Assisting Rebels Abroad: The Ethics of Violence at the Limits of the Defensive Paradigm

In this article, I analyse the theory and practice of interventions in foreign civil wars to assist rebels ﬁ ghting against violently oppressive government. I argue that the indirect nature of this kind of intervention gives rise to political complications that are either absent from or less obvious in humanitarian interventions aimed chie ﬂ y at defending human rights from imminent threats. An adequate theory must therefore accommodate three additional components. First, it requires a theory of indirect warfare accounting for how the ends of the interveners ’ added violence relate to those of the rebels ’ violence. Second, it requires a theory of indirect political leadership, paying careful attention both to the political status of the rebel leaders vis- (cid:1) a-vis the people on whose behalf they ﬁ ght and to the relationship between those leaders and the interveners. Third, the peculiarities of indirect military intervention mean that the violence it introduces isn ’ t easily explained purely in terms of defensive goals. An adequate account needs additionally to pay attention to the role of violence in shaping new political movements and institutions. The value of such goals is less easily quanti ﬁ ed than those of humanitarian intervention, making it harder to set upper limits on permissible ‘ proportionate ’ harm.

interveners bring about change by acting on and through other political agents, the rebels. Indirect intervention therefore requires careful attention to the ethical complications arising from questions of legitimate political leadership. Secondly, both reform and humanitarian interventions employ violence chiefly to eliminate or deter threats, thus opening safe spaces within which the positive tasks of political reconstruction can be pursued nonviolently. By contrast, I argue that indirect military intervention, properly understood, gives violence a more creative role in generating and shaping political possibilities. The multiple functions that violence serves in indirect military intervention put a strain on an ethical theory modelled on actions that are defensive in a more restrictive sense.
I begin in the next section (Part 2) by questioning the 'parallelism' of recent philosophical treatments of intervention and rebellion and propose a different starting point for analysis. I then identify three distinctive components that a theory of indirect military intervention needs. First (Part 3), it must address how the ends of the interveners' added violence relate to those of rebel violence. Second (Part 4), a theory of indirect political agency pays careful attention to the relationships of rebel leaders to the rebel people and of the interveners to the rebel leaders. Both of these issues complicate the theory by comparison with theories of both humanitarian intervention and revolution. Third (Part 5), I argue that the peculiarities of indirect military intervention mean that the violence it introduces isn't easily explained purely in defensive terms. Moreover, the value of its goalslike those of revolutionis less easily quantified than those of humanitarian intervention. It is therefore harder to set clear upper limits on permissible collateral harm with reference to proportionality in cases of indirect military intervention compared with humanitarian intervention.
Before I begin, some stipulations and clarifications: First, the word 'rebels' refers here to armed, nonstate actors aiming at revolutionary change with potential to secure self-determination through a human-rights-respecting democracy ('democracy,' for short). Second, I presume that this purpose can sometimes justify armed rebellion. 7 Third, my claims concern any attempts to assist rebels militarily, including active participation (e.g. air support); supplying military experts; and training, arming, and funding rebels. 8 While each raises distinctive issues, all share enough features to require a unified theoretical framework. Once we have one, then we might give the differences more fine-grained attention (a problem for another day).

Humanitarian Intervention and Rebellion: From Parallelism to Complementarity
According to the 'prevailing' liberal view, Ned Dobos writes, the set of cases where a state's domestic actions justify armed, international intervention and that in which its own citizens can justify rebellion may overlap extensively but are not quite the same: 'armed intervention is not always justified even where rebellion with similar aims, employing similar means, is acknowledged to be a legitimate option for the victims of tyranny'. 9 Defenders of this asymmetry variously cite differences in how ad bellum proportionality applies to each, different thresholds of oppression justifying each, and in the compatibility of each with self-determination. Yet advocates of both asymmetry and symmetry share an assumption, which is that humanitarian intervention and rebellion against oppressive governments are ethically parallel. That is, both are means of securing the same ends, and there is therefore a range of cases in which either might be chosen, depending on which seems best suited to a case's specifics. 10 Dobos, again, writes: Both radical rebellion and humanitarian intervention aim to defend citizens against tyranny and human rights abuses at the hands of their government. The only difference is that rebellion is waged by the oppressed subjects themselves, while humanitarian intervention is carried out by foreigners on their behalf. 11 Asymmetry and symmetry advocates differ about whether there is a set of cases in which only one sort of action is permissible. But both presume that in other cases, the same oppression might justify either rebellion or humanitarian intervention.
