact reasonable and rational,political conflict would still occur. No doubt this istrue, but unless it can be shown that in a more justsociety people would be less prejudiced, biased andblind – and Rawls drops in PL his previous attempt toargue for this (TJ: 453-512) –, it is not veryinteresting. Removing grounds for reasonable conflictsimply leaves unreasonable conflict as the only threat.Perhaps this is not really a problem of justice. Fromthe point of view of social order, however, it is verymuch the problem. It is hardly controversial tosuggest that civil peace may be threatened byunreasonable people with guns, and that no sharedprinciples of justice are likely to be available tojustify restricting their activities. Indeed,unreasonable people without guns are just as much of aproblem: less dangerous, perhaps, but more difficultto deal with in an approximately just society.

What is at issue here is ultimately whether theconcept of “reasonableness” – and therefore theunderlying idea of “reason” – can do the work Rawlsintends it to. Certainly his own ambivalence hardly

26helps his case. How does the idea of a “comprehensivedoctrine” express “rationality” – the moral power todefine conceptions of the good? In Isaiah Berlin’swell known version of moral pluralism, moral knowledgeis inherently limited by value pluralism, which isneither individually nor collectively subjective, butobjective, so that the idea of a “comprehensivedoctrine” is incoherent. Rawls is aware of this, andindeed pays lip service to it (PL: 197). As he says,“Only ideologues and visionaries fail to experiencedeep conflicts of political values and conflictsbetween these and nonpolitical values.” (PL: 44). Buthe is uncomfortable with the idea, and elides itsimplications. Ultimately Rawls’ pluralism is quasi-monistic, in the sense that it postulates a pluralityof value monisms, and he apparently speaks for himselfwhen he says “that there are (...) many reasonablecomprehensive doctrines affirmed by reasonable peoplemay seem surprising, as we like to see reason leadingto the truth and to think of the truth as one.” (PL:64). Skepticism about one’s beliefs is in this respectan important issue, and Rawls is ambivalent betweenthe (plausible) view that no form of politicalliberalism can legitimately demand skepticism and the

23. This is the point of Joshua Cohen’s article “Moralpluralism and political consensus” [in David Copp et al.(ed.), The idea of democracy (Cambridge: Cambridge UniversityPress, 1993)], which was a critique of Rawls’ “The domainof the political and overlapping consensus” (New YorkUniversity Law Review, 64(2) (1989): 233-55; also reprinted inCopp et al., The idea of democracy, 245-69), in which thedistinction between reasonable pluralism and pluralism toutcourt was not made. It was directly responsible for Rawls’adoption of the idea of reasonable pluralism in PL. Asimilar critique of the principles of justice in TJ wasmade by Brian Barry [The liberal theory of justice (Oxford:Clarendon Press, 1973)], on the grounds that the“difference principle” takes greed, which is unjust, forgranted and implicitly validates it. This, however, is nota criticism that Rawls has incorporated.

War, politics, and sovereign violence 27(unsubstantiated) view that beliefs held skepticallycannot really be moral beliefs at all. There is noroom in Rawls for “extreme views weakly held”,24

although many people would I think find them familiar– and perhaps congenial. While Rawls is not consistenthere, one aspect of what he is saying is that onlyideologues and visionaries subscribe to comprehensivedoctrines. If so, if rationality is ultimatelyincompatible with reasonableness, much of what isdistinctive about the later Rawls simply evaporates.

Such inconsistencies are hardly trivial, but it ispossible to interpret them more charitably. Let ustherefore redefine the problem – much more narrowlythan Rawls himself – as the threat to a just socialorder posed by doctrinal differences betweenreasonable people. The characteristic postulates thatother threats are less significant and that non-moralsocial orders are inherently unstable are notessential to this formulation. The familiar solutionoffered by Rawls is the idea of an “overlappingconsensus”. However, Rawls fails to prove theeffectiveness of an overlapping consensus in promotingsocial and political order; and furthermore thepolitical efficacy of liberalism in this respect, in-cluding the kind of consensus it involves, derivesprecisely from those features that Rawls feels mostuneasy about defending.

