WARLORDS AND STATES IN TIMES OF CIVIL CRISIS

Although warlords may emerge even in relatively stable countries, they tend to thrive in situations of civil war and state weakness, failure, or collapse, where they are able to establish control over more or less geographically demarcated territories while maintaining autonomy from the states in those territories lie (Duffield Reference Duffield1997). Warlords take different forms in different settings, all of which are characterized by “multiple and overlapping centers of sovereignty” (Duffield Reference Duffield1997, 97): paramilitary groups in Colombia (Duncan Reference Duncan2006; Ronderos 2014), “strongman governors” in Afghanistan (Mukhopadhyay Reference Mukhopadhyay2014), dozos in Cote d'Ivoire (Hellweg Reference Hellweg2011), and clan leaders and elders in Somalia (Menkhaus Reference Menkhaus2007), among other examples.

The term “warlord” is an unfortunately loaded one. It has become an “ugly, detrimental expression, evoking brutality, racketeering and terrorism” (Jackson Reference Jackson2003, 131). Warlords have been characterized as “brutal” and “parochial,” and as a “plague” that “deprives countries of the chance for lasting security and economic growth” (Marten Reference Marten2007, 41). They are thought to be archaic, “prior to the state” both historically and analytically. They are “forms of social organization that the state was meant to supplant,” by nature incapable of using force justifiably (Ahram and King Reference Ahram and King2012, 173; Gross Reference Gross2015, 37–45).

Yet warlords are not always so dissimilar from states themselves (Duffield Reference Duffield1998). Warlords, like states, are capable of organizing and deploying collective violence. Like states, they make and enforce rules, adjudicate disputes, collect taxes, exploit natural resources, and, in many cases, provide social services. In so doing, they may even appeal to their “subjects” in the name of a “shared community or destiny” (Hansen and Stepputat Reference Hansen, Stepputat, Hansen and Stepputat2001, 16). They are, like states, an “attempt to re-establish stability within anarchy” (Jackson Reference Jackson2003, 147).

Our aim is not to glorify or romanticize warlord rule. Many warlords are violent, corrupt, and exclusionary, by no means deserving of recognition as legitimate authorities. Many commit human rights abuses or traffic drugs and weapons, threatening the stability not only of their own countries but also of neighboring ones. Yet the same or worse is often true of the states to which warlords are nominally subject. In some cases warlords may surpass the state in the quality of governance they provide, and may do so with significantly less oversight or support from international intermediaries, which by default tend to favor the state against its nonstate competitors (Luck Reference Luck2010). As we will see later, in Afghanistan's Herat Province, the warlord Ismail Kahn provided social services unrivaled anywhere else in the country, including Kabul. In Puntland, an autonomous but nonsecessionist state in Somalia, clan leaders and elders established the most functional government in the country, contrasting markedly with the central government's “inorganic, top-down state-building projects” (Menkhaus Reference Menkhaus2007, 82). In Cote d'Ivoire, dozos provided security and restituted victims of crime even as the state “all but abdicated its law-preserving function” (Hellweg Reference Hellweg2011, 18). Unlike states, warlords perform these functions without formal or legal recognition either by the state or international organizations.

To fix ideas, consider the following stylized case. Assume that S is an internationally recognized state that has been debilitated by civil strife, in the context of which it has either committed gross human rights violations against a significant portion of its population or has failed to protect a significant portion of its population from gross violations committed by others. Assume that W is a warlord that has succeeded in establishing order and a near monopoly of force over a community C in territory T. T lies inside S, but does not have internationally recognized borders of its own; it is, however, demarcated in such a way that its residents—those whose lives and livelihood are rooted there—can be distinguished from those of other territories. C is thus a “political community” in the sense of Geuss (Reference Geuss2001, 15–19): It occupies a well-defined region (T) and is ruled by an agent (W) that is committed to and claims responsibility for maintaining social order among C members.