Thinking about intervention and rebellion in this way gives rise to some problems. On that view, two different entitiesa domestic rebel leadership and a foreign intervenercan, in principle, have the same jus belli at the same time, at least prima facie. Motivated by the same objective, their ultimate ends won't conflict. But, given that they are likely to adopt different means and perhaps different strategic objectives, there is a possibility of militaryand politicalconfliction. If they had parallel prima facie justifications, one way to resolve their different claims would be by appealing more general principles of jus ad bellum rather than to a theory designed to make sense of problems specific to interventions. So, for instance, even if each, considered independently, could satisfy proportionality and had a reasonable prospect of success, 'necessity' might help decide: if rebellion had the same prospect of success as intervention, but would secure it at a higher price, then necessity would dictate favouring intervention instead. Or if the ratio between expected morally relevant costs and benefits was equal between them, one might be chosen over the other due to a better chance (or expected degree) of success, albeit at a higher (but still morally acceptable) price in relevant collateral costs. Sometimes cooperation might satisfy best the conditions of jus ad bellum. Or, if all else is equal between alternatives, one might harmonize better with another principle, such as autonomy or fair distribution of costs. But, whichever is chosen, deciding by applying ad bellum criteria case-by-case implies that whether or not a political or military relationship occurs between external and internal actors (and what form it might take) is theoretically and normatively secondary.
I think this underestimates the question's importance. Its significance is thrown into sharper relief if we start from the thought that, instead of presenting parallel alternatives, intervention and rebellion are likely in many cases to be complementary. Let's assume, first, that the core set of rights that might legitimately motivate either intervention or rebellion are those contributing, as the Declaration of the Rights of Man and Citizen (1789) puts it, to a more general 'right of resistance against oppression'. As formulated, this is not the same as a 'right to resist': instead, it captures the idea that people have a right to be protected from oppression and suggests that there may be agents who have placed themselves under a duty so to protect them. I presume, secondly, that the first-best possibility in the current international order is that this duty is discharged by peoples' own governments. Governments may be regarded as being under a perfect duty to protect (at the very least) their own citizens. They frequently fail to do so, either by failing to resist other oppressors or by themselves actively oppressing. In such cases, the responsibility for discharging this duty might devolve to or be taken up by another entity. For reasons I will discuss more fully in the third and fourth sections, it is better, ceteris paribus, that people secure their rights in a self-determining way: it respects better the value of autonomy and is more likely to achieve deep political change with stable outcomes. That being the case, where no viable peaceful, constitutional route to change exists, then the second-best possibility is for competent domestic leaders to challenge the rights-violating government through legitimate rebellion. It is therefore only if domestic initiative fails (or fails to materialize) that the third best alternative is justifiable: to seek the assistance of foreign actors in resisting oppression by means of intervention. 12 On this view, cases like the Rwandan or Darfur or Bosnian genocides fall into the third category: domestic government failed to resist the violence or perpetrated it, and no other suitable domestic actor was available to defeat it. Therefore, the right of resistance against oppression could arguably be claimed against international actors.
On this complementarity view, then, intervention and rebellion are not parallel possibilities, arising simultaneously as rival options in a range of similar cases. Instead, they complement each other, the less ideal alternative arising only where the more ideal fails. Indirect military intervention arises as a means of addressing one particular type of failure, when a domestic rebellion seeks to defeat oppression but needs outside assistance. 13 On this view, a justification for indirect military intervention occurs as a function of (prima facie) justified domestic rebellion. It is not first and foremost a function of an international right (or duty) to intervene or an extension of a prior duty of humanitarian intervention.
On the view I propose, then, the right to engage in intervention therefore cannot be parallel to (nor can it be independent of) the right of rebellion because it is a condition of someone's having a right to intervene that domestic rebellion has either failed already or is expected to fail. Intervention's permissibility thus turns on justifications for rebellion in a way that's quite different from what other scholars have supposed.

Dilemmas of Indirect Agency I: Military Engineering versus Victory
Michael Walzer's noninterventionism rejects parallelism and the idea of symmetry in particular. Drawing on Mill, he places a value on domestic self-determination that defeats liberal arguments for intervening in most cases. Secessionists might request assistance on the basis that they represented a distinct people whose self-determination was stymied within a larger state. And 'crimes that shock the conscience of mankind' might trigger a duty of rescue by demonstrating that there was no natural, self-determined 'fit' between government and governed. Otherwise, the only residual exception is where counterintervention prevents a wrongful intervention by another state from distorting self-determination. 14 Many liberal theorists reject Walzer's emphasis on the value of self-determination. 15 But I think he and Mill nevertheless understood something important about any intervention aiming at political reform. This has less to do with self-determination as a moral value and more to do with its importance as a de facto political dimension of those cases in which indirect military intervention seems most promising. Properly appraising its significance (first two subsections of Part 3) brings to light a dilemma (Subsection 3) for the theory and practice of indirect military intervention between (a) embracing the rebels' political goals and aiming strategically at military victory and (b) keeping the rebels' goals at arm's length while concentrating strategically on engineering the balance of forces within the state in order to facilitate political transition.

Mill and Walzer on Nonintervention
Mill's and Walzer's noninterventionism addresses two ways of altering the political institutions of another state: reform intervention simpliciter, which occurs without prior initiation of rebellion within the target state, and indirect military intervention. But, as Mill puts it, ' [o]f these cases, that of a people in arms for liberty is the only one of any nicety, or which theoretically at least, is likely to present conflicting moral considerations'. 16 Intuitively, such cases might seem to require that capable democratic states offer military assistance to the oppressed. But both Mill and Walzer resist that conclusion. Their common ground is in emphasizing the importance of political self-determination to defining what is permissible. Both maintain that it points towards a general principle of nonintervention in international affairs, albeit with some important exceptions. But their arguments differ in emphasis.