The idea of an overlapping consensus arises in PLat the conjunction of two requirements: thatprinciples of justice should be stable,25 in other

24. A self-description reportedly given by the Englishhistorian A. J. P. Taylor [Alan Watkins, Brief lives (London:Hamish Hamilton, 1982): 178].

25. As we have seen, “stability” may also characterizethe society regulated by principles of justice. The extentto which this elastic usage is deliberate and legitimateis unclear. On the resulting difficulties, see furtherDavid Copp, “Pluralism and stability in liberal theory,”Journal of Political Philosophy, 4(3) (1996): 191-206.

28words that their implementation should promote a senseof justice, and should therefore not be self-defeating; and that they should actually be acceptableto all reasonable citizens. However these ratheruncontroversial requirements are also, in a furthertwist, subject to the logic of a “well-ordered”society, i.e. one “effectively regulated by a publicconception of justice” (TJ: 5). The reason for thetwist is Rawls’ sharp distinction between a (non-moral) modus vivendi and an overlapping consensus “moralin both its object and its content” (PL: 126).

Rawls defines a modus vivendi as any set ofinstitutions supported by the conjunction of privateadvantage rather than by public argument. He advancestwo distinct reasons for his vehement rejection ofsuch an idea. On the one hand, he expresses skepticismabout its effectiveness in principle: “social unity isonly apparent, as its stability is contingent oncircumstances remaining such as not to upset thefortunate convergence of interests.” (PL: 147). On theother hand, by definition, no modus vivendi can be thefocus for a well-ordered society. Presumably they areintended to reinforce each other, but the first is infact vacuous: no one would seriously suggest that anysociety could be maintained solely by its members’individual material satisfaction with their share ofthe fruits of social cooperation. The objection toviewing justice in terms of a modus vivendi thereforedepends entirely on the theoretical persuasiveness ofthe idea of a well-ordered society. It was initiallyderived from the “Aristotelian principle”, but this isvalid only within comprehensive liberalism. Within theterms of PL, therefore, and for the purposes of thisdiscussion, the idea of a well-ordered society must bejudged by its practical implications. There are threedifficulties in this respect. First, Rawls’ emphasison publicity (PL: 66-67) ultimately commits him to thetransparency of the political sphere as a criterion ofa good society. Yet, on his own terms (PL: 133-172),

War, politics, and sovereign violence 29the same political principles may be supported bydifferent reasons from the perspective of the variousavailable conceptions of the good – although therequirement that such reasons be articulated publiclylimits the extent of difference. Furthermore, trans-parency is not simply unnecessary, but potentiallyself-destructive. So long as it is assumed that peopleentertain serious disagreements (and otherwise Rawls’whole exercise would be pointless), prudence wouldseem to advise against bringing it all out into theopen. The limited but genuine success of currentpolitical transitions in South Africa and NorthernIreland suggests that there is more to be said foropaque and immoral political fudges than Rawls isprepared to envisage.26

Secondly, the effectiveness of an overlappingconsensus in reducing political division is notsatisfactorily established by Rawls. The general idea– which is plausible enough – is that people withdivergent comprehensive views should be able equallyto accommodate a common political perspective. Rawls’distinctive contribution, however, is to make thecontent of the consensus moral and to restrict itsscope to those comprehensive views eligible on groundsof reasonableness. Yet the capacity of such aconsensus to be politically effective is impossible toevaluate in the absence of an analytical theory of thecauses of social conflict, which Rawls completelyfails to provide.

Finally, a characteristic feature of PL is Rawls’suggestion that the idea of an overlapping consensusis simply a more systematic and sophisticated formu-lation of something that is already available withinthe practical political tradition of the West, namelythe principle of tolerance. His attempts to prove this

26. For a broader theoretical perspective, see JohnCrowley, “La transparence du politique et ses limites:négociations d’intérêts et pluralisme ‘moral’,” Politique etSociétés, 17(3) (1998): 37-58.