Assume, further, that S and W have conflicting claims to rule over C. This conflict may take many forms, including disputes over taxation, conscription, adjudication, and the like. Whatever the content of the dispute, S claims the right to depose W and (re-)gain jurisdiction over T, by force if necessary. W, in turn, claims the right to repel S and retain jurisdiction over T, also by force if necessary. Intuitively, if S has a record of hostility toward C and of failing to protect C members from internal or external threats—or if W, by contrast, has a record of reliably providing security in C, even amid civil strife—then the validity of W's claim seems plausible at least. Under what conditions, if any, might the claim be valid?

In answering this question, we limit the scope of our analysis in two ways—one related to S, the other to W. First, as suggested earlier, we assume that S has proven incapable of, or unwilling to, secure social order and protect basic rights in T and that latent threats to C persist, generated either by S or by third parties. In these contexts, we argue, S forfeits its right to rule over C, thus opening the way for other claimants to legitimate authority. Second, we focus on cases in which armed nonstate groups seek autonomy without attempting to capture or secede from the state. Given the upheaval that secessions and state-capturing civil wars cause for civilians and for neighboring countries—not to mention for the states involved in them—we assume they need to be justified according to more stringent conditions than the ones we develop here.

These scope conditions are not as narrow as they may seem. In Afghanistan, warlords exploited the collapse of the state in 2001 to consolidate territorial control, but in the decade following the U.S. invasion, none advocated secession or explicitly rejected the (potential) reemergence of a strong central government (Barfield Reference Barfield2011). During a period of “partial state collapse” in Colombia, paramilitaries achieved “politico-military hegemony” within local “proto-states,” but never attempted an “absolute taking of power within the state itself” (Duncan Reference Duncan2006, 24). In Somalia—arguably the most severe case of state collapse in the world—Puntland has long sought autonomy from the central government while consistently maintaining allegiance to a unified Somali state (Menkhaus Reference Menkhaus2007). In these and other cases, warlords aim neither to capture nor to secede from the state, but rather to preserve autonomy within it. It is in these settings that we wish to assess the nature and extent of W's coercive rights.

PUBLIC AUTHORITY AND THE USE OF FORCE

Contemporary just war theory, which has mostly focused on the rights of states and state-seeking secessionists, provides little analytical traction for understanding the normative structure of warlordism. We begin, instead, from late medieval and early modern accounts of just war, produced during the early stages of the consolidation of sovereign statehood (Osiander Reference Osiander2001). Although these accounts are too dated to directly provide a theory of warlords’ rights, the conditions of contested sovereignty and competing jurisdictional claims under which they were created are not so unlike those of contemporary weak and war-torn states. Indeed, some scholars have described politics in these states as “neo-medieval,” characterized by “competing institutions” and multiple “zones of authority with overlapping boundaries” and no “universal centre of competence” (Duffield Reference Duffield1998, 70).Footnote ² Late medieval and early modern accounts were designed to clarify the scope of just war under circumstances much like these.

These accounts yield two powerful insights for our purposes. First, they distinguish between two types of war-making powers: defensive and “offensive” (Neff Reference Neff2005, 59–62; Reichberg Reference Reichberg, Rodin and Shue2008, 196–98). Although virtually all accounts from this period endorsed a universal right to use defensive force for the sake of self or others, the scope of this right was extremely limited. Defensive war was justified only in the face of imminent or ongoing violations of right, and only the minimum amount of force sufficient to avoid or stop the violation was permissible. If an injury had not materialized or a threat was not imminent, it was mandatory to wait or appeal to public authorities, rather than to take defense into one's own hands (see, e.g., Grotius Reference Grotius1925, 173; Vitoria Reference Vitoria1991, 300).

Offensive wars, in contrast, could be waged for purposes beyond these limited and short-term goals.Footnote ³ They were “fought for the public weal,” which “set [them] apart from wars fought over purely private issues” (Keen Reference Keen1965, 78). Their defining purpose was to protect the public good and safeguard social order in the longer run; as such, their powers included actively seeking redress and punishment for past injuries even deposing foreign rulers and annexing their territories if necessary to consolidate future peace (e.g., Grotius Reference Grotius1925, 184; Vitoria Reference Vitoria1991, 300). Whereas force used strictly in self-defense required no special form of authority, offensive wars were reserved for public powers alone: their goals and permissions were far broader than those of self- and other-defense and presupposed both a distinct concern for the common good and the capacity to mobilize collective resources.