In Just and Unjust Wars, Walzer characterizes self-determination as a value that grounds a moral claim right against interference by political outsiders: 'Self-determination [. . .] is the right of a people "to become free by their own efforts" if they can'. 17 It is connected, Walzer implies, to the 'right' that 'the citizens of a sovereign state have [. . .] to suffer only at one another's hands' when seeking to 'shape the domestic arrangements or alter the conditions of life' in their country. 18 In a later restatement, Walzer expanded on the normative status of self-determination. Judged from the perspective of the international community, he maintains, the legitimacy of states is 'pluralist in character' and reflects 'our recognition of diversity and our respect for communal integrity and for different patterns of cultural and political development'. 19 The value of self-determination 'derives its moral and political force from the rights of contemporary men and women to live as members of a historic community and to express their inherited culture through political forms worked out by themselves '. 20 To open up space within which to revisit the question of indirect military intervention, I want to contrast Walzer's treatment of self-determination with two others. First, what we might call democratic self-determination is a strongly normative ideal that is realized where all competent members of a unified political community engage in an undistorted, egalitarian process of collective self-determination. Like Walzer's, this ideal is grounded in the individual's right to autonomy, embracing both narrowly individual pursuits and choices arising from and affecting groups that individuals might (choose to) belong to. But, as such, self-determination really only occurs under certain conditions. Allen Buchanan, for instance, maintains that self-determination implies 'group agency' which requires 'a structure or process that coordinates the actions of the individual members in such a way as to justify the claim that there is a collective agent'. 21 Where one part of a people systematically dictates to another without consultation, then agency in this normative sense simply isn't attributable to the group as a whole. The category of societies that genuinely experience democratic self-determination therefore excludes authoritarian states, the very cases in relation to which the question of indirect military intervention arises. By contrast, Walzer thinks that even under authoritarian conditions a de facto process of self-determination might yet occur, yielding outcomes that are intrinsically valuable even though they fall far short of the ideal of democratic self-determination: authoritarian self-determination 'has value even if it is not always pretty, and even if its outcome does not conform to philosophical standards of political and social justice'. 22 Walzer's ideal falls somewhere between democratic self-determination and the political idea at the heart of Mill's relatively pragmatic and normatively parsimonious account. In 'Nonintervention,' Mill presents self-determination chiefly as a political constraint that it would be morally wrong to neglect rather than as a moral right that must be honoured or a moral value to be protected. He doesn't use the term 'self-determination' in the text let alone define it as a right, but he does highlight the way a community may negotiate its political commitments endogenously rather than having them imposed from outside. The normative significance of self-determination in this sense arises from empirical assumptions about its importance for founding free government.
Mill thinks self-determination important for two reasons. First, where it is permitted to occur, it can tell parties outside the state something about the moral and political condition of the people inside. If a people successfully replaces domestic tyranny with stable democracy, for instance, then outsiders know it has cultivated conditions suitable for 'free government'. But if it remains subject to authoritarian rule, then outsiders cannot be sure about its readiness for change. And they should be fairly adamant about this: absent successful revolution, outsiders know that they don't know. Second, Mill thought that domestic, self-determining struggle was necessary not only instrumentally to defeat tyrants but also as a 'school' through which to learn democratic virtues and values. 23 Only by completing the struggle, Mill thought, do the political community and its members undergo the deep changes necessary to support a new constitutional order. True and lasting freedom depends on a resulting 'balance of forces' that outside interference can only 'disturb'. This idea of balance informs Mill's account of both the conditions for 'the permanent maintenance of freedom' and the struggle through which people may 'contend successfully' for it. 24 So, if the country lacks a balance suitable to 'free institutions,' then only by internal struggle between its opposing elements can it achieve one. 25 Recent empirical work on campaigns for democracy bears out Mill's assumptions about both the transformative effects of engaging in struggle and their importance for successful postconflict democratic institutions. In their influential analysis of nonviolent resistance, Erica Chenoweth and Maria Stephan argue that the success of both campaigns for democracy and their results is a function of the nature of the political struggle itself. Crucially, a movement's success in achieving mass participation and support is vital to its ability to sustain the struggle and disrupt the opposing regime and to building support for new institutions that will last beyond the conflict. This is partly a matter of mobilizing more and more opponents of the regime and creating 'loyalty shifts' that undermine its power base and convert more people to democracy. Like Mill, Chenoweth and Stephan therefore think that external interventions are 'never substitutes' for internal struggle. 26 In a similar vein, Ali Kadivar argues that 'the duration of unarmed mobilization during democratic transitions predicts the success of democratic consolidation'. The longer the struggle, the stronger and more stable is the resulting democracy. 