30are unconvincing. Thus, in the pre-liberal era, hesuggests, “Intolerance was accepted as a condition ofsocial order and stability. The weakening of that be-lief helps to clear the way for liberal institutions.Perhaps the doctrine of free faith developed becauseit is difficult, if not impossible, to believe in thedamnation of those with whom we have, with trust andconfidence, long and fruitfully cooperated inmaintaining a just society.” (PL: xxv) This suggeststhat free faith involves no change in the nature offaith itself. Yet the historical development ofliberalism is inconceivable without the decline inreligious faith characteristic of the skeptical anddeistic 18th century. “Belief” in intolerance did notsimply decay in a vacuum. By contrast it is strikingthat early understandings of (legal) toleration were –in Britain at least – not the opposite of but premisedupon doctrinal intolerance.27 Conversely it isarguable, as a matter of historical interpretation,that legal toleration was a condition for thedevelopment of doctrinal tolerance rather than a con-sequence of it. Again, the link between socialcooperation and belief in damnation seems merelyfacetious. After all, some religious doctrinesprescribe belief in the damnation of the majority ofone’s coreligionists, and to the extent that such a beliefis socially self-contradictory, it tends as much tothe erosion of faith as to “free faith”. Finally,toleration was born in societies that by no standardrelevant to Rawls’ discussion could be regarded asjust.

A different but perfectly plausible interpretationof the historical development of liberalism, along thelines sketched in the previous paragraph, would infact be incompatible with the whole thrust of Rawls’theory. In PL, Rawls considers various possibleobjections to the idea of an overlapping consensus.

27. Iain Hampsher-Monk, “Is there an English form oftoleration?,” New Community, 21(2) (1995): 227-40.

War, politics, and sovereign violence 31One that he clearly believes would, if valid, bedamaging is that it necessarily implies skepticismabout the truth of comprehensive doctrines, and isthus incompatible with the holding of any suchdoctrine. His statement, and rebuttal, of thisobjection involves however the fallacy, explicitlyendorsed in PL, that “belief” is a purely cognitivenotion.28 In fact, convictions have degrees of com-mitment attached to them, not simply degrees ofintellectual certainty, and the practical effect ofskepticism is usually diffidence with regard toimplementation rather than intellectual doubt. Indeed,if toleration is, plausibly enough, the “only workablealternative to endless and destructive civil strife”(PL: 159), what makes it work is ultimately thefeeling that things may be true in some sense withoutbeing worth killing for, whereas in previous centuriescommitment and belief were inseparable. Thepossibility that rationality might ultimately beunreasonable (using both words in the Rawlsian sense)is within the logic of PL, but Rawls refuses to admitit. He does of course recognize that liberalism haseffects on society, to the point, ultimately, ofturning many or most people into liberals of somekind. What he is not prepared to accept is that theseeffects, because non-liberals clearly have no reasonto endorse them, might provide a justification forliberalism. However, contrary to what he suggests, the

28. As he puts it: “it is vital to the idea of politicalliberalism that we may with perfect consistency hold thatit would be unreasonable to use political power to enforceour own comprehensive view, which we must, of course, affirmas either reasonable or true.” (PL: 138; my italics). Thesame point crops up in much of the current literature onmoral pluralism, and seems to have axiomatic status. Yetit is hardly self-evident. Of course, “I believe that x”is trivially equivalent to “I believe that x is true”. Butthis is not what is relevant here. Sentences of the form“I believe in x” cannot generally be rephrased as “Ibelieve that y is true” without loss of content.

32historical evidence of the development of liberalismdoes not support the view that a derivation ofliberalism acceptable to non-liberals is available.

A similar tendency to set the limits of theory byreference primarily or exclusively to justice isapparent in Habermas. Although Habermas presents hisdebates with Rawls as a “family quarrel”, theirtheories actually differ quite sharply. In particular,while Habermas does start from value plurality as asocial fact, the ethic of discussion is not reallypluralistic, and certainly does not postulate theincommensurability of values. (Arguably, of course, ifmy earlier suggestion is correct that Rawls, despitehis Berlinian references, is some kind of closetmonist, the difference may not be so sharp. This is acentral theme in their recent debate, Habermas re-garding their positions as closer than Rawls isprepared to accept.) Furthermore, he explicitly dis-claims a close theoretical link between justice andsocial order, or stability in the modified Rawlsiansense of PL. In so doing, he obviously denies thatjustice based on reasonable agreement is a sufficientcondition of order, although it is fairly clear thatdoes envisage some kind of necessary relation. Howeverthe way in which he does so is revealing. The extentto which justice may contribute to social order is, hesays, an empirical question – about which theory haslittle to say (Débat sur la justice politique: 41). Socialorder is thus ultimately not “thinkable” but ratherthe condition prior and radically external to rationalthought that makes politics and political theorypossible. (Habermas does not say this, but thecombination of his silences and what he does sayimplies it.) In times of civil war, therefore – orsimply of severe social conflict involving, throughbad faith or other mechanisms, the utter failure tocommunicate –, theory can only be silent.