The second, related insight is that influential late medieval and early modern sources incorporated the possibility of de jure fragmented political power within, not just across, customarily recognized boundaries (Barnes Reference Barnes, Kretzmann, Kenny and Pinborg1982, 775–77; Keen Reference Keen1965, 63–81; Russell Reference Russell1975, 100–3, 143–47).Footnote ⁴ In particular, they recognized the possibility that actors other than “supreme rulers” might, on grounds of custom or necessity, have the right to wage offensive war. These actors included rulers of communities who were formally under suzerainty but were autonomous in practice—barons, counts, dukes, the podestà of Italian city-states—or of communities that were not autonomous but whose formal sovereign was unable or unwilling to provide protection and administer justice (Grotius Reference Grotius1925, 97–101; Vitoria Reference Vitoria1991, 300–2). Even pirates and “brigands,” Grotius wrote, who initially “banded together for wrongdoing” and private enrichment could become public powers if their leaders became ministers of justice and the public good, “so that those who have only been robbers upon embracing another mode of life become a state” (Grotius Reference Grotius1925, 631–32).Footnote ⁵

If we abstract from their specific historical context, these late medieval and early modern just war theories leave us with the suggestive proposition that in situations of contested sovereignty, armed nonstate actors with a public orientation—i.e., a credible commitment to protect the existence and rights of a political community—may justifiably use force not only to stop ongoing or imminent serious violations of right but also, more broadly to secure public order and reduce citizens’ vulnerability in the longer run. In adapting this proposition to contemporary settings, we must ask under what conditions, if any, warlords might qualify as public authorities of this sort, justified in using not only strictly defensive but also offensive force to protect the communities over which they claim the right to rule.

Contemporary literatures offer limited guidance in answering this question. The prevailing view in the empirical peace-building and state-building literatures overwhelmingly favors statist claims to (re-)establish uncontested control over territory and excludes in principle the validity of a warlord's counterclaim (e.g., Fukuyama Reference Fukuyama2004; Reference Fukuyama2005; Rotberg Reference Rotberg and Rotberg2003; Zartman Reference Zartman and Zartman1995). A statist bias is also prominent in contemporary international humanitarian law, which limits the status of lawful combatant to state security forces alone—a framework that “privileges [states’] own police and armed forces against those who oppose them” (Fleck Reference Fleck and Fleck2008, 613).

Contemporary just war theorists are divided on the question of legitimate authority, but are similarly unhelpful for assessing the scope of warlords’ rights. In keeping with the early moderns, some have endorsed a universal right to self- and other-defense in cases of imminent or ongoing massive violations of right; given the urgency of rescue, no special authority should be required in such cases (Boyle Reference Boyle, Nardin and Williams2006; Fabre Reference Fabre2012; Finlay Reference Finlay2010; Reitberger Reference Reitberger2013). Others, including promoters of the doctrine of responsibility to protect (R2P) and some proponents of humanitarian intervention, have defended a more restrictive view. According to their institutional accounts, the ultimate bearer of other-defensive rights should be the “international community,” ideally in the form of a coalition of UN-authorized member states. If a state proves unwilling or unable to protect its own population, then the UN must step in and coordinate state-based efforts to rescue populations under threat or attack, even to the point of using offensive force to stabilize crisis situations (ICISS 2001; Roberts Reference Roberts and Welsh2004; Welsh Reference Welsh, Knight and Egerton2012).

Neither of these approaches responds adequately to our stylized case. On the one hand, R2P doctrine generally views warlords as threats, potentially serious enough to validate international intervention, rather than as legitimate sources of defensive (much less offensive) force. This view is inconsistent with the doctrine's own grounding of authority in the de facto capacity to provide protection to populations at risk.Footnote ⁶ The doctrine, moreover, applies only to the most extreme and serious cases of violations of right; for more distant or less serious threats, it affords states the freedom to disrupt effective and already existing forms of nonstate governance or to attack and harass citizens as long as such abuses do not rise to the level of a “core international crime.” On the other hand, universalist just war accounts of defensive rights do not discriminate among potential bearers of other-defensive rights, and therefore they neglect the normatively relevant features of the relationship between C, W, and S.