27 Studies of violent movements support similar conclusions. Although recent comparative research presents a bleak picture so far as the correlation between civil wars in general and the emergence of democracy is concerned, work on insurgencies that aim specifically at that outcome is more optimistic. Elizabeth Jean Wood, for instance, argues that, although nonviolent mass action may be a more promising route more generally to positive political change, 'in oligarchic societies, sustained insurgency [. . .] may lead to democratic rule'. 28 Moreover, recent empirical work, including Wood's, suggests that various outcomes tending to deepen, support, and enrich democracy are correlated with involvement in struggle, including 'more egalitarian gender roles, a more equitable distribution of property rights, an empowered civil society, and unprecedented political participation'. Even the experience of violence itself during insurgency can sometimes have 'pro-social' effects that benefit democratic prospects after civil war. 29 None of these observations recommend violent insurgency over nonviolent forms of action. All else being equal, nonviolence offers better prospects when opportunities exist to pursue it, both for the struggle itself and for the longevity of democratic outcomes. 30 But, while armed struggle may be less auspicious, it too sometimes succeeds. 31 And, of course, the choice isn't always entirely in the hands of rebels; violence is often the government's initiative. 32 Where violence succeeds, it is when, alongside armed struggle, the opposition has the time and means of bringing about political transformationthrough 'diverse, mass mobilization' and by instigating 'loyalty shifts within the ruling [regime's] economic and military elites' which build the 'mass support sufficient to bring about revolutionary change'. 33 Where armed rebellion breaks out during prodemocracy struggles, then, there are reasons to favour Mill's assumptions. Without a further modification to his theory that I propose in the next section, both militate in favour of leaving rebels to themselves, the second one above all. Let's call that process whereby a people decides its own political institutions without interference concrete self-determination. It might be doubted that self-determination in Walzer's (and Mill's) normatively undemanding sense has enough intrinsic moral value to outweigh the value of rescuing individuals from basic rights violations. But Mill's idea that concrete self-determination is an important part of the problem that indirect intervention faces is independent of this claim, and I think it should be taken seriously. What it posits is twofold: on the one hand, concrete self-determination is an empirical constraint that limits the scope for international action; but on the other, by the same reasoning its importance in shaping political outcomes means that promoting it must also be a normatively important part of the political aim of both rebellion and, if it can be justified, intervention.
I'll set aside democratic self-determination because it elides the distinction that Mill asks us to make. 'Self-determination' will henceforth refer to Mill's concern unless otherwise stated.

Internal Counterintervention
Mill's emphasis on concrete self-determination is important, but his focus on how external actors can vitiate it distracts from something else: this is the distorting effect of high concentrations of armed capability wielded by particular domestic actors. Mill's account can be made to yield a plausible case for indirect military intervention if we introduce a further premisethat coercive means are artificial factors that can balk and distort self-determination, regardless of whether they originate domestically or abroad. The resulting dichotomy between 'artificial' and, as it were, 'natural' factors isn't meant to imply any sort of underlying natural law teleology. It indicates ways in which military force can exert influence in shaping political change that is disproportionate to the number of people who wield it. Absent its role in multiplying their salience, what remains is a process of political struggle which is natural at least insofar as that source of distortion has been negated.
Let's review how this might work by distinguishing two sources of a regime's power. The first may be called 'sociological' legitimacy. 34 Taken in a broad sense, this encompasses both the rational endorsement of the government's rule by those with objective reasons to support itits beneficiariesand the acquiescence of those who uphold it even though it doesn't serve their interests. The latter may uphold a government simply by not resisting or even by actively supporting it. Either might be due to a prevalent belief in the moral legitimacy of the government, or to fear, a lack of organization, or adapted preferences. Together, these factors shape the layer of personal qualities and values whose orientation, rootedness, and distribution are Mill's concern in 'Non-Intervention.' The second source of power is an ability to coerce citizens by drawing on a (near) monopoly on the (best) means of violence. If the government has the preponderance of arms (considering quantity, technological quality, and trained personnel to use them), it may be able to shore up a shortfall in support by threatening would-be dissenters. This might also account for habits of acquiescence and adapted preferences, if large-scale force has been used to imprint the population with a collective memory of the dire consequences of resistance. 35 Hannah Arendt's analysis of revolution relies on an account of political power along these lines. 'Power' in the true sense of the word comes from below, she argues, ideally constituted by people coming together and freely coordinating their actions. In this way, they can 'empower' a democratic government. But even autocrats rely on some degree of power in this sense, albeit in an impoverished form. Theirs generally combines the active support of a smaller subset of the population, including the armed forces, and the passive acquiescence of most of the remainder. Ruling purely by coercion is impossible since autocrats need part of the population to support them in applying whatever force is required in order to control the rest. They enjoy greater security when they can rely on demoralization or habit to ensure that most people just go along with their rule. 36 One way, then, that revolution can occur is when mass protest demonstrates a declining willingness to obey, which threatens a catastrophic encroachment on the regime's power base and which might eventually force it to choose: either capitulate or commit to the difficult labour of sustained mass coercion. 