Habermas is understandably unhappy with Rawls’unsatisfactory conflation of justice and social order

War, politics, and sovereign violence 33via the notion of stability. But in denying that orderis an appropriate subject for theory at all, heremains within the same format. “Reason” and“violence”, as he presents them in passing at thebeginning of Faktizität und Geltung – without justificationor expansion – are polar opposites. There is no spacehere for violence as a mode of discourse, or for theoverlap or politics and war, that characterize mostpressing political problems. The bounds of reason areset by “serious” participation in discussion, therebytacitly assuming, in the formal structure of thetheory, the motivations, the “sense of justice” in theRawlsian sense, that are supposed to be internallygrounded. The best indicator of this difficulty is theundue emphasis Habermas is forced to put on thedevelopmental psychology he borrows from Piaget, andlater from Kohlberg. The link between thetranscendental conditions of public discourse and theempirical, socio-psychological disposition to engagein it is ultimately a spectacular leap of faith.Habermas is thus vulnerable to the same charge asRawls: to put it very crudely, that he offers atheoretical approach to justice that only worksagainst the kind of institutional background – itselfa contingent and fragile historical achievement – thatmakes such an approach, if not strictly unnecessary,at least optional. Justice in approximately just so-cieties, so long as they are approximately insulatedfrom a deeply unjust world, is an important moralissue, but not, by definition, a burning practicalone.

Foundation, justification, and compliance

These internal difficulties in Rawls and Habermas,which could be equally illustrated from a wide varietyof other thinkers, show how the journalistic clichésabout “politics” and “war” have a close parallel in

34contemporary theory. To put things in more abstracttheoretical terms, what is ultimately at stake is theambivalence at the heart of the notion of“compliance”, which seeks to conjoin two ideas: thefact of acceptance (of authority generally or ofspecific instructions), and the reasons for it. Thepresumption, implicit in Rawls, Habermas and manyothers, is that there is a close internal connectionbetween the two: that ungrounded acceptance (say defacto “social peace”) is different in practice (“lessstable”), and not simply morally, from acceptancebased on recognized legitimacy. Conversely, the moralcharacter of social institutions contributes to theirpractical stability. There is nothing wrong with thisas a normative aspiration. As a sociological analysis,however, it is less plausible than it appears at firstsight.

The participants in any activity may have a widevariety of reasons for following its rules. Does suchdiversity make a difference, or is the fact of com-pliance of independent significance? It is easy enoughto agree that identification of a rule as a rulerequires evidence of compliance over time in a rangeof circumstances, and some grounds for postulatingfuture compliance. In other words, reasons forcompliance are constitutive of rules, and not simplyincidental to them. In particular, the ability ingeneral to punish non-compliance depends on theexistence and plausibility of such reasons.

From the point of view of political theory, lawsare a highly significant category of rules (taking lawin a broad sense, regardless of, say, the detailedhierarchy of legal norms). Let us therefore considerthe question: how do the reasons for compliance withlaw – the basis of civil, or political, obligation, asthey are often termed – relate to actual compliance?It is immediately apparent that this question is inprinciple indeterminate unless the nature of theactivity presumptively regulated by law is further

War, politics, and sovereign violence 35specified. This may not seem to be of great practicalrelevance. Yet a glance at contemporary theory revealsa tension between two distinct, but conceptuallyrelated, answers. On the one hand, the centralactivity to which law pertains is considered to be therealization of social justice, following Rawls’ dictumthat “justice is the first virtue of socialinstitutions, as truth is of systems of thought.”. Theissues are, therefore, the nature of justice, thereasons for behaving justly, their capacity to producejust behaviour, and the implications of their failureto do so. On the other hand, law is theorized in termsof social order, the key issues being mutatis mutandisidentical.