With few notable exceptions, contemporary just war theorists have thus failed to examine the use of offensive force by nonstate actors.Footnote ⁷ Fabre and Finley do concede that nonstate actors might be justified in using force in the absence of ongoing massive violations of human rights; for instance, in response to serious but distant threats or in pursuit of legitimate political goals. But in the absence of ongoing mass atrocity, they both argue that nonstate armed groups are justified in using offensive force only when there is an explicit mandate to that effect from the community for the sake of which they claim to act (Fabre Reference Fabre2012, 154–56; Finley Reference Finlay2010). We argue that such case-by-case authorization would be excessively onerous under the circumstances of civil war and state collapse. Condemning a warlord who succeeds in pacifying a territory but lacks an explicit communal mandate would amount to putting democratic values ahead of the protection of basic rights.

Of course, determining whether or not a particular armed nonstate actor might have standing to wage offensive war is a politically and theoretically contentious matter. Authority to wage offensive war includes a normative power to decide what coercive actions may be required to protect a political community, in addition to and beyond the urgent imperative of rescue. This power, we argue, entails the capacity to alter other agents’ rights and obligations and, in particular, creates an obligation in S and other threatening agents to abstain from entering T, as well as rights in W to collect taxes and recruit militia members. The question, then, is how to decide whether a given warlord is entitled to such a power. How can we know if W is committed and able to protect the interests of C members? How can we determine whether W constitutes a public authority, capable of deciding under uncertain circumstances when force on behalf of C is warranted?

TWO CRITERIA OF LEGITIMACY FOR WARLORDS

We defend a minimalist answer to these questions based on two criteria inspired by, but significantly weaker than, the conditions posited as necessary for states to legitimately defend their territorial rights through force (e.g., Stilz Reference Stilz, Fabre and Lazar2014). When a warlord W satisfies these two criteria, we argue, states and international organizations should recognize W as a public authority in T on a limited and provisional basis, with implications that we discuss in the following section.

First, to be considered a legitimate authority, W must guarantee the basic right to security of C members. According to Henry Shue's influential construction, this is the right “not to be subjected to murder, torture, mayhem, rape, or assault” (Reference Shue1996, 20–21). The capacities (armed and otherwise) required to make such a guarantee will vary across contexts, but must at least be sufficient to maintain public order and be proportional to the capacities of potential aggressors such that W can preempt foreseeable threats to C. W's capacities must also be subject to self-disciplining mechanisms to ensure that they are not used against C members themselves.Footnote ⁸ Although this criterion is primarily oriented toward the present and future, it excludes warlords that rose to power by systematically committing gross human rights violations. W cannot plausibly claim legitimate authority by perpetrating gross violations that would cause S to lose its right to rule in the first place. Engagement in such violations, moreover, undermines W's credibility as a protector of basic rights.

Second, W must have popular support within C, or “social legitimacy” in the sense of Geuss (Reference Geuss2001, 31–37). This means that C members generally comply with W not (or not only) because of the threat of coercion, but because they interpret W's commands as having some “rudimentary normative force” (33). Popular support may be grounded in long-standing customary or traditional practices; may derive from ascriptive ties of ethnicity, religion, or language; may be a byproduct of the social order that emerges during civil conflict; or may be the result of mechanisms deliberately designed to elevate W's status among C members. Whatever its source, popular support denotes something more than reluctant acquiescence) It reveals some degree of “fit” between W and C (Walzer Reference Walzer1980).

In the extreme circumstances of civil conflict and state collapse, neither S nor W may be perfectly able to guarantee basic rights or generate popular support, so our criteria are best understood as bases for comparison (although a minimal baseline should nonetheless be met, such that the social order W provides would exclude a justified foreign humanitarian intervention). If W is more capable and reliable than S in protecting basic rights and enjoys greater popular support among C members, then W should displace S as the legitimate ruler in T on a limited and provisional basis, with implications we discuss in the next section. W's limited and provisional authority in T can last only for as long as W continues to outperform S according to our two criteria, and then only up to the point in which a negotiated settlement pacifies the relationship between W and S and formalizes W's rule in T.