37 Given the nature of modern armed forces and technology, the extent to which the state's armed forces can intensify its coercive powerwhat R.B. Gregg calls 'power over'will frequently be sufficient to destroy the 'power with' of popular opposition: 'In a head-on clash between violence and power,' Arendt writes, 'the outcome is hardly in doubt.' 38 On Arendt's view, then, the logic of violence in an armed confrontation is at odds with the logic of political self-determination as Mill conceives it. The greater the asymmetry of arms between opponents, the more likely it is that coercion rather than popular power and political engagements between opponents will determine outcomes and, hence, the more self-determination will be distorted. 39 The success of popularly selfdetermining political revolution relies on convincing enough people to defect to the opposition, withdrawing their acquiescence with the incumbent powers. 40 But even if opponents of government predominated across the state, it is no guarantee that revolution can succeed. As Stathis Kalyvas writes, 'it is not necessarily the case that political majorities enjoy a military advantage over minorities; in fact, the opposite may be true'. 41 So even without foreign interference, the progress and outcome of domestic conflict (the balances emphasized by Mill) are shaped by influences acting contrary to the logic of political self-determination. This points to a plausible objective for indirect military interveners, similar to that of counterintervention in Mill's and Walzer's analyses: intervention should compensate for the artificial effects of a distribution of violent means favouring the regime. 42 Let's call the theory that this sort of domestic counterintervention is legitimately a neo-Millian account. While Arendt commented extensively on revolution, she didn't offer an account of legitimate intervention. But there is another historical theory of foreign intervention in revolutionary wars that helps fill out the neo-Millian account. Mikhail Tukhachevsky's article, 'Revolution from Without', contributed to Bolshevik debates during the Russian Civil War about supporting popular communist revolution abroad. It shares with Mill's and Walzer's analyses (and, indeed, Chenoweth and Stephan's) a belief that revolution must essentially be the work of domestic political forces and cannot be imposed externally. What it adds is twofold: first, Tukhachevsky's recognition that government arms are likely to crush incipient but politically promising revolts; and second, his appreciation of the role of military success in shaping the war's political outcomes.
'In general,' Tukhachevsky writes, 'the conduct of socialist war must be an extension of the revolutionary work of the communists of the other State, if the sum of conditions there prevents the development of revolution without assistance.' Mirroring Mill's worry, he writes that, '[t]o impose a socialist revolution by force is impossible.
[It] can only be the handing over of armed force to a revolutionary working class which is unable to get the better of the army and police of its own bourgeoisie'. 43 But Tukhachevsky nevertheless offers an account of legitimate foreign intervention drawing on a Leninist theory that the state simply is a concentration of coercive power. Its essential components are the 'special bodies of armed men' employed to maintain a system of domination enabling a small minority to exploit an expanding and increasingly impoverished majority. On Lenin's account, asymmetry in the ownership of productive means necessitates a radical asymmetry in the ownership of destructive means. 44 Consequently, if revolutionary insurrection breaks out, Tukhachevsky writes: One sidethe bourgeoisiepossesses a regular, well-trained and wellequipped army, whose commanding officers are firmly attached to the bourgeois class and, like it, accustomed to hate any proletarian movement wholeheartedly. On the other side, a workers' army just coming into existence, an army which has to be created from nothing, having neither officers, arms nor equipment.
Provided the state doesn't hesitate in using violence, the asymmetry can be decisive. Then revolution can succeed only if benevolent outside forces support the insurgents' fight long enough for the balance to shift in their favour. 45 What this adds to a neo-Millian account is a suggestion about how a contribution of military force by an international actor can facilitate struggle without vitiating its integrity as a process of concrete political self-determination. First and foremost, the purpose of intervening is to prevent overwhelming domestic military superiority from crushing it. But doing so has a second beneficial effect. By delaying the state's repressive forces directly, intervention facilitates indirectly an emerging domestic preponderance of political forces favourable to revolution that can, in turn, promote a parallel shift in the balance of military force. Like recent work on the theory of nonviolent mass opposition, Tukhachevsky's focus is on the importance of shifting political allegiances, particularly those of soldiers. 46 Whereas counterrevolutionary officers are dependable, the poorer, ordinary soldiers are prone to defect. So if Soviet intervention deflects the initial bourgeois repression, the ensuing flow of defectors eventually tilts the balance of violence over to the revolution, aligning it with the distribution of political will across the country while further expanding the revolutionary political base. So if the bourgeoisie fails to prevent civil war from unfolding, 'little by little all the advantages will swing over decisively to the side of the revolution'. 47

A Theory of Indirect Warfare
Whether expressed in liberal or Marxist terms, the upshot of a neo-Millian theory is the same: it opens up the possibility of a form of war that is indirect in the sense that it contributes to another ongoing war, one initiated and led by agents other than the interveners. What it therefore needs is an account of how the additional violence interveners introduce ought to be integrated with that of the domestic rebels. In this connection, interveners face a twofold dilemma: concerning how they contribute to the rebels' political ends and how they frame their strategic-military objectives in doing so.