What makes this tension interesting is preciselythat the two sets of problems are not strictlydistinct. A theorist of order could perhaps, inprinciple, disregard justice, although even here thepossibility that unjust arrangements might beinherently unstable needs explicit discussion andcannot be dismissed out of hand. A theorist ofjustice, however, is inescapably confronted with theproblem of order. Justice is, indeed, a kind ofordering, in something like the mathematical sense.Specifically, a set of just arrangements ischaracterized, among other things, by patterns ofmotivation and behaviour that are, at the very least,widespread; and by institutions that effectivelypermit the correction of injustice. Human rights, forinstance, require application – often involving force– and not simply enunciation. How rules might becomesocially embedded and effective is thus the keyproblem in any theory of justice. In practice, theo-ries can broadly be divided into those for which orderis some kind of background condition (of a “cultural”nature, for instance), and those that regard order asderiving from the operation of the rules that itunderwrites (what scientists might call a “bootstrap”structure).

36The bootstrap version, which is clearly the more

theoretically ambitious, is characteristic of the mostinteresting contemporary theories. Justice and orderthus emerge simultaneously, and the principles offoundation and justification must in some sense converge. Inother words, whatever might in principle be regardedas justifying some social arrangement must also beassumed to sustain it in practice.

The ambivalence about the nature and status of lawnaturally extends to compliance, and thus brings usback to our starting point. It would be tempting, butwrong, to distinguish neatly between the perspectiveof social order, from which only the capacity toenforce compliance matters, and the perspective ofjustice, from which the ability to justify suchenforcement is vital. In fact the two issues blur,because the distinction depends on the theoreticallyindeterminate question of the reasons for non-compliance. We may agree with the currently standardliberal paradigm that actual or latent conflict inwhich all parties are “reasonable” can be fullyresolved by reference to justice. But what makespeople reasonable, and what are the consequences oftheir not being reasonable?

A fairly standard answer, which Habermas and Rawlsin particular both reflect without being entirelycomfortable with it, is that reasonableness only makessense within a set of established and generallyaccepted institutional constraints, which in turn makejust coercion possible. But what this means is thatjustice flourishes, if at all, in the shadow ofviolence. Foundation in principle may be consistentwith justification, but actual historical origin isgenerally unjustifiable. This is the core ofFoucault’s critique of the social-contracttradition.29 Considering the relation between politicsand war, Foucault seeks to show how a theorist who isoften assumed to have addressed the issue explicitly

29. Foucault, Il faut défendre la société.

War, politics, and sovereign violence 37(Hobbes) in fact evades it. Although war – in thepeculiar Hobbesian conception of the “state of war”,where actual fighting is almost immaterial – is con-ceptually linked to the origins of political order,all trace of actual conflict is removed within theCommonwealth. Foucault generalizes this observation tothe social-contract tradition generally, which heviews as handicapped by its legalistic conception ofpower, equated with sovereignty and in turn with theenunciation of law. On the contrary, the political-historical writers that Foucault analyzes, and that herelates to the conceptual origins of racism, regardlaw as a decorative structure behind or beneath whichthe traces of past war, and its present-daycontinuation, are clearly visible if one chooses tolook. While he does not exactly endorse this view, hedoes regard it as more consistent with his ownperspective on power than the legalistic traditionwhich remains the mainstream within political theory.The social contract, as Foucault puts it, is signed,if at all, in blood.

And even this understates the problem, since a justorder must, in principle, be founded anew each time itaddresses non-compliance, which cannot be presumed tobe reasonable. Thus, the justification of violence isnot an incidental or occasional feature of politicalauthority, but something very close to the inherentnature of justice itself. No system of law operates byreason alone – otherwise it would not be necessary atall. Violence and reason, which Habermas and Rawlsdistinguish strictly, therefore have, at the veryleast, an inherent instrumental relationship within,and not simply as a derogation from, the idea of jus-tice. War is an aspect of politics, and not itsnegation. Thus, contemporary liberalism is haunted bythe ghost of sovereignty – in the strong Schmittiansense that Agamben appropriates for rather differentpurposes – and yet is entirely unequipped to exorciseit. The sword of justice is an inherent attribute of

38the sovereign because it is quite simply what justicemeans in an unjust world. Yet the swish of that swordis precisely the point at which war and politicsmerge, and justification and foundation part company.