Although we expect these two criteria to generally correlate positively, hard cases may arise when comparisons conflict. W may have less popular support than S but a greater capacity to protect basic rights, or vice versa. We are reluctant to give lexical priority to either of our two criteria because it seems intuitively clear that extreme differences in one could override minor countervailing differences in the other.Footnote ⁹ In indeterminate cases, when it is impossible to establish whether one actor outperforms the other, we believe a presumption in favor of the state as the default mode of international order and individual rights protection should weigh against W's claim. However, W's accomplishments in protecting basic rights or maintaining popular support imply that S should, at a minimum, refrain from simply eradicating W. Instead it must build on the foundation of public order that W provides—a recommendation that resonates with the small but growing empirical literature on “bottom-up” state-building (Autesserre Reference Autesserre2010; Bowles and Chopra Reference Bowles, Chopra, Call and Wyeth2008; Brown and Gusmao Reference Brown and Gusmao2009; Reno Reference Reno, Call and Wyeth2008).

One might object that our criteria are too permissive. Although we draw on Shue's conception of basic rights, we stop short of requiring his more exacting standard of “subsistence rights,” which would include access to “unpolluted water, adequate food, adequate clothing, adequate shelter, and minimal preventive public health care” (Shue Reference Shue1996, 23). Nor do we require that W guarantee the freedoms of expression and association or uphold the “principle of moral equality,” much less that it “[possess] the capacity to sustain a legitimate state” (cf. Buchanan Reference Buchanan2004, 247–59, 266–68; Stilz Reference Stilz, Fabre and Lazar2014, 218–24). In conflict-torn countries, neither states nor warlords are generally capable of providing all of these goods or protecting all of these rights simultaneously; to insist on their provision as a necessary condition for legitimacy may be normatively appealing, but is empirically untenable.Footnote ¹⁰

Moreover, although we draw on Geuss's concept of social legitimacy, we do not require that W have a formal mandate from C, much less that W be authorized through free and fair elections. Few states meet even the most lenient standards of democratic representation in the wake of civil war, even years after the fighting stops;Footnote ¹¹ the same is almost certainly true of warlords. We submit only that W must be viewed as legitimate by most C members, and we follow several authors in considering that more demanding standards should not be required in contexts of civil strife (e.g. Bhuta Reference Bhuta2010; Doyle and Sambanis Reference Doyle and Sambanis2006; Rotberg Reference Rotberg and Rotberg2003). Such standards tend to conceal the great value of the sheer existence of actors (state or otherwise) capable of protecting basic rights and generating popular support while having their commands obeyed. We view such capacities as sufficient to support claims of legitimate authority, if only on a limited and provisional basis.

Finally, our criteria are conceived to progressively foster forms of rule that exceed the minimal standards we impose. A warlord that can provide public order and protect basic rights while maintaining popular support is in a promising position to make the transition to more representative forms of rule with more demanding rights regimes. Although we do not believe democratic representation and protection of the full array of human rights are viable standards to impose in the midst of civil conflict, once a negotiated settlement between W and S is reached and their state of war ended, latent threats to C should become a lesser concern, allowing more demanding criteria—a point to which we return in the concluding section.

PROVISIONAL TERRITORIAL RIGHTS

Assume that W meets these two criteria of legitimacy. What rights might this entail for W? What duties might it impose on others, including not only S but third parties as well?

Intuitively, if W's legitimate authority is justified in part by the capacity to provide security in T, then W must also have the powers necessary to continue securing T in the future. These include the right to oversee the adjudication of disputes among C members and to regulate the movement of people and goods into and out of the territory. And because the provision of security requires revenue, W must also have the right to tax C members and exploit other sources of funding, including possibly natural resources located in T (however, for the complexities involved see Okowa Reference Okowa2013). Most important for our purposes, W must have the right to organize and maintain a militia and, when necessary, to use force against S, foreign forces acting in support of S, and other armed nonstate groups.