Indirect military interveners must choose or find an appropriate compromise between two ways of orientating their actions vis-a-vis the rebels' political ends. On the one hand, they might embrace them, fighting for those ends as their own: violence, then, is justified insofar as those goals are just, the means are proportionate and necessary, and it can help secure them. The strategic aim should then be victory for the rebels. There are two arguments for adopting this direct approach to the rebellion. First, if the ultimate aim of indirect military intervention is to support democratization, then choices tending towards success are preferable, ceteris paribus. Second, swift victory might reduce the overall human cost of civil war (especially in light of Allen Buchanan's analysis, discussed below (Part 4). 48 If so, then indirect military intervention should not only help rebels win but also seek to ensure that the number of casualties averted by shortening war exceeds any caused by the additional force needed to expedite victory. 49 But on the other hand, a neo-Millian might limit interveners' political goals to facilitating concrete self-determination, which suggests that an arm's-length relationship with rebel goals would be more appropriate. The goal of indirect military intervention is then distinguishable from the freedoms that revolutionaries pursue, even if helps secure them indirectly. While aiming to negate factors that vitiate political self-determination, indirect military intervention must avoid adding further vitiating factors. This restricts the inputs of force it adds, limiting them to those necessary to counteract the effects of asymmetry in the distribution of force, and counsels against further inputs that might enhance the chance of rebel victory or accelerate it. Rather than expediting rebel victory, indirect military intervention might therefore have to allow the civil war to continue for its full naturalor quasi-naturalduration.
Tukhachevsky's analysis of defection supports the view that expediting victory could have disastrous political implications. Underlying the defections on which he focuses and the changing military balance they cause is a more fundamental shift, leading back towards Mill's concerns. As Gregg argued in his pacifist critique of Leninism, genuine revolutionary change occurs when enemies are persuaded to change perspective and rethink their values. The direct application of violence is likely to have limited value in achieving thiscoercion typically only creates the appearance of conversion and results in superficial political change. 50 Contemporary analysts of democratic revolution agree that the encouragement of military defections is important to the success of campaigns for change. 51 They also recognize that the attachment of the military to the state and a failure to convert it to democratic values and to civilian control constitute major obstacles to the fulfilment of revolutionary hopes after the overt struggle ends. 52 Military defections contribute a vital part to the sort of moral and political dynamic that Gregg indicates whereby enemy soldiers are converted to revolutionary values.
Prioritizing swift victory at the cost of reducing the number of converts risks jeopardizing political transformations that can contribute to the legitimacy of a postrevolutionary order. Given the terrible destructiveness of civil wars, however, a sensible theory ought to give due attention to all three ends to which intervention might contribute: first, limiting military destruction; second, helping revolutionaries secure their goals; and third, facilitating political transition through Mill's 'school' of struggle (understood, following Tukhachevsky and Gregg, as a process of defection and conversion) by rebalancing domestic military forces. Where these ends conflict, deciding which way to turn demands of interveners a difficult judgement about both military and political matters. The right decision as to which ends to prioritize, resolving the dilemma, is unlikely to be obvious.

Dilemmas of Indirect Agency II: Self-determination versus Armed Entelechy
Let 'subject' refer to the political entity directing violence as opposed to the agents of violence whom it directs. The first set of dilemmas concerned the ends and means of legitimate violence; the second concerns its subjects. In cases of humanitarian intervention simpliciter, there is only one subject, the intervener. Likewise, in a purely domestic revolution, those who successfully claim the leadership of rebellion form a subject of legitimate violence. But indirect military intervention envisages two subjects, the intervener and the rebel leadership. And since they are not responding in parallel to the rights claims of the oppressed, we need an account of what their relationship should be. 53 This question poses another dilemma. Insofar as indirect military intervention aims to prevent artificial factors from distorting a natural process of self-determination, its actions should passively track rebel needs where doing so is compatible with supporting an appropriate rebalancing of forces. However, it is likely that, in some cases, indirect military interveners will have to make choices between rival rebel leaderships. When they do, interveners play an unavoidably creative role in shaping the political complexion of the rebel side as a whole.

Legitimacy in Rebellion
Let's start with a simpler case: a people fights for liberty without outside assistance but with reasonable hope of success. What requirements must rebel leaders satisfy to claim moral and political legitimacy? The assumption that something corresponding to what the just war tradition calls 'legitimate' or 'right' authority is truly required by the jus ad bellum has been questioned by some recent work, but it also has its defenders. 54 Based on some of these arguments, I shall assume that legitimacy is at least partly scalar and depends on six factors: (1) Strategic intelligence: This epistemic requirement refers to the ability to identify military goals suited to achieving the rebellion's political aims and to formulate a credible military strategy for securing them by calculating the effects of alternatives and comparing them. 55 (2) Coercive ability / authority to command: Leadership must be able to coordinate enough people reliably enough to execute its strategy. This might ideally be due to people's belief in its competence and good faith. But it might also rely on coercion, especially during the incipient phases of a civil war. 56 (3) Moral probity: Relatedly, legitimacy is affected by a commitment (and ability, per [2]) to satisfy the requirements of just conduct in war as far as possible, whether defined directly by the legal jus in bello or in other ways. 57 (4) Consent: Legitimacy is strengthened by a willingness of the people on whose behalf it acts to be led by it into and during war. In part, this may be the other side of the same coin as (2) in a subset of the cases where leadership relies on authority rather than mere coercion. It is partly about consenting to being led by this particular subject and partly about consenting to its particular decisions to initiate war and to fight it in a particular way. 58 (5) Input legitimacy: Leaders gain legitimacy according to their epistemic capacity to gauge the character and preferences of those it leads. It benefits from knowledge of the degree to which people (and how many people) are willing to risk death, injury, bereavement, and impoverishment for the sake of political goals. This is an important indication of the value that ought to be put on their lives (which, I presume, they may be able to discount) and what value to put on their political aspirations when applying proportionality and necessity standards to different strategies. 59 (6) Orientation: The rebel leadership needs to have sincerely embraced the appropriate political objectives and moral ends.
I take it that all of these conditions are directly relevant to the credibility of a rebel group's claim to legitimacy as the leadership of a people in revolt against their government. All indicate qualities that it must have or facets of its relationship to those people that affect its claim to be competent to act in their name and their interests. Where a rebel group satisfied all of these conditions, it would have the strongest possible claim to recognition as legitimate leadership. But each may be satisfied partially rather than wholly. Establishing a minimum threshold of legitimacy above which an agent can claim the right to use force on behalf of others is, of course, a vexed question, but this is more than I require for present purposes. 60 Suffice to say that any faults under any of the six headings will weaken a group's claim, a fact likely to be particularly important where there are competing claims for legitimate authority between rival groups of leaders. If these are plausible criteria for cases where there is not (yet) a case for indirect military intervention, then it points towards two types of scenario in which intervention might begin to appear necessary. In the best case, rebel leaders satisfy these criteria well enough to claim a high degree of legitimacy but lack something else. In the more difficult case, leaders fail to satisfy these criteria more comprehensively.

Learning from al Qaeda
So, what does this suggest about the relationship between the interveners and rebels? Where rebel leaders enjoy a high degree of legitimacy, then interveners should work through their leadership. The reasons for connecting with the rebel leadership in the first case are twofold. First, if leaders satisfy conditions (4) to (6) in particular, then both the leadership and the people it represents have an autonomy-based moral claim to political recognition. Second, the rebel leadership is itself a central part of the political objective of an indirect military intervention. If an intervener fails to establish a suitable relationship with a legitimate, competent, domestic subject, then its actions will constitute a simple reform intervention with all the problems that come with that. 61 To illustrate, imagine two states fighting a defensive war against a third. Whether they maintain an effective alliance may affect their ability to win and satisfy proportionality and necessity. But if they do so without an alliance, then their relationship doesn't matter: victory is still victory. Not so with indirect military intervention: if you don't have an appropriate domestic partner among the rebels, and if your strategy for intervention doesn't take its development into account, then this vitiates the outcomes even if you win the war. 62 This is because the aims of indirect intervention are of a specific political kind: to secure democracy, it is necessary to help establish institutions, the people needed to operate them, and an appropriate political relationship between those people and the wider population. Indirect interveners must therefore try to bring about a constellation of forces through which the violence they introduce can help create and give shape to political institutions. Without an appropriate relationship with the right rebel leaders, indirect military intervention is therefore unlikely to succeed. By contrast, the two states fighting an aggressor succeed when they put a barrier between political institutions, which already exist and which they wish to preserve, and the effects of violence.
The lines of argument considered so far indicate that indirect interveners ought to adopt a more or less passive attitude, politically, towards the rebel leaderships. If political initiative, as Mill argues, should be entirely endogenous, then external assistance should track domestic initiatives. Interveners might conceivably provide leadership in purely military, strategic matters, but, as far as unfolding political balances are concerned, the rebel leaders lead, the indirect military intervener follows. This does not, however, exhaust the problem. Other factors point in the opposite direction.