W's use of force in such cases is subject to the same constraints of proportionality and just cause that apply to wars generally. Just causes would include, most obviously, attempts by S or its proxies to forcibly remove W from power. Ex hypothesi, such attempts would threaten the basic rights of C members by subverting the public order that W provides and replacing it with a more brutal or less effective form of rule. These attempts need not constitute violations of right so severe that they would justify an international humanitarian intervention. As noted earlier, W's legitimate authority implies rights that extend beyond strictly defensive war—waged to prevent or stop gross human rights violations in the short term—to include offensive war as well, aimed at securing public order in the longer run. W may, for example, be justified in resisting by force a deployment of S's military to garrisons within T; although such a deployment might not pose a serious threat to C, it would exacerbate the risk of violence between forces loyal to each side, facilitate future attempts to oust W, and more generally erode W's control over T on which the security of C members depends. Proportionality, in turn, would constrain W's right to use force according to the severity and immediacy of the threat. It would, for example, preclude executing soldiers sent only to establish a physical presence for S in T. W would be justified in expelling these soldiers, but not in killing them.Footnote ¹²

The rights just discussed constitute a set of limited and provisional territorial rights analogous to, but significantly weaker than, those identified in recent theoretical work on the territorial rights of states (D. Miller Reference Miller2012; Moore Reference Moore2015; Nine Reference Nine2012; Stilz Reference Stilz2011). W's rights are limited in that they fall short of the full legislative and international powers typically afforded to fully functioning states; in particular, they exclude borrowing privileges and the right to participate in international institutions or to engage in diplomatic relations with other states. W's rights are also provisional in that they must be formalized in a negotiated settlement with S as soon as feasible. Provisionality reflects an imperative to end the unstable relationship between W and S, which stand, in effect, in a state of war, ready to use force in response to perceived threats and in pursuit of conflicting claims over people and territory. Formalizing W's rule serves not only to resolve the normative problems raised by W's liminal status but also to stabilize an inherently unstable situation.

Provisional and limited territorial rights nonetheless give warlords a sui generis normative standing and generate a set of connected obligations in others. We have already discussed the implications for S, but equally or perhaps more important are the implications for third parties, including the UN and other states. Third parties often play a decisive role in determining state-building success or failure, providing financial and logistical aid to war-wracked states, rehabilitating infrastructure, recruiting and retraining civil servants, and—most controversially for our purposes—supporting efforts to eradicate warlords and other armed nonstate groups that contest the state's rule.

Our theory suggests that these efforts are not always as benign as they may seem. If S wrongfully attacks W using the resources provided by third parties, then those third parties indirectly undermine basic rights in T; if third parties orchestrate the attacks themselves, then they become perpetrators of injustice. Third parties must take all necessary precautions to prevent this from occurring. They must make efforts to mediate the conflict between S and W, deter future attacks, and possibly fund reconstruction if violence occurs. More controversially and speculatively, third parties may provide aid directly to warlords. The greater the warlord's capacity to protect basic rights and the deeper its popular support, the stronger the case for aid.

Finally, these obligations presuppose that third parties must determine whether a warlord meets the conditions of legitimate authority described earlier. Such judgments may be highly demanding epistemically and politically contentious. Predictably, states will see warlords as insurgents or terrorists rather than as agents of protection, and states and warlords will both claim superior will and ability to protect populations at risk. As Anne Orford has noted in the context of R2P, adjudication of such claims requires not only an “assessment of facts” but also an “account of the social conditions under which protection can best be guaranteed;” such an account will, in turn, depend on “who interprets what protection or the safety of the people means in a particular time and place” (Reference Orford2011, 137, 183). Third parties must judge the nature and severity of potential threats to civilians, must gauge warlords’ capacity to provide protection, and must estimate the level of popular support that warlords enjoy in the communities they claim to oversee.

Contentious though they may be, there are reasons to believe such judgments are possible. First and most important, our criteria are in principle observable and operationalizable. The quality of nonstate governance has been successfully documented in a variety of settings (Arjona Reference Arjona2014; Mampilly Reference Mampilly2011; Weinstein Reference Weinstein2006); similar approaches could be taken to assess warlord regimes. Moreover, civil war provides warlords with opportunities to demonstrate their capacities and thus to “reveal”—both to third parties and to civilians—the otherwise latent quality of their rule. Although warlords may have incentives to misrepresent their intentions—for example, to restrict third-party access to areas they control or to coerce citizens into expressing support for their regimes—they must allow scrutiny by third parties to assess the plausibility of their claims. Persisting attempts at obfuscating will be grounds for questioning the warlords’ legitimacy.

AN EMPIRICAL ILLUSTRATION: AFGHANISTAN

How useful is our argument for evaluating warlords’ claims to legitimate authority in real-world cases of civil war and state collapse? To illustrate the theory's potential, we consider an application to Afghanistan, one of the most important examples of externally sponsored state-building in recent memory. We focus on the case of Ismail Khan, one of Afghanistan's most powerful warlords who, at the height of his power, ruled over an “autonomous empire” in Herat Province (Wimmer and Schetter Reference Wimmer and Schetter2003, 530). Then-president Hamid Karzai repeatedly and ultimately successfully sought to remove Khan from power, culminating in Khan's ouster in 2004.

Of course, our very brief account cannot conclusively adjudicate the legitimacy of Khan's rule or the justifiability of Karzai's actions. Nonetheless, we provide a variety of reasons to believe that Khan met and exceeded our two criteria of legitimacy—protecting basic rights and winning popular support even amid decades of civil war and state predation—and that he therefore deserved autonomy from Kabul, formalized through a negotiated settlement with Karzai's regime.

FORMALIZING WARLORDS’ RIGHTS

We have argued that, under some conditions, warlords should be recognized as public authorities within the territories they control and therefore as bearers of limited and provisional territorial rights, including the use of force. This recognition is premised on two fundamental values: the capacity to protect basic rights—in particular, to enforce commands, adjudicate disputes, and preserve public order—and the ability to maintain popular support among those over whom warlords claim the right to rule. In developing this argument we have recovered from late medieval and early modern just war theorists the powerful insights that, in situations of fragmented political control and contested sovereignty, (1) the right to wage war may reside in subnational actors, and (2) war-making powers may extend beyond short-term self- and other-defense to include preservation of public order and protection of basic rights in the longer run.

We have also argued that both states and warlords have a duty to formalize W's territorial rights in a negotiated settlement ending their state of war. This raises the question of what, exactly, such an arrangement should entail. Although a detailed discussion is beyond the scope of this article, we conclude with some general observations about a negotiated settlement's structure and possible effects.

First and most important, the settlement should usually recognize internal autonomy only, not independent statehood. In other words, provisional and limited territorial rights are not tantamount to a right to secession. This is a crucial but potentially controversial stipulation, which may seem either too restrictive or too permissive. On the one hand, it might be objected that since W's consolidation of territorial control indicates a significant dissociation between C and S, secession should be a readily accessible option. Allen Buchanan, for example, has argued that in contexts of state failure, virtually any group for which independence would improve security should be allowed to create a new state—such “sauve qui peut separatism” would be valid whenever states have failed (Reference Buchanan2004, 366–69).

This strikes us as an excessively permissive approach. Secession can radically worsen the security situation in both the new and remainder states, as well as in the surrounding region (Ratner Reference Ratner and Wippman1998; Wellman Reference Wellman2005, 36–38); as such, it should be seen as a last resort, to be pursued only when there are very compelling reasons to think that the basic rights of C members cannot be secured as long as T remains legally within the boundaries of S. A better balance can and should be struck between the value of sovereign statehood and the value of preserving social order and basic individual rights. Autonomy is a less internationally destabilizing alternative to secession and has good prospects for preserving social order and basic rights in the long term.Footnote ¹⁵

On the other hand, it might be objected that even the prospect of autonomy will incentivize violence and foster divergences in S, or that granting limited and provisional territorial rights will only lead to more extensive and destabilizing claims, including secession. It might also be objected that the promise of legitimacy will encourage warlords to appropriate assets belonging to the state, such as taxes, natural resources, and customs revenues (as was the case with Ismail Khan in Afghanistan). Warlords may thus be instrumental in the very institutional collapse that underpins their alleged legitimacy.