In the 'best case', there already exists a domestically legitimate leadership for interveners to engage with. But this doesn't always occur. Allen Buchanan identifies two common problems. First, revolutions frequently exhibit multiple, rival leaderships. Each might satisfy some criteria to some degree, but not all. Second is when the challenges facing incipient revolt compel leaders to violate individual rights and the principles of jus in bello. For instance, leaders needing to build support for rebellion against a regime that puts a high price on disobedience may feel compelled to terrorize their own people more than the enemy does. And the need to establish unified leadership might also motivate violence between rival groups. Both problems lead Buchanan to recommend earlier interventions, all else being equal, lending further weight to the reasoning for aiming at swift victory discussed in Part 3. They also support leniency in applying criterion (3) when evaluating the eligibility of rebel leaders for external support. 63 But the second problem Buchanan highlights adds a further complication: how to decide between rival rebel leaderships. This problem arose in Syria when the Free Syrian Army fought alongside a swathe of factions, larger and smaller. One response might be to let the rebels fight amongst themselves and to back the horse that proves itself by winning. But doing so faces a similar objection to nonintervention. The fastest horse might be the best funded and armed, not the most morally promising. Allowing internecine fighting to run its course permits artificial factors to influence outcomes. And the most successful groups might not be the most legitimate or those most likely to pursue democracy and human rights: a group that satisfied conditions (1) and (2) but not (3), (4), (5), or (6) might attract more recruits in a desperate struggle than groups with the opposite bundle. Over the course of the Syrian conflict, for instance, more secular, nationalist rebels found themselves outgunned by the regime. By contrast, al Qaeda's soldiersmany with long experience of insurgency elsewherehad greater funds and better weapons. By forging an umbrella movement like the Jabhat al-Nusra Front, al Qaeda was able to use military proficiency to encourage large numbers to compromise on ideology and defect from less radical groups. 64 Al Qaeda's strategy capitalized on a phenomenon identified by the comparative sociology of civil war. As Kalyvas writes, 'military resources generally trump the population's prewar political and social preferences in spawning control. In turn, control has a decisive impact on the population's collaboration with a political actor'. 65 Al Qaeda wasn't representative of wider Syrian opinion at first but leveraged its fighting capacity to increase the proportion of people willing to espouse its goals. This is a form of leadership that looks beyond the passivity of tracking de facto public opinion, instead seeking to reshape it through conversionin this regard, it recalls Tukhachevsky. Highlighting the dangers of a wait-and-see approach, it suggests that, rather than waiting for the race to finish, indirect interveners ought to cultivate militarily the morally most promising rebels, judged according to the degree to which they satisfy the six criteria of internal legitimacy and especially (3)-(6). This way, conversions might be made to flow back from groups like al Qaeda and towards more benign elements.
Deliberately cultivating internal political possibilities creates a tension between this method of intervention and Millian principles. Like Leninism, it recognizes the duty of revolutionary leadershipboth domestic and foreign -'to be that of guiding and developing popular consciousness, not passively reflecting it'. 66 But it is also driven by a neo-Millian ambition to counterbalance the artificial effects of military capacities while ensuring that leaders endogenous to the state are wedded to external normative standards of democracy. If indirect military intervention is likely to have to give a steer to political change in this way, this further complicates how much force indirect military intervention should contribute: in addition to that required by balancing conflict between opponents, it is also necessary to correct the balance within the rebel side. And trying to attract and convert rebels on the ground will tend towards increasing the armed contributions beyond that required by the first balance.

Conclusions: Calculating Proportionate Costs
With its demand for careful attention to the landscape of political allegiances and 'existing social forces,' as Yves Winter puts it, indirect military intervention takes us into Machiavellian theoretical territory. 67 It doesn't act purely on abstract humanitarian ideals, opposing force with force in order to preserve existing lives, but it rather seeks concrete political possibilities to cultivate, promoting the most promising at the expense of the others. Through this sort of armed entelechy, political relationships between interveners and rebels become central and essential parts of normative analysis, not contingent, peripheral issues.
I conclude with a third issue, one distancing indirect military intervention from humanitarian intervention understood as a purely defensive measure. Parts 3 and 4 identified a series of difficulties determining how much force indirect military intervention should introduce. The third part identified a difficult choice between victory-directed violence and balance-directed engineering. The fourth part showed how the need to bolster one group at the expense of rivals might complicate matters further. But the problem is even more intractable. This is because of the diverse functions that violence performs on the neo-Millian account: insofar as the aim is re-engineering an inauspicious military imbalance, it doesn't only preserve lives from violence but also protects the integrity of a political balance from distortion. If it also contributes to armed entelechy, then violence isn't narrowly defensive (orientated towards preserving things that exist) but creative of new possibilities (by growing them from indigenous seeds). The institutions that this contributes to might, in due course, defend people from threatsviolence introduced now could therefore be interpreted as indirectly defensive of people in the future. But its function is not literally and directly defensive: it aims at a future in which something valuable exists where nothing as valuable existed before. 68 And the value of the new entity is much less determinate than that of the lives of a finite number of existing people.
Insofar as a democratic government created with help from indirect military intervention offers ongoing protections and opportunities to future generations, its value is of indefinite extent. It is therefore less clear how we should calibrate judgements about the proportionality of actions that will collaterally harm innocent people. If you are defending a finite number of existing (innocent) people from imminent harms, as you might be in a direct, humanitarian intervention, then it is excessive to cause similar harms to a greater number of innocent people. But it is less clear how to specify that point beyond which harms to the innocent cease to be the lesser evil when they are the by-product of violence aimed at creating things whose value is not directly commensurable with innocent lives. In place of, in principle, the quite determinate applicability of proportionality to humanitarian intervention, therefore, indirect military intervention seems to demand practical moral and political judgements without the bannisters of determinate numerical figures.