Although these are valid concerns, any damaging effects that may be associated with the recognition of a warlord's legitimacy—violence against the state but also against other armed nonstate actors or civilians; attempts by W to pilfer resources from S—are in all likelihood determined already by the dynamics of civil war and the weakness of state institutions in which warlords operate. Moreover, the benefits of recognition that we prescribe do not include those of external sovereignty (borrowing privileges, diplomatic representation, and so on), which make sovereign status particularly attractive. It seems unlikely that the possibility of such limited autonomy conditioned on local support and the protection of basic rights would make a warlord's behavior any worse than it otherwise would be.

On the contrary, in most cases we would expect the prospect of territorial rights to disincentivize actions that threaten civilians living in the territories under warlords’ control. Warlords seeking to demonstrate their capacity to protect basic rights should be less inclined to violate them; warlords seeking to generate popular support should be less inclined to abuse or otherwise alienate civilians on whose endorsement their claim to legitimacy ultimately depends. Nor is it obvious that granting autonomy would incentivize more extensive claims later on. Although the empirical literature is divided on this question, recent studies suggest that concessions of autonomy tend to mitigate rather than exacerbate the risk of future secessions or center-seeking civil wars (Cederman et al. Reference Cederman, Hug, Schädel and Wucherpfennig2015; Martin Reference Martin2013; Siroky and Cuffe Reference Siroky and Cuffe2015). These findings are not specific to warlords, but they support our conjecture that the recognition of autonomy may have pacifying effects. If W can achieve autonomy without seceding or further waging war, then recognition may actually attenuate the risk of future conflict between W and S.

Lastly, there is nothing in our approach that precludes S from attacking or otherwise stopping W before W has met our criteria of legitimacy. Nor does our theory preclude the possibility of warlords losing their claim to legitimacy before a negotiated settlement is reached. As noted before, if S gains greater capacity to protect basic rights and generates greater popular support than W while the state of war is ongoing between them, then W would act wrongly by failing to relinquish power; W's use of force would then be unjust, and S would be justified in retaking control of T. However, in cases where S proves too weak or negligent to prevent W from consolidating power, we agree with those who argue that for the sake of stability, de facto victors deserve de jure rights within the territories they control (Toft Reference Toft2009; Weinstein Reference Weinstein2005)—provided they have also established the capacity to protect basic rights while generating popular support. The key challenge is then to transform de facto into de jure warlord rule under more demanding forms of public scrutiny and accountability.

Thus seen, the negotiated settlement between S and W must convert W's provisional territorial rights into legal ones within S's internationally recognized borders. The settlement must therefore be both backward- and forward-looking. It must address each of the provisional rights discussed earlier, in particular the rights to maintain an independent militia and to collect taxes; must define the form those rights will take within S's constitutional order; and must include provisions for adjudicating subsequent disputes between W and S. Moreover, although we assume that the arrangement between W and S will be binding unless and until its terms are breached by one of the parties, like any negotiated settlement, it must accommodate potential changes in the relative standing of W and S. In particular, if a settlement grants significant autonomy to W in T, it should nonetheless specify conditions under which W might be required to forgo autonomy fully or partially—say, by allowing S or third parties to intervene if the quality of W's rule has deteriorated while that of S outside T has greatly improved. To that end, the negotiated settlement might, for example, designate a neutral third party—such as a constitutional court or an international organization—to consult with C members on a regular basis and evaluate the quality of W's rule in T relative to other areas in S.

Finally, once the conflict between W and S is settled, more demanding criteria of political legitimacy can and should be applied to both, including progressively more democratic forms of representation and protection of an increasing array of human rights. Indeed, to the extent that the negotiated settlement effectively ends a state of war, it signals a degree of stability in which public order and the protection of basic security rights gradually cease to be overriding concerns. More exacting criteria then become appropriate, not just for S but for